Compare commits

..

46 Commits

Author SHA1 Message Date
Sam Machin
3b4e0bcd60 Merge branch 'main' into fix/1155 2025-05-13 15:44:33 +01:00
Dave Horton
610c9af274 update db-helpers 2025-05-13 10:32:30 -04:00
Dave Horton
c0a35af591 update to 0.2.10 speech-utils (#1199) 2025-05-13 10:11:26 -04:00
Hoan Luu Huu
9585018147 support whisper instructions (#1198)
* support whisper instructions

* wip

* update speech utils and verb specification
2025-05-13 09:44:00 -04:00
Hoan Luu Huu
d7884a837a update deepgram voice agent (#1191)
* update deepgram voice agent

* fix lint

* wip

* wip
2025-05-13 07:43:48 -04:00
Dave Horton
ca0bf36815 dont apply snake casing to either env vars or tool call args (#1194) (#1197) 2025-05-12 12:56:58 -04:00
Sam Machin
6b68d32e2c end if last_word_end is -1 (#1196)
* end if last_word_end is -1

* lint
2025-05-12 12:11:32 -04:00
rammohan-y
8217a76697 Removed this.name from Task constructor, as LLM's names are populated post calling the base construction (#1192)
Also fixed a jslint error
2025-05-12 09:14:33 -04:00
rammohan-y
5c8237b382 Feat 1179 race issue with play verb (#1183)
* Fixed race issue between queueCommand false and queueCommand true when play task is involved

https://github.com/jambonz/jambonz-feature-server/issues/1179

* removed unnecessary emitter

* added destroy mechanism for stickyEventEmitter

* clearing stickyEventEmitter

* memory leak fix
2025-05-11 20:25:48 -04:00
Vasudev Anubrolu
4ff5c845de feat/864 update speech utils for playht on prem (#1187)
* feat/864 update speech utils for playht on prem

* feat/864 update speech utils version package lock
2025-05-09 12:34:14 -04:00
Anton Voylenko
78ebd08490 feat: prioritize JAMBONES_LOGLEVEL over db setting (#1188) 2025-05-09 09:41:23 -04:00
Hoan Luu Huu
8b18532f31 fixed tts streaming buffer cannot reset timeoutwhen lastUpdateTime is short (#1184)
* fixed tts streaming buffer cannot reset timeoutwhen lastUpdateTime is short

* wip
2025-05-07 10:26:11 -04:00
rammohan-y
e4bb00b382 Send stop-playback event (#1186)
* Send stop-playback event

https://github.com/jambonz/jambonz-feature-server/issues/1185

* check if not notified in playback-stop, ensure that the stop-playback is sent when kill-playback is sent
2025-05-07 08:59:59 -04:00
Hoan Luu Huu
14295dcebc support google s2s (#1169)
* support google s2s

* wip
2025-05-07 07:20:33 -04:00
Hoan Luu Huu
4d68c179ea sip_decline release callSession if ws requestor is used (#1182) 2025-05-06 10:01:36 -04:00
Hoan Luu Huu
6205959f53 fix microsoft stt max client buffer size error for transcribe verb (#1173) 2025-04-29 09:41:24 -04:00
Hoan Luu Huu
ed92cb2632 update speech utils 0.2.7 (#1177)
* update speech utils 0.2.7

* wip
2025-04-29 08:26:09 -04:00
Sam Machin
3098e04ed6 send env_vars in callHook (#1175)
* send env_vars in callHook

* lint

* add try/catch
2025-04-28 09:51:37 -04:00
Hoan Luu Huu
7e2fe72b6c fix say verb cannot failover if tts_response-code != 2xx (#1174) 2025-04-28 08:46:08 -04:00
Hoan Luu Huu
c2666b7a09 fixed deepgram gather cannot be timeout on empty transcription with continueAsr (#1171) 2025-04-28 08:36:31 -04:00
Hoan Luu Huu
9d54ca8116 Jambonz support Model context protocol (MCP) (#1150)
* Jambonz support Model context protocol (MCP)

* merged mcp tools with existing llmOptions.tools

* support list of mcp servers

* wip

* wip

* wip

* fix voice agent

* fix open-ai

* fix review comment

* fix deepgram voice agent

* update verb specification version
2025-04-24 06:50:53 -04:00
Sam Machin
472f4f4532 clientTools over webhooks (#1167)
* clientTools over webhooks

* lint

* simpler toolHook response
2025-04-23 09:15:16 -04:00
Hoan Luu Huu
63899d0091 update speech utils version 0.2.6 (#1172) 2025-04-23 08:22:47 -04:00
rammohan-y
31e6997746 Updated drachtio-srf version (#1170)
https://github.com/drachtio/drachtio-server/issues/424
2025-04-22 11:06:59 -04:00
Hoan Luu Huu
15b583ef2c only connect to drachtio server if connected to freeswitch (#1123)
* only connect to drachtio server if connected to freeswitch

* wip

* wip
2025-04-22 09:55:39 -04:00
Sam Machin
0bf2013934 add default model (#1147) 2025-04-22 09:49:24 -04:00
rammohan-y
182c310191 remove video from sdp in case of reInvite if the call is audio call (#1159)
https://github.com/jambonz/jambonz-feature-server/issues/1158
2025-04-18 09:33:16 -04:00
Sam Machin
4e74bab728 handle errors in createing call (#1164)
* move createCall into the try/catch and add a completionReason to results for errors

* add default completionReason

fixes #1165

* lint
2025-04-17 07:43:22 -04:00
Sam Machin
9bbc76df12 ensure that all actions play at least once 2025-04-15 11:44:13 +01:00
rammohan-y
87195b6444 Stop tts streaming, when bargeIn is enabled for gather verb and input is detected (#1154)
https://github.com/jambonz/jambonz-feature-server/issues/1153
2025-04-14 09:18:50 -04:00
rammohan-y
eb5e6fa515 Updated db-helper to 0.9.11 (#1152)
https://github.com/jambonz/jambonz-feature-server/issues/1151
2025-04-14 08:22:28 -04:00
Dave Horton
305facb03b Fix/11labs no client config (#1149)
* update to verb specs

* add parameter to api call when there is not client config provided
2025-04-12 10:36:35 -04:00
Dave Horton
d310ba0ed1 reduce verbosity of logging (#1145) 2025-04-09 15:36:58 -04:00
Hoan Luu Huu
77f0fc85a3 ell tts support speech and pronunciation_dictionary_locators (#1137) 2025-04-09 12:32:06 -04:00
Sam Machin
c708b7d007 fix initial message format (#1144) 2025-04-09 10:43:07 -04:00
Hoan Luu Huu
343b382373 fixed ws-requestor missing hook for dial:confirm (#1143) 2025-04-09 07:29:08 -04:00
rammohan-y
0a541e089d Fix for https://github.com/jambonz/jambonz-feature-server/issues/1138 (#1139) 2025-04-04 09:02:18 -04:00
rammohan-y
d910981b1a Allow hangup verb on siprec call (#1136)
* Allow hangup verb on siprec call
https://github.com/jambonz/jambonz-feature-server/issues/1135

* added sip:decline to AllowedSipRecVerbs
2025-04-04 08:23:39 -04:00
Hoan Luu Huu
3f2744f032 fixed replaceEndpoint offer single codec that callee does not support (#1131) 2025-04-03 07:58:39 -04:00
Dave Horton
fcaf2e59e7 initial changes for openai stt (#1127)
* initial changes for openai stt

* wip

* wip

* wip

* wip

* wip

* make minBargeinWordCount work for openai

* wip

* wip

* wip

* wip

* wip

* wip

* wip

* wipp

* wip

* wip

* wip

* openai stt: support for prompt templates

* lint

* wip

* support openai semantic_vad

* wip

* transcribe supports openai stt

* sip

* wip

* wip

* refactor list of stt vendors that dont need to be restarted after a final transcript

* cleanup

* wip

* cleanup

* wip

* wip

* wip

* remove credentials from log

* comment
2025-03-28 13:14:58 -04:00
rammohan-y
ee846b283d Feat 1120 video call - remove video media from SDP if the call is audio call (#1124)
* sending jambonz:error when the incoming message is not parsable

https://github.com/jambonz/jambonz-feature-server/issues/1094

* writing an alert when incoming paylod is invalid

* added content to the jambonz:error payload

* removing video media from sdp if the call is an audio call. This is to avoid sending video media to destination if the incoming call is an audio call

* calling removeVideoSdp only when the environment variable JAMBONES_VIDEO_CALLS_ENABLED_IN_FS is set to true, this will ensure there are no regression issues for audio calls

* fixed jslint errors
2025-03-28 12:56:08 -04:00
Hoan Luu Huu
acdb8695a0 allow cartesia model_id is override from synthesizer option (#1130) 2025-03-27 13:37:57 -04:00
Hoan Luu Huu
f33f197e8d gather say support ttstream (#1128) 2025-03-27 07:19:19 -04:00
Sam Machin
9c437ab687 use deepgramOptions.model (#1126)
* use deepgramOptions.model

* lint

* Update transcription-utils.js
2025-03-24 12:25:29 -04:00
Dave Horton
1873694784 update to dractio-fsmrf@4.0.2 2025-03-17 08:50:10 -04:00
rammohan-y
d36e6b4c22 set the detected language as language_code when deepgram detects the language (#1116)
https://github.com/jambonz/jambonz-feature-server/issues/1115
2025-03-11 12:16:29 -04:00
34 changed files with 2173 additions and 410 deletions

95
app.js
View File

@@ -27,8 +27,61 @@ const pino = require('pino');
const logger = pino(opts, pino.destination({sync: false}));
const {LifeCycleEvents, FS_UUID_SET_NAME, SystemState, FEATURE_SERVER} = require('./lib/utils/constants');
const installSrfLocals = require('./lib/utils/install-srf-locals');
installSrfLocals(srf, logger);
const createHttpListener = require('./lib/utils/http-listener');
const healthCheck = require('@jambonz/http-health-check');
logger.on('level-change', (lvl, _val, prevLvl, _prevVal, instance) => {
if (logger !== instance) {
return;
}
logger.info('system log level %s was changed to %s', prevLvl, lvl);
});
// Install the srf locals
installSrfLocals(srf, logger, {
onFreeswitchConnect: (wraper) => {
// Only connect to drachtio if freeswitch is connected
logger.info(`connected to freeswitch at ${wraper.ms.address}, start drachtio server`);
if (DRACHTIO_HOST) {
srf.connect({host: DRACHTIO_HOST, port: DRACHTIO_PORT, secret: DRACHTIO_SECRET });
srf.on('connect', (err, hp) => {
const arr = /^(.*)\/(.*)$/.exec(hp.split(',').pop());
srf.locals.localSipAddress = `${arr[2]}`;
logger.info(`connected to drachtio listening on ${hp}, local sip address is ${srf.locals.localSipAddress}`);
});
}
else {
logger.info(`listening for drachtio requests on port ${DRACHTIO_PORT}`);
srf.listen({port: DRACHTIO_PORT, secret: DRACHTIO_SECRET});
}
// Start Http server
createHttpListener(logger, srf)
.then(({server, app}) => {
httpServer = server;
healthCheck({app, logger, path: '/', fn: getCount});
return {server, app};
})
.catch((err) => {
logger.error(err, 'Error creating http listener');
});
},
onFreeswitchDisconnect: (wraper) => {
// check if all freeswitch connections are lost, disconnect drachtio server
logger.info(`lost connection to freeswitch at ${wraper.ms.address}`);
const ms = srf.locals.getFreeswitch();
if (!ms) {
logger.info('no freeswitch connections, stopping drachtio server');
disconnect();
}
}
});
if (NODE_ENV === 'test') {
srf.on('error', (err) => {
logger.info(err, 'Error connecting to drachtio');
});
}
// Init services
const writeSystemAlerts = srf.locals?.writeSystemAlerts;
if (writeSystemAlerts) {
writeSystemAlerts({
@@ -54,24 +107,6 @@ const {
const InboundCallSession = require('./lib/session/inbound-call-session');
const SipRecCallSession = require('./lib/session/siprec-call-session');
if (DRACHTIO_HOST) {
srf.connect({host: DRACHTIO_HOST, port: DRACHTIO_PORT, secret: DRACHTIO_SECRET });
srf.on('connect', (err, hp) => {
const arr = /^(.*)\/(.*)$/.exec(hp.split(',').pop());
srf.locals.localSipAddress = `${arr[2]}`;
logger.info(`connected to drachtio listening on ${hp}, local sip address is ${srf.locals.localSipAddress}`);
});
}
else {
logger.info(`listening for drachtio requests on port ${DRACHTIO_PORT}`);
srf.listen({port: DRACHTIO_PORT, secret: DRACHTIO_SECRET});
}
if (NODE_ENV === 'test') {
srf.on('error', (err) => {
logger.info(err, 'Error connecting to drachtio');
});
}
srf.use('invite', [
initLocals,
createRootSpan,
@@ -97,27 +132,20 @@ sessionTracker.on('idle', () => {
}
});
const getCount = () => sessionTracker.count;
const healthCheck = require('@jambonz/http-health-check');
let httpServer;
const createHttpListener = require('./lib/utils/http-listener');
createHttpListener(logger, srf)
.then(({server, app}) => {
httpServer = server;
healthCheck({app, logger, path: '/', fn: getCount});
return {server, app};
})
.catch((err) => {
logger.error(err, 'Error creating http listener');
});
const monInterval = setInterval(async() => {
srf.locals.stats.gauge('fs.sip.calls.count', sessionTracker.count);
try {
const systemInformation = await srf.locals.dbHelpers.lookupSystemInformation();
if (systemInformation && systemInformation.log_level) {
logger.level = systemInformation.log_level;
const envLogLevel = logger.levels.values[JAMBONES_LOGLEVEL.toLowerCase()];
const dbLogLevel = logger.levels.values[systemInformation.log_level];
const appliedLogLevel = Math.min(envLogLevel, dbLogLevel);
if (logger.levelVal !== appliedLogLevel) {
logger.level = logger.levels.labels[Math.min(envLogLevel, dbLogLevel)];
}
}
} catch (err) {
if (process.env.NODE_ENV === 'test') {
@@ -133,6 +161,7 @@ const disconnect = () => {
httpServer?.on('close', resolve);
httpServer?.close();
srf.disconnect();
srf.removeAllListeners();
srf.locals.mediaservers?.forEach((ms) => ms.disconnect());
});
};

View File

@@ -15,6 +15,7 @@ const {
JAMBONES_DISABLE_DIRECT_P2P_CALL
} = require('./config');
const { createJambonzApp } = require('./dynamic-apps');
const { decrypt } = require('./utils/encrypt-decrypt');
module.exports = function(srf, logger) {
const {
@@ -348,11 +349,10 @@ module.exports = function(srf, logger) {
}
req.locals.application = app2;
// eslint-disable-next-line no-unused-vars
const {call_hook, call_status_hook, ...appInfo} = app; // mask sensitive data like user/pass on webhook
// eslint-disable-next-line no-unused-vars
const {requestor, notifier, ...loggable} = appInfo;
const {requestor, notifier, env_vars, ...loggable} = appInfo;
logger.info({app: loggable}, `retrieved application for incoming call to ${req.locals.calledNumber}`);
req.locals.callInfo = new CallInfo({
req,
@@ -417,10 +417,22 @@ module.exports = function(srf, logger) {
...(app.fallback_speech_recognizer_language && {fallback_language: app.fallback_speech_recognizer_language})
}
};
let env_vars;
try {
if (app.env_vars) {
const d_env_vars = JSON.parse(decrypt(app.env_vars));
logger.info(`Setting env_vars: ${Object.keys(d_env_vars)}`); // Only log the keys not the values
env_vars = d_env_vars;
}
} catch (error) {
logger.info('Unable to set env_vars', error);
}
const params = Object.assign(['POST', 'WS'].includes(app.call_hook.method) ? { sip: req.msg } : {},
req.locals.callInfo,
{ service_provider_sid: req.locals.service_provider_sid },
{ defaults });
{ defaults },
{ env_vars }
);
logger.debug({ params }, 'sending initial webhook');
const obj = rootSpan.startChildSpan('performAppWebhook');
span = obj.span;

View File

@@ -23,6 +23,7 @@ const HttpRequestor = require('../utils/http-requestor');
const WsRequestor = require('../utils/ws-requestor');
const ActionHookDelayProcessor = require('../utils/action-hook-delay');
const TtsStreamingBuffer = require('../utils/tts-streaming-buffer');
const StickyEventEmitter = require('../utils/sticky-event-emitter');
const {parseUri} = require('drachtio-srf');
const {
JAMBONES_INJECT_CONTENT,
@@ -79,6 +80,10 @@ class CallSession extends Emitter {
this.callGone = false;
this.notifiedComplete = false;
this.rootSpan = rootSpan;
this.stickyEventEmitter = new StickyEventEmitter();
this.stickyEventEmitter.onSuccess = () => {
this.taskInProgress = null;
};
this.backgroundTaskManager = new BackgroundTaskManager({
cs: this,
logger,
@@ -135,6 +140,15 @@ class CallSession extends Emitter {
this.requestor.on('handover', handover.bind(this));
this.requestor.on('reconnect-error', this._onSessionReconnectError.bind(this));
}
/**
* Currently this is used for openai STT, which has a prompt paramater and
* we have an experimental feature where you can send the conversation
* history in the prompt
*/
this.conversationTurns = [];
this.on('userSaid', this._onUserSaid.bind(this));
this.on('botSaid', this._onBotSaid.bind(this));
}
/**
@@ -953,7 +967,7 @@ class CallSession extends Emitter {
(type === 'tts' && credential.use_for_tts) ||
(type === 'stt' && credential.use_for_stt)
)) {
this.logger.info(
this.logger.debug(
`${type}: ${credential.vendor} ${credential.label ? `, label: ${credential.label}` : ''} `);
if ('google' === vendor) {
if (type === 'tts' && !credential.tts_tested_ok ||
@@ -1106,12 +1120,17 @@ class CallSession extends Emitter {
};
}
else if ('speechmatics' === vendor) {
this.logger.info({credential}, 'CallSession:getSpeechCredentials - speechmatics credential');
return {
api_key: credential.api_key,
speechmatics_stt_uri: credential.speechmatics_stt_uri,
};
}
else if ('openai' === vendor) {
return {
api_key: credential.api_key,
model_id: credential.model_id,
};
}
else if (vendor.startsWith('custom:')) {
return {
speech_credential_sid: credential.speech_credential_sid,
@@ -1166,7 +1185,9 @@ class CallSession extends Emitter {
const taskNum = ++this.taskIdx;
const stackNum = this.stackIdx;
const task = this.tasks.shift();
this.logger.info(`CallSession:exec starting task #${stackNum}:${taskNum}: ${task.name}`);
this.isCurTaskPlay = TaskName.Play === task.name;
this.taskInProgress = task;
this.logger.info(`CallSession:exec starting task #${stackNum}:${taskNum}: ${task.name} : {task.taskId}`);
this._notifyTaskStatus(task, {event: 'starting'});
// Register verbhook span wait for end
task.on('VerbHookSpanWaitForEnd', ({span}) => {
@@ -1240,7 +1261,7 @@ class CallSession extends Emitter {
this.logger.info('CallSession:exec all tasks complete');
this._stopping = true;
this._onTasksDone();
this._clearResources();
await this._clearResources();
if (!this.isConfirmCallSession && !this.isSmsCallSession) sessionTracker.remove(this.callSid);
@@ -1445,7 +1466,7 @@ class CallSession extends Emitter {
}
if (tasks) {
const t = normalizeJambones(this.logger, tasks).map((tdata) => makeTask(this.logger, tdata));
this.logger.info({tasks: listTaskNames(t)}, 'CallSession:_lccCallHook new task list');
this.logger.debug({tasks: listTaskNames(t)}, 'CallSession:_lccCallHook new task list');
this.replaceApplication(t);
if (this.wakeupResolver) {
//this.logger.debug({resolution}, 'CallSession:_onCommand - got commands, waking up..');
@@ -1712,10 +1733,10 @@ Duration=${duration} `
this.currentTask.ep :
this.ep;
const db = parseDecibels(opts);
this.logger.info(`_lccBoostAudioSignal: boosting audio signal by ${db} dB`);
this.logger.debug(`_lccBoostAudioSignal: boosting audio signal by ${db} dB`);
const args = [ep.uuid, 'setGain', db];
const response = await ep.api('uuid_dub', args);
this.logger.info({response}, '_lccBoostAudioSignal: response from freeswitch');
this.logger.debug({response}, '_lccBoostAudioSignal: response from freeswitch');
}
async _lccMediaPath(desiredPath) {
@@ -1768,7 +1789,6 @@ Duration=${duration} `
let res;
try {
res = await this.ttsStreamingBuffer?.bufferTokens(tokens);
this.logger.info({id, res}, 'CallSession:_lccTtsTokens - tts:tokens-result');
} catch (err) {
this.logger.info(err, 'CallSession:_lccTtsTokens');
}
@@ -1903,9 +1923,11 @@ Duration=${duration} `
this.tasks = tasks;
this.taskIdx = 0;
this.stackIdx++;
this.logger.info({tasks: listTaskNames(tasks)},
this.logger.debug({tasks: listTaskNames(tasks)},
`CallSession:replaceApplication reset with ${tasks.length} new tasks, stack depth is ${this.stackIdx}`);
if (this.currentTask) {
this.logger.debug('CallSession:replaceApplication - killing current task ' +
this.currentTask?.name + ', taskId: ' + this.currentTask.taskId);
this.currentTask.kill(this, KillReason.Replaced);
this.currentTask = null;
}
@@ -1914,11 +1936,15 @@ Duration=${duration} `
this.wakeupResolver({reason: 'new tasks'});
this.wakeupResolver = null;
}
if ((!this.currentTask || this.currentTask === undefined) && this.isCurTaskPlay) {
this.logger.debug(`CallSession:replaceApplication - emitting uuid_break, taskId: ${this.taskInProgress?.taskId}`);
this.stickyEventEmitter.emit('uuid_break', this.taskInProgress);
}
}
kill(onBackgroundGatherBargein = false) {
if (this.isConfirmCallSession) this.logger.debug('CallSession:kill (ConfirmSession)');
else this.logger.info('CallSession:kill');
else this.logger.debug('CallSession:kill');
this._endVerbHookSpan();
if (this.currentTask) {
this.currentTask.kill(this);
@@ -1983,7 +2009,7 @@ Duration=${duration} `
task.synthesizer.label :
this.speechSynthesisLabel;
this.logger.info({vendor, language, voice, label},
this.logger.debug({vendor, language, voice, label},
'CallSession:_preCacheAudio - precaching audio for future prompt');
task._synthesizeWithSpecificVendor(this, this.ep, {vendor, language, voice, label, preCache: true})
.catch((err) => this.logger.error(err, 'CallSession:_preCacheAudio - error precaching audio'));
@@ -2053,7 +2079,7 @@ Duration=${duration} `
}
async _onCommand({msgid, command, call_sid, queueCommand, tool_call_id, data}) {
this.logger.info({msgid, command, queueCommand, data}, 'CallSession:_onCommand - received command');
this.logger.debug({msgid, command, queueCommand, data}, 'CallSession:_onCommand - received command');
let resolution;
switch (command) {
case 'redirect':
@@ -2062,18 +2088,18 @@ Duration=${duration} `
const t = normalizeJambones(this.logger, data)
.map((tdata) => makeTask(this.logger, tdata));
if (!queueCommand) {
this.logger.info({tasks: listTaskNames(t)}, 'CallSession:_onCommand new task list');
this.logger.debug({tasks: listTaskNames(t)}, 'CallSession:_onCommand new task list');
this.replaceApplication(t);
}
else if (JAMBONES_INJECT_CONTENT) {
if (JAMBONES_EAGERLY_PRE_CACHE_AUDIO) this._preCacheAudio(t);
this._injectTasks(t);
this.logger.info({tasks: listTaskNames(this.tasks)}, 'CallSession:_onCommand - updated task list');
this.logger.debug({tasks: listTaskNames(this.tasks)}, 'CallSession:_onCommand - updated task list');
}
else {
if (JAMBONES_EAGERLY_PRE_CACHE_AUDIO) this._preCacheAudio(t);
this.tasks.push(...t);
this.logger.info({tasks: listTaskNames(this.tasks)}, 'CallSession:_onCommand - updated task list');
this.logger.debug({tasks: listTaskNames(this.tasks)}, 'CallSession:_onCommand - updated task list');
}
resolution = {reason: 'received command, new tasks', queue: queueCommand, command};
resolution.command = listTaskNames(t);
@@ -2340,10 +2366,32 @@ Duration=${duration} `
// Destroy previous ep if it's still running.
if (this.ep?.connected) this.ep.destroy();
this.ep = await this.ms.createEndpoint({remoteSdp: this.dlg.remote.sdp});
/* Codec negotiation issue explanation:
*
* Problem scenario:
* 1. Initial negotiation:
* - FreeSWITCH → SBC: offers multiple codecs (PCMU, PCMA, G722)
* - SBC → Callee: passes all codecs (PCMU, PCMA, G722)
* - Callee → SBC: responds with PCMA (its supported codec)
* - SBC → FreeSWITCH: responds with PCMU (after transcoding)
*
* 2. After endpoint replacement:
* - If we only offer PCMU in the new endpoint
* - FreeSWITCH → SBC: offers only PCMU
* - SBC → Callee: offers only PCMU
* - Call fails: Callee rejects since it only supports PCMA
*
* Solution:
* Always have FreeSWITCH offer multiple codecs to the SBC, don't pass remote sdp here to ensure
* the SBC can reoffer the same codecs that the callee originally accepted.
* This prevents call failures during media renegotiation.
*/
this.ep = await this.ms.createEndpoint();
this._configMsEndpoint();
await this.dlg.modify(this.ep.local.sdp);
const sdp = await this.dlg.modify(this.ep.local.sdp);
await this.ep.modify(sdp);
this.logger.debug('CallSession:replaceEndpoint completed');
return this.ep;
}
@@ -2351,9 +2399,16 @@ Duration=${duration} `
/**
* Hang up the call and free the media endpoint
*/
_clearResources() {
async _clearResources() {
this.stickyEventEmitter.destroy();
this.stickyEventEmitter = null;
this.taskInProgress = null;
for (const resource of [this.dlg, this.ep, this.ep2]) {
if (resource && resource.connected) resource.destroy();
try {
if (resource && resource.connected) await resource.destroy();
} catch (err) {
this.logger.info({err}, 'CallSession:_clearResources - error clearing resources');
}
}
this.dlg = null;
this.ep = null;
@@ -2567,7 +2622,7 @@ Duration=${duration} `
if (json && Array.isArray(json)) {
const tasks = normalizeJambones(this.logger, json).map((tdata) => makeTask(this.logger, tdata));
if (tasks && tasks.length > 0) {
this.logger.info('CallSession:handleRefer received REFER, get new tasks');
this.logger.debug('CallSession:handleRefer received REFER, get new tasks');
this.replaceApplication(tasks);
if (this.wakeupResolver) {
this.wakeupResolver({reason: 'CallSession: referHook new taks'});
@@ -2614,14 +2669,14 @@ Duration=${duration} `
if (typeof this.queueEventHookRequestor === 'undefined') {
const pp = this._pool.promise();
try {
this.logger.info({accountSid: this.accountSid}, 'performQueueWebhook: looking up account');
this.logger.debug({accountSid: this.accountSid}, 'performQueueWebhook: looking up account');
const [r] = await pp.query(sqlRetrieveQueueEventHook, [this.accountSid]);
if (0 === r.length) {
this.logger.info({accountSid: this.accountSid}, 'performQueueWebhook: no webhook provisioned');
this.queueEventHookRequestor = null;
}
else {
this.logger.info({accountSid: this.accountSid, webhook: r[0]}, 'performQueueWebhook: webhook found');
this.logger.debug({accountSid: this.accountSid, webhook: r[0]}, 'performQueueWebhook: webhook found');
this.queueEventHookRequestor = new HttpRequestor(this.logger, this.accountSid,
r[0], this.webhook_secret);
this.queueEventHook = r[0];
@@ -2635,7 +2690,7 @@ Duration=${duration} `
/* send webhook */
const params = {...obj, ...this.callInfo.toJSON()};
this.logger.info({accountSid: this.accountSid, params}, 'performQueueWebhook: sending webhook');
this.logger.debug({accountSid: this.accountSid, params}, 'performQueueWebhook: sending webhook');
this.queueEventHookRequestor.request('queue:status', this.queueEventHook, params)
.catch((err) => {
this.logger.info({err, accountSid: this.accountSid, obj}, 'Error sending queue notification event');
@@ -2770,7 +2825,7 @@ Duration=${duration} `
async handleReinviteAfterMediaReleased(req, res) {
assert(this.dlg && this.dlg.connected && !this.ep);
const sdp = await this.dlg.modify(req.body);
this.logger.info({sdp}, 'CallSession:handleReinviteAfterMediaReleased - reinvite to A leg returned sdp');
this.logger.debug({sdp}, 'CallSession:handleReinviteAfterMediaReleased - reinvite to A leg returned sdp');
res.send(200, {body: sdp});
}
@@ -2851,7 +2906,7 @@ Duration=${duration} `
// if final transcription is received but endpoint is already closed,
// freeswitch module will not be able to send the transcription
this.logger.info('callSession:_configMsEndpoint -' +
this.logger.debug('callSession:_configMsEndpoint -' +
' transcribe task, wait for some time before destroy');
await sleepFor(JAMBONES_TRANSCRIBE_EP_DESTROY_DELAY_MS);
}
@@ -2905,7 +2960,7 @@ Duration=${duration} `
_awaitCommandsOrHangup() {
assert(!this.wakeupResolver);
return new Promise((resolve, reject) => {
this.logger.info('_awaitCommandsOrHangup - waiting...');
this.logger.debug('_awaitCommandsOrHangup - waiting...');
this.wakeupResolver = resolve;
if (this._actionHookDelayProcessor) {
@@ -2925,7 +2980,7 @@ Duration=${duration} `
this.ep.play(this.fillerNoise.url);
this.ep.once('playback-start', (evt) => {
if (evt.file === this.fillerNoise.url && !this._isPlayingFillerNoise) {
this.logger.info('CallSession:_awaitCommandsOrHangup - filler noise started');
this.logger.debug('CallSession:_awaitCommandsOrHangup - filler noise started');
this.ep.api('uuid_break', this.ep.uuid)
.catch((err) => this.logger.info(err, 'Error killing filler noise'));
}
@@ -2936,7 +2991,7 @@ Duration=${duration} `
_clearTasks(backgroundGather, evt) {
if (this.requestor instanceof WsRequestor && !backgroundGather.cleared) {
this.logger.info({evt}, 'CallSession:_clearTasks on event from background gather');
this.logger.debug({evt}, 'CallSession:_clearTasks on event from background gather');
try {
backgroundGather.cleared = true;
this.kill(true);
@@ -3014,6 +3069,43 @@ Duration=${duration} `
this._jambonzHangup('Max Call Duration');
this._maxCallDurationTimer = null;
}
_onUserSaid(transcript) {
const count = this.conversationTurns.length;
if (count === 0 || this.conversationTurns[count - 1].type === 'assistant') {
this.conversationTurns.push({
type: 'user',
text: transcript
});
}
else {
this.conversationTurns[count - 1].text += ` ${transcript}`;
}
}
_onBotSaid(transcript) {
const count = this.conversationTurns.length;
if (count === 0 || this.conversationTurns[count - 1].type === 'user') {
this.conversationTurns.push({
type: 'assistant',
text: transcript
});
}
else {
this.conversationTurns[count - 1].text += ` ${transcript}`;
}
}
getFormattedConversation(numTurns) {
const turns = this.conversationTurns.slice(-numTurns);
if (turns.length === 0) return null;
return turns.map((t) => {
if (t.type === 'user') {
return `user: ${t.text}`;
}
return `assistant: ${t.text}`;
}).join('\n');
}
}
module.exports = CallSession;

View File

@@ -63,7 +63,7 @@ class RestCallSession extends CallSession {
this.callInfo.callTerminationBy = terminatedBy;
const duration = moment().diff(this.dlg.connectTime, 'seconds');
this.emit('callStatusChange', {callStatus: CallStatus.Completed, duration});
this.logger.debug(`RestCallSession: called party hung up by ${terminatedBy}`);
this.logger.info(`RestCallSession: called party hung up by ${terminatedBy}`);
this._callReleased();
}

View File

@@ -550,7 +550,8 @@ class TaskDial extends Task {
proxy: `sip:${sbcAddress}`,
callingNumber: this.callerId || fromUri.user,
...(this.callerName && {callingName: this.callerName}),
opusFirst: isOpusFirst(this.cs.ep.remote.sdp)
opusFirst: isOpusFirst(this.cs.ep.remote.sdp),
isVideoCall: this.cs.ep.remote.sdp.includes('m=video')
};
const t = this.target.find((t) => t.type === 'teams');

View File

@@ -12,6 +12,7 @@ const {
JambonzTranscriptionEvents,
AssemblyAiTranscriptionEvents,
VoxistTranscriptionEvents,
OpenAITranscriptionEvents,
VadDetection,
VerbioTranscriptionEvents,
SpeechmaticsTranscriptionEvents
@@ -83,6 +84,7 @@ class TaskGather extends SttTask {
this._bufferedTranscripts = [];
this.partialTranscriptsCount = 0;
this.bugname_prefix = 'gather_';
}
get name() { return TaskName.Gather; }
@@ -110,6 +112,12 @@ class TaskGather extends SttTask {
return this.fillerNoise.startDelaySecs;
}
get isStreamingTts() { return this.sayTask && this.sayTask.isStreamingTts; }
getTtsVendorData() {
if (this.sayTask) return this.sayTask.getTtsVendorData(this.cs);
}
get summary() {
let s = `${this.name}{`;
if (this.input.length === 2) s += 'inputs=[speech,digits],';
@@ -233,6 +241,7 @@ class TaskGather extends SttTask {
const {span, ctx} = this.startChildSpan(`nested:${this.sayTask.summary}`);
const process = () => {
this.logger.debug('Gather: nested say task completed');
this.playComplete = true;
if (!this.listenDuringPrompt) {
startDtmfListener();
}
@@ -263,6 +272,7 @@ class TaskGather extends SttTask {
const {span, ctx} = this.startChildSpan(`nested:${this.playTask.summary}`);
const process = () => {
this.logger.debug('Gather: nested play task completed');
this.playComplete = true;
if (!this.listenDuringPrompt) {
startDtmfListener();
}
@@ -553,6 +563,31 @@ class TaskGather extends SttTask {
break;
case 'openai':
this.bugname = `${this.bugname_prefix}openai_transcribe`;
this.addCustomEventListener(
ep, OpenAITranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, OpenAITranscriptionEvents.SpeechStarted, this._onOpenAISpeechStarted.bind(this, cs, ep));
this.addCustomEventListener(
ep, OpenAITranscriptionEvents.SpeechStopped, this._onOpenAISpeechStopped.bind(this, cs, ep));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.Error,
this._onOpenAIErrror.bind(this, cs, ep));
/* openai delta transcripts are useful only for minBargeinWordCount eval */
if (this.minBargeinWordCount > 1) {
this.openaiPartials = [];
opts.OPENAI_WANT_PARTIALS = 1;
this.addCustomEventListener(
ep, OpenAITranscriptionEvents.PartialTranscript, this._onOpenAIPartialTranscript.bind(this, cs, ep));
}
this.modelSupportsConversationTracking = opts.OPENAI_MODEL !== 'whisper-1';
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
@@ -584,6 +619,25 @@ class TaskGather extends SttTask {
bugname: this.bugname
}, 'Gather:_startTranscribing');
/* special feature for openai: we can provide a prompt that includes recent conversation history */
let prompt;
if (this.vendor === 'openai') {
if (this.modelSupportsConversationTracking) {
prompt = this.formatOpenAIPrompt(this.cs, {
prompt: this.data.recognizer?.openaiOptions?.prompt,
hintsTemplate: this.data.recognizer?.openaiOptions?.promptTemplates?.hintsTemplate,
// eslint-disable-next-line max-len
conversationHistoryTemplate: this.data.recognizer?.openaiOptions?.promptTemplates?.conversationHistoryTemplate,
hints: this.data.recognizer?.hints,
});
this.logger.debug({prompt}, 'Gather:_startTranscribing - created an openai prompt');
}
else if (this.data.recognizer?.hints?.length > 0) {
prompt = this.data.recognizer?.hints.join(', ');
}
}
/**
* Note: we don't need to ask deepgram for interim results, because they
* already send us words as they are finalized (is_final=true) even before
@@ -595,6 +649,7 @@ class TaskGather extends SttTask {
interim: this.interim,
bugname: this.bugname,
hostport: this.hostport,
prompt
}).catch((err) => {
const {writeAlerts, AlertType} = this.cs.srf.locals;
this.logger.error(err, 'TaskGather:_startTranscribing error');
@@ -630,7 +685,9 @@ class TaskGather extends SttTask {
}
_startAsrTimer() {
if (this.vendor === 'deepgram') return; // no need
// Deepgram has a case that UtteranceEnd is not sent to cover the last word end time.
// So we need to wait for the asrTimeout to be sure that the last word is sent.
// if (this.vendor === 'deepgram') return; // no need
assert(this.isContinuousAsr);
this._clearAsrTimer();
this._asrTimer = setTimeout(() => {
@@ -775,7 +832,11 @@ class TaskGather extends SttTask {
const bugname = fsEvent.getHeader('media-bugname');
const finished = fsEvent.getHeader('transcription-session-finished');
this.logger.debug({evt, bugname, finished, vendor: this.vendor}, 'Gather:_onTranscription raw transcript');
if (bugname && this.bugname !== bugname) return;
if (bugname && this.bugname !== bugname) {
this.logger.debug(
`Gather:_onTranscription - ignoring transcript from ${bugname} because our bug is ${this.bugname}`);
return;
}
if (finished === 'true') return;
if (this.vendor === 'ibm' && evt?.state === 'listening') return;
@@ -786,7 +847,8 @@ class TaskGather extends SttTask {
}
else {
const utteranceTime = evt.last_word_end;
if (utteranceTime && this._dgTimeOfLastUnprocessedWord && utteranceTime < this._dgTimeOfLastUnprocessedWord) {
// eslint-disable-next-line max-len
if (utteranceTime && this._dgTimeOfLastUnprocessedWord && utteranceTime < this._dgTimeOfLastUnprocessedWord && utteranceTime != -1) {
this.logger.debug('Gather:_onTranscription - got UtteranceEnd with unprocessed words, continue listening');
}
else {
@@ -1078,6 +1140,33 @@ class TaskGather extends SttTask {
this._onVendorError(cs, _ep, {error: JSON.stringify(e)});
}
async _onOpenAIErrror(cs, _ep, evt) {
// eslint-disable-next-line no-unused-vars
const {message, ...e} = evt;
this._onVendorError(cs, _ep, {error: JSON.stringify(e)});
}
async _onOpenAISpeechStarted(cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onOpenAISpeechStarted');
}
async _onOpenAISpeechStopped(cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onOpenAISpeechStopped');
}
async _onOpenAIPartialTranscript(cs, _ep, evt) {
if (!this.playComplete) {
const words = evt.delta.split(' ').filter((w) => /[A-Za-z0-0]/.test(w));
this.openaiPartials.push(...words);
this.logger.debug({words, partials: this.openaiPartials, evt}, 'TaskGather:_onOpenAIPartialTranscript - words');
if (this.openaiPartials.length >= this.minBargeinWordCount) {
this.logger.debug({partials: this.openaiPartials}, 'killing audio due to speech (openai)');
this._killAudio(cs);
this.notifyStatus({event: 'speech-bargein-detected', words: this.openaiPartials});
}
}
}
async _onVendorError(cs, _ep, evt) {
super._onVendorError(cs, _ep, evt);
if (!(await this._startFallback(cs, _ep, evt))) {
@@ -1187,6 +1276,7 @@ class TaskGather extends SttTask {
}
}
else if (reason.startsWith('speech')) {
this.cs.emit('userSaid', evt.alternatives[0].transcript);
if (this.parentTask) this.parentTask.emit('transcription', evt);
else {
this.emit('transcription', evt);

View File

@@ -221,7 +221,7 @@ class TaskListen extends Task {
}
}
_onConnect(ep) {
this.logger.debug('TaskListen:_onConnect');
this.logger.info('TaskListen:_onConnect');
}
_onConnectFailure(ep, evt) {
this.logger.info(evt, 'TaskListen:_onConnectFailure');

View File

@@ -4,6 +4,8 @@ const TaskLlmOpenAI_S2S = require('./llms/openai_s2s');
const TaskLlmVoiceAgent_S2S = require('./llms/voice_agent_s2s');
const TaskLlmUltravox_S2S = require('./llms/ultravox_s2s');
const TaskLlmElevenlabs_S2S = require('./llms/elevenlabs_s2s');
const TaskLlmGoogle_S2S = require('./llms/google_s2s');
const LlmMcpService = require('../../utils/llm-mcp');
class TaskLlm extends Task {
constructor(logger, opts) {
@@ -18,6 +20,8 @@ class TaskLlm extends Task {
// delegate to the specific llm model
this.llm = this.createSpecificLlm();
// MCP
this.mcpServers = this.data.mcpServers || [];
}
get name() { return this.llm.name ; }
@@ -28,14 +32,32 @@ class TaskLlm extends Task {
get ep() { return this.cs.ep; }
get mcpService() {
return this.llmMcpService;
}
get isMcpEnabled() {
return this.mcpServers.length > 0;
}
async exec(cs, {ep}) {
await super.exec(cs, {ep});
// create the MCP service if we have MCP servers
if (this.isMcpEnabled) {
this.llmMcpService = new LlmMcpService(this.logger, this.mcpServers);
await this.llmMcpService.init();
}
await this.llm.exec(cs, {ep});
}
async kill(cs) {
super.kill(cs);
await this.llm.kill(cs);
// clean up MCP clients
if (this.isMcpEnabled) {
await this.mcpService.close();
}
}
createSpecificLlm() {
@@ -59,6 +81,10 @@ class TaskLlm extends Task {
llm = new TaskLlmElevenlabs_S2S(this.logger, this.data, this);
break;
case 'google':
llm = new TaskLlmGoogle_S2S(this.logger, this.data, this);
break;
default:
throw new Error(`Unsupported vendor ${this.vendor} for LLM`);
}
@@ -82,8 +108,15 @@ class TaskLlm extends Task {
await this.cs?.requestor.request('llm:event', this.eventHook, data);
}
async sendToolHook(tool_call_id, data) {
await this.cs?.requestor.request('llm:tool-call', this.toolHook, {tool_call_id, ...data});
const tool_response = await this.cs?.requestor.request('llm:tool-call', this.toolHook, {tool_call_id, ...data});
// if the toolHook was a websocket it will return undefined, otherwise it should return an object
if (typeof tool_response != 'undefined') {
tool_response.type = 'client_tool_result';
tool_response.invocation_id = tool_call_id;
this.processToolOutput(tool_call_id, tool_response);
}
}
async processToolOutput(tool_call_id, data) {

View File

@@ -164,7 +164,9 @@ class TaskLlmElevenlabs_S2S extends Task {
try {
const {host, path} = await this.getSignedUrl();
const args = [ep.uuid, 'session.create', this.input_sample_rate, this.output_sample_rate, host, path];
const args = this.conversation_initiation_client_data ?
[ep.uuid, 'session.create', this.input_sample_rate, this.output_sample_rate, host, path] :
[ep.uuid, 'session.create', this.input_sample_rate, this.output_sample_rate, host, path, 'no_initial_config'];
await this._api(ep, args);
} catch (err) {
this.logger.error({err}, 'TaskLlmElevenlabs_S2S:_startListening');
@@ -189,7 +191,7 @@ class TaskLlmElevenlabs_S2S extends Task {
if (this.conversation_initiation_client_data) {
if (!await this._sendClientEvent(ep, {
type: 'conversation_initiation_client_data',
conversation_initiation_client_data: this.conversation_initiation_client_data
...this.conversation_initiation_client_data
})) {
this.notifyTaskDone();
}
@@ -242,13 +244,36 @@ class TaskLlmElevenlabs_S2S extends Task {
/* tool calls */
else if (type === 'client_tool_call') {
this.logger.debug({evt}, 'TaskLlmElevenlabs_S2S:_onServerEvent - function_call');
if (!this.toolHook) {
const {tool_name: name, tool_call_id: call_id, parameters: args} = evt.client_tool_call;
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools.some((tool) => tool.name === name)) {
this.logger.debug({name, args}, 'TaskLlmElevenlabs_S2S:_onServerEvent - calling mcp tool');
try {
const res = await this.parent.mcpService.callMcpTool(name, args);
this.logger.debug({res}, 'TaskLlmElevenlabs_S2S:_onServerEvent - function_call - mcp result');
this.processToolOutput(ep, call_id, {
data: {
type: 'client_tool_result',
tool_call_id: call_id,
result: res.content?.length ? res.content[0] : res.content,
is_error: false
}
});
return;
}
catch (err) {
this.logger.info({err, evt}, 'TaskLlmElevenlabs_S2S - error calling mcp tool');
this.results = {
completionReason: 'client error calling mcp function',
error: err
};
endConversation = true;
}
} else if (!this.toolHook) {
this.logger.warn({evt}, 'TaskLlmElevenlabs_S2S:_onServerEvent - no toolHook defined!');
}
else {
const {client_tool_call} = evt;
const {tool_name: name, tool_call_id: call_id, parameters: args} = client_tool_call;
try {
await this.parent.sendToolHook(call_id, {name, args});
} catch (err) {

View File

@@ -0,0 +1,313 @@
const Task = require('../../task');
const TaskName = 'Llm_Google_s2s';
const {LlmEvents_Google} = require('../../../utils/constants');
const ClientEvent = 'client.event';
const SessionDelete = 'session.delete';
const google_server_events = [
'error',
'session.created',
'session.updated',
];
const expandWildcards = (events) => {
const expandedEvents = [];
events.forEach((evt) => {
if (evt.endsWith('.*')) {
const prefix = evt.slice(0, -2); // Remove the wildcard ".*"
const matchingEvents = google_server_events.filter((e) => e.startsWith(prefix));
expandedEvents.push(...matchingEvents);
} else {
expandedEvents.push(evt);
}
});
return expandedEvents;
};
class TaskLlmGoogle_S2S extends Task {
constructor(logger, opts, parentTask) {
super(logger, opts, parentTask);
this.parent = parentTask;
this.vendor = this.parent.vendor;
this.vendor = this.parent.vendor;
this.model = this.parent.model || 'gemini-2o';
this.auth = this.parent.auth;
this.connectionOptions = this.parent.connectOptions;
const {apiKey} = this.auth || {};
if (!apiKey) throw new Error('auth.apiKey is required for Google S2S');
this.apiKey = apiKey;
this.actionHook = this.data.actionHook;
this.eventHook = this.data.eventHook;
this.toolHook = this.data.toolHook;
const {setup} = this.data.llmOptions;
if (typeof setup !== 'object') {
throw new Error('llmOptions with an initial setup is required for Google S2S');
}
this.setup = {
...setup,
// make sure output is always audio
generationConfig: {
...(setup.generationConfig || {}),
responseModalities: 'audio'
}
};
this.results = {
completionReason: 'normal conversation end'
};
/**
* only one of these will have items,
* if includeEvents, then these are the events to include
* if excludeEvents, then these are the events to exclude
*/
this.includeEvents = [];
this.excludeEvents = [];
/* default to all events if user did not specify */
this._populateEvents(this.data.events || google_server_events);
this.addCustomEventListener = parentTask.addCustomEventListener.bind(parentTask);
this.removeCustomEventListeners = parentTask.removeCustomEventListeners.bind(parentTask);
}
get name() { return TaskName; }
async _api(ep, args) {
const res = await ep.api('uuid_google_s2s', `^^|${args.join('|')}`);
if (!res.body?.startsWith('+OK')) {
throw new Error({args}, `Error calling uuid_openai_s2s: ${res.body}`);
}
}
async exec(cs, {ep}) {
await super.exec(cs);
await this._startListening(cs, ep);
await this.awaitTaskDone();
/* note: the parent llm verb started the span, which is why this is necessary */
await this.parent.performAction(this.results);
this._unregisterHandlers();
}
async kill(cs) {
super.kill(cs);
this._api(cs.ep, [cs.ep.uuid, SessionDelete])
.catch((err) => this.logger.info({err}, 'TaskLlmGoogle_S2S:kill - error deleting session'));
this.notifyTaskDone();
}
_populateEvents(events) {
if (events.includes('all')) {
/* work by excluding specific events */
const exclude = events
.filter((evt) => evt.startsWith('-'))
.map((evt) => evt.slice(1));
if (exclude.length === 0) this.includeEvents = google_server_events;
else this.excludeEvents = expandWildcards(exclude);
}
else {
/* work by including specific events */
const include = events
.filter((evt) => !evt.startsWith('-'));
this.includeEvents = expandWildcards(include);
}
this.logger.debug({
includeEvents: this.includeEvents,
excludeEvents: this.excludeEvents
}, 'TaskLlmGoogle_S2S:_populateEvents');
}
async _startListening(cs, ep) {
this._registerHandlers(ep);
try {
const args = [ep.uuid, 'session.create', this.apiKey];
await this._api(ep, args);
} catch (err) {
this.logger.error({err}, 'TaskLlmGoogle_S2S:_startListening');
this.notifyTaskDone();
}
}
async _sendClientEvent(ep, obj) {
let ok = true;
this.logger.debug({obj}, 'TaskLlmGoogle_S2S:_sendClientEvent');
try {
const args = [ep.uuid, ClientEvent, JSON.stringify(obj)];
await this._api(ep, args);
} catch (err) {
ok = false;
this.logger.error({err}, 'TaskLlmGoogle_S2S:_sendClientEvent - Error');
}
return ok;
}
async _sendInitialMessage(ep) {
const setup = this.setup;
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools && mcpTools.length > 0) {
const convertedTools = mcpTools.map((tool) => {
return {
functionDeclarations: {
name: tool.name,
description: tool.description,
parameters: tool.inputSchema,
}
};
}
);
// merge with any existing tools
setup.tools = [...convertedTools, ...(this.setup.tools || [])];
}
if (!await this._sendClientEvent(ep, {
setup,
})) {
this.logger.debug(this.setup, 'TaskLlmGoogle_S2S:_sendInitialMessage - sending session.update');
this.notifyTaskDone();
}
}
_registerHandlers(ep) {
this.addCustomEventListener(ep, LlmEvents_Google.Connect, this._onConnect.bind(this, ep));
this.addCustomEventListener(ep, LlmEvents_Google.ConnectFailure, this._onConnectFailure.bind(this, ep));
this.addCustomEventListener(ep, LlmEvents_Google.Disconnect, this._onDisconnect.bind(this, ep));
this.addCustomEventListener(ep, LlmEvents_Google.ServerEvent, this._onServerEvent.bind(this, ep));
}
_unregisterHandlers() {
this.removeCustomEventListeners();
}
_onError(ep, evt) {
this.logger.info({evt}, 'TaskLlmGoogle_S2S:_onError');
this.notifyTaskDone();
}
_onConnect(ep) {
this.logger.debug('TaskLlmGoogle_S2S:_onConnect');
this._sendInitialMessage(ep);
}
_onConnectFailure(_ep, evt) {
this.logger.info(evt, 'TaskLlmGoogle_S2S:_onConnectFailure');
this.results = {completionReason: 'connection failure'};
this.notifyTaskDone();
}
_onDisconnect(_ep, evt) {
this.logger.info(evt, 'TaskLlmGoogle_S2S:_onConnectFailure');
this.results = {completionReason: 'disconnect from remote end'};
this.notifyTaskDone();
}
async _onServerEvent(ep, evt) {
let endConversation = false;
this.logger.debug({evt}, 'TaskLlmGoogle_S2S:_onServerEvent');
const {toolCall /**toolCallCancellation*/} = evt;
if (toolCall) {
this.logger.debug({toolCall}, 'TaskLlmGoogle_S2S:_onServerEvent - toolCall');
if (!this.toolHook) {
this.logger.info({evt}, 'TaskLlmGoogle_S2S:_onServerEvent - no toolHook defined!');
}
else {
const {functionCalls} = toolCall;
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
const functionResponses = [];
if (mcpTools && mcpTools.length > 0) {
for (const functionCall of functionCalls) {
const {name, args, id} = functionCall;
const tool = mcpTools.find((tool) => tool.name === name);
if (tool) {
const response = await this.parent.mcpService.callMcpTool(name, args);
functionResponses.push({
response: {
output: response,
},
id
});
}
}
}
if (functionResponses && functionResponses.length > 0) {
this.logger.debug({functionResponses}, 'TaskLlmGoogle_S2S:_onServerEvent - function_call - mcp result');
this.processToolOutput(ep, 'tool_call_id', {
toolResponse: {
functionResponses
}
});
} else {
try {
await this.parent.sendToolHook('function_call_id', {type: 'toolCall', functionCalls});
} catch (err) {
this.logger.info({err, evt}, 'TaskLlmGoogle_S2S - error calling function');
this.results = {
completionReason: 'client error calling function',
error: err
};
endConversation = true;
}
}
}
}
this._sendLlmEvent('llm_event', evt);
if (endConversation) {
this.logger.info({results: this.results},
'TaskLlmGoogle_S2S:_onServerEvent - ending conversation due to error');
this.notifyTaskDone();
}
}
_sendLlmEvent(type, evt) {
/* check whether we should notify on this event */
if (this.includeEvents.length > 0 ? this.includeEvents.includes(type) : !this.excludeEvents.includes(type)) {
this.parent.sendEventHook(evt)
.catch((err) => this.logger.info({err}, 'TaskLlmGoogle_S2S:_onServerEvent - error sending event hook'));
}
}
async processLlmUpdate(ep, data, _callSid) {
try {
this.logger.debug({data, _callSid}, 'TaskLlmGoogle_S2S:processLlmUpdate');
await this._api(ep, [ep.uuid, ClientEvent, JSON.stringify(data)]);
} catch (err) {
this.logger.info({err, data}, 'TaskLlmGoogle_S2S:processLlmUpdate - Error processing LLM update');
}
}
async processToolOutput(ep, tool_call_id, data) {
try {
this.logger.debug({tool_call_id, data}, 'TaskLlmGoogle_S2S:processToolOutput');
const {toolResponse} = data;
if (!toolResponse) {
this.logger.info({data},
'TaskLlmGoogle_S2S:processToolOutput - invalid tool output, must be functionResponses');
}
else {
await this._api(ep, [ep.uuid, ClientEvent, JSON.stringify(data)]);
}
} catch (err) {
this.logger.info({err, data}, 'TaskLlmGoogle_S2S:processToolOutput - Error processing tool output');
}
}
}
module.exports = TaskLlmGoogle_S2S;

View File

@@ -235,6 +235,23 @@ class TaskLlmOpenAI_S2S extends Task {
/* send immediate session.update if present */
else if (this.session_update) {
if (this.parent.isMcpEnabled) {
this.logger.debug('TaskLlmOpenAI_S2S:_sendInitialMessage - mcp enabled');
const tools = await this.parent.mcpService.getAvailableMcpTools();
if (tools && tools.length > 0 && this.session_update) {
const convertedTools = tools.map((tool) => ({
name: tool.name,
type: 'function',
description: tool.description,
parameters: tool.inputSchema
}));
this.session_update.tools = [
...convertedTools,
...(this.session_update.tools || [])
];
}
}
obj = {type: 'session.update', session: this.session_update};
this.logger.debug({obj}, 'TaskLlmOpenAI_S2S:_sendInitialMessage - sending session.update');
if (!await this._sendClientEvent(ep, obj)) {
@@ -299,13 +316,37 @@ class TaskLlmOpenAI_S2S extends Task {
/* tool calls */
else if (type === 'response.output_item.done' && evt.item?.type === 'function_call') {
this.logger.debug({evt}, 'TaskLlmOpenAI_S2S:_onServerEvent - function_call');
if (!this.toolHook) {
const {name, call_id} = evt.item;
const args = JSON.parse(evt.item.arguments);
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools.some((tool) => tool.name === name)) {
this.logger.debug({call_id, name, args}, 'TaskLlmOpenAI_S2S:_onServerEvent - calling mcp tool');
try {
const res = await this.parent.mcpService.callMcpTool(name, args);
this.logger.debug({res}, 'TaskLlmOpenAI_S2S:_onServerEvent - function_call - mcp result');
this.processToolOutput(ep, call_id, {
type: 'conversation.item.create',
item: {
type: 'function_call_output',
call_id,
output: res.content[0]?.text || 'There is no output from the function call',
}
});
return;
} catch (err) {
this.logger.info({err, evt}, 'TaskLlmOpenAI_S2S - error calling function');
this.results = {
completionReason: 'client error calling mcp function',
error: err
};
endConversation = true;
}
}
else if (!this.toolHook) {
this.logger.warn({evt}, 'TaskLlmOpenAI_S2S:_onServerEvent - no toolHook defined!');
}
else {
const {name, call_id} = evt.item;
const args = JSON.parse(evt.item.arguments);
try {
await this.parent.sendToolHook(call_id, {name, args});
} catch (err) {

View File

@@ -39,6 +39,10 @@ class TaskLlmUltravox_S2S extends Task {
this.eventHook = this.data.eventHook;
this.toolHook = this.data.toolHook;
this.results = {
completionReason: 'normal conversation end'
};
/**
* only one of these will have items,
* if includeEvents, then these are the events to include
@@ -63,7 +67,50 @@ class TaskLlmUltravox_S2S extends Task {
}
}
/**
* Converts a JSON Schema to the dynamic parameters format used in the Ultravox API
* @param {Object} jsonSchema - A JSON Schema object defining parameters
* @param {string} locationDefault - Default location value for parameters (default: 'PARAMETER_LOCATION_BODY')
* @returns {Array} Array of dynamic parameters objects
*/
transformSchemaToParameters(jsonSchema, locationDefault = 'PARAMETER_LOCATION_BODY') {
if (jsonSchema.properties) {
const required = jsonSchema.required || [];
return Object.entries(jsonSchema.properties).map(([name]) => {
return {
name,
location: locationDefault,
required: required.includes(name)
};
});
}
return [];
}
async createCall() {
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools && mcpTools.length > 0) {
const convertedTools = mcpTools.map((tool) => {
return {
temporaryTool: {
modelToolName: tool.name,
description: tool.description,
dynamicParameters: this.transformSchemaToParameters(tool.inputSchema),
// use client tool that ultravox call tool via freeswitch module.
client: {}
}
};
}
);
// merge with any existing tools
this.data.llmOptions.selectedTools = [
...convertedTools,
...(this.data.llmOptions.selectedTools || [])
];
}
const payload = {
...this.data.llmOptions,
model: this.model,
@@ -85,10 +132,10 @@ class TaskLlmUltravox_S2S extends Task {
});
const data = await body.json();
if (statusCode !== 201 || !data?.joinUrl) {
this.logger.error({statusCode, data}, 'Ultravox Error registering call');
throw new Error(`Ultravox Error registering call: ${data.message}`);
this.logger.info({statusCode, data}, 'Ultravox Error registering call');
throw new Error(`Ultravox Error registering call:${statusCode} - ${data.detail}`);
}
this.logger.info({joinUrl: data.joinUrl}, 'Ultravox Call registered');
this.logger.debug({joinUrl: data.joinUrl}, 'Ultravox Call registered');
return data;
}
@@ -106,12 +153,11 @@ class TaskLlmUltravox_S2S extends Task {
async _startListening(cs, ep) {
this._registerHandlers(ep);
const data = await this.createCall();
const {joinUrl} = data;
// split the joinUrl into host and path
const {host, pathname, search} = new URL(joinUrl);
try {
const data = await this.createCall();
const {joinUrl} = data;
// split the joinUrl into host and path
const {host, pathname, search} = new URL(joinUrl);
const args = [ep.uuid, 'session.create', host, pathname + search];
await this._api(ep, args);
// Notify the application that the session has been created with detail information
@@ -120,7 +166,8 @@ class TaskLlmUltravox_S2S extends Task {
...data
});
} catch (err) {
this.logger.error({err}, 'TaskLlmUltraVox_S2S:_startListening');
this.logger.info({err}, 'TaskLlmUltraVox_S2S:_startListening - Error sending createCall');
this.results = {completionReason: `connection failure - ${err}`};
this.notifyTaskDone();
}
}
@@ -148,7 +195,7 @@ class TaskLlmUltravox_S2S extends Task {
}
_onConnect(ep) {
this.logger.debug('TaskLlmUltravox_S2S:_onConnect');
this.logger.info('TaskLlmUltravox_S2S:_onConnect');
}
_onConnectFailure(_ep, evt) {
this.logger.info(evt, 'TaskLlmUltravox_S2S:_onConnectFailure');
@@ -164,7 +211,7 @@ class TaskLlmUltravox_S2S extends Task {
async _onServerEvent(_ep, evt) {
let endConversation = false;
const type = evt.type;
this.logger.info({evt}, 'TaskLlmUltravox_S2S:_onServerEvent');
this.logger.debug({evt}, 'TaskLlmUltravox_S2S:_onServerEvent');
/* server errors of some sort */
if (type === 'error') {
@@ -178,12 +225,35 @@ class TaskLlmUltravox_S2S extends Task {
/* tool calls */
else if (type === 'client_tool_invocation') {
this.logger.debug({evt}, 'TaskLlmUltravox_S2S:_onServerEvent - function_call');
if (!this.toolHook) {
this.logger.warn({evt}, 'TaskLlmUltravox_S2S:_onServerEvent - no toolHook defined!');
const {toolName: name, invocationId: call_id, parameters: args} = evt;
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools.some((tool) => tool.name === name)) {
this.logger.debug({
name,
input: args
}, 'TaskLlmUltravox_S2S:_onServerEvent - function_call - mcp tool');
try {
const res = await this.parent.mcpService.callMcpTool(name, args);
this.logger.debug({res}, 'TaskLlmUltravox_S2S:_onServerEvent - function_call - mcp result');
this.processToolOutput(_ep, call_id, {
type: 'client_tool_result',
invocation_id: call_id,
result: res.content
});
return;
} catch (err) {
this.logger.info({err, evt}, 'TaskLlmUltravox_S2S - error calling mcp tool');
this.results = {
completionReason: 'client error calling mcp function',
error: err
};
endConversation = true;
}
} else if (!this.toolHook) {
this.logger.info({evt}, 'TaskLlmUltravox_S2S:_onServerEvent - no toolHook defined!');
}
else {
const {toolName: name, invocationId: call_id, parameters: args} = evt;
try {
await this.parent.sendToolHook(call_id, {name, args});
} catch (err) {
@@ -221,13 +291,14 @@ class TaskLlmUltravox_S2S extends Task {
if (!data.type || ![
'input_text_message'
].includes(data.type)) {
this.logger.info({data}, 'TaskLlmUltravox_S2S:processLlmUpdate - invalid mid-call request');
this.logger.info({data},
'TaskLlmUltravox_S2S:processLlmUpdate - invalid mid-call request, only input_text_message supported');
}
else {
await this._api(ep, [ep.uuid, ClientEvent, JSON.stringify(data)]);
}
} catch (err) {
this.logger.info({err}, 'TaskLlmUltravox_S2S:processLlmUpdate');
this.logger.info({err, data}, 'TaskLlmUltravox_S2S:processLlmUpdate - Error processing LLM update');
}
}
@@ -243,7 +314,7 @@ class TaskLlmUltravox_S2S extends Task {
await this._api(ep, [ep.uuid, ClientEvent, JSON.stringify(data)]);
}
} catch (err) {
this.logger.info({err}, 'TaskLlmUltravox_S2S:processToolOutput');
this.logger.info({err, data}, 'TaskLlmUltravox_S2S:processToolOutput - Error processing tool output');
}
}

View File

@@ -29,7 +29,7 @@ class TaskLlmVoiceAgent_S2S extends Task {
this.parent = parentTask;
this.vendor = this.parent.vendor;
this.model = this.parent.model;
this.model = this.parent.model || 'voice-agent';
this.auth = this.parent.auth;
this.connectionOptions = this.parent.connectOptions;
@@ -41,25 +41,25 @@ class TaskLlmVoiceAgent_S2S extends Task {
this.actionHook = this.data.actionHook;
this.eventHook = this.data.eventHook;
this.toolHook = this.data.toolHook;
const {settingsConfiguration} = this.data.llmOptions;
const {Settings} = this.data.llmOptions;
if (typeof settingsConfiguration !== 'object') {
throw new Error('llmOptions with an initial settingsConfiguration is required for VoiceAgent S2S');
if (typeof Settings !== 'object') {
throw new Error('llmOptions with an initial Settings is required for VoiceAgent S2S');
}
// eslint-disable-next-line no-unused-vars
const {audio, ...rest} = settingsConfiguration;
const cfg = this.settingsConfiguration = rest;
const {audio, ...rest} = Settings;
const cfg = this.Settings = rest;
if (!cfg.agent) throw new Error('llmOptions.settingsConfiguration.agent is required for VoiceAgent S2S');
if (!cfg.agent) throw new Error('llmOptions.Settings.agent is required for VoiceAgent S2S');
if (!cfg.agent.think) {
throw new Error('llmOptions.settingsConfiguration.agent.think is required for VoiceAgent S2S');
throw new Error('llmOptions.Settings.agent.think is required for VoiceAgent S2S');
}
if (!cfg.agent.think.model) {
throw new Error('llmOptions.settingsConfiguration.agent.think.model is required for VoiceAgent S2S');
if (!cfg.agent.think.provider?.model) {
throw new Error('llmOptions.Settings.agent.think.provider.model is required for VoiceAgent S2S');
}
if (!cfg.agent.think.provider?.type) {
throw new Error('llmOptions.settingsConfiguration.agent.think.provider.type is required for VoiceAgent S2S');
throw new Error('llmOptions.Settings.agent.think.provider.type is required for VoiceAgent S2S');
}
this.results = {
@@ -92,7 +92,7 @@ class TaskLlmVoiceAgent_S2S extends Task {
const {path} = this.connectionOptions || {};
if (path) return path;
return '/agent';
return '/v1/agent/converse';
}
async _api(ep, args) {
@@ -193,7 +193,20 @@ class TaskLlmVoiceAgent_S2S extends Task {
}
async _sendInitialMessage(ep) {
if (!await this._sendClientEvent(ep, this.settingsConfiguration)) {
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (mcpTools && mcpTools.length > 0 && this.Settings.agent?.think) {
const convertedTools = mcpTools.map((tool) => ({
name: tool.name,
description: tool.description,
parameters: tool.inputSchema
}));
this.Settings.agent.think.functions = [
...convertedTools,
...(this.Settings.agent.think?.functions || [])
];
}
if (!await this._sendClientEvent(ep, this.Settings)) {
this.notifyTaskDone();
}
}
@@ -254,17 +267,43 @@ class TaskLlmVoiceAgent_S2S extends Task {
/* tool calls */
else if (type === 'FunctionCallRequest') {
this.logger.debug({evt}, 'TaskLlmVoiceAgent_S2S:_onServerEvent - function_call');
if (!this.toolHook) {
const mcpTools = this.parent.isMcpEnabled ? await this.parent.mcpService.getAvailableMcpTools() : [];
if (!this.toolHook && mcpTools.length === 0) {
this.logger.warn({evt}, 'TaskLlmVoiceAgent_S2S:_onServerEvent - no toolHook defined!');
}
else {
const {function_name:name, function_call_id:call_id} = evt;
const args = evt.input;
} else {
const {functions} = evt;
const handledFunctions = [];
try {
await this.parent.sendToolHook(call_id, {name, args});
if (mcpTools && mcpTools.length > 0) {
for (const func of functions) {
const {name, arguments: args, id} = func;
const tool = mcpTools.find((tool) => tool.name === name);
if (tool) {
handledFunctions.push(name);
const response = await this.parent.mcpService.callMcpTool(name, JSON.parse(args));
this.logger.debug({response}, 'TaskLlmVoiceAgent_S2S:_onServerEvent - function_call - mcp result');
this.processToolOutput(_ep, id, {
data: {
type: 'FunctionCallResponse',
id,
name,
content: response.length > 0 ? response[0].text : 'There is no output from the function call'
}
});
}
}
}
for (const func of functions) {
const {name, arguments: args, id} = func;
if (!handledFunctions.includes(name)) {
await this.parent.sendToolHook(id, {name, args: JSON.parse(args)});
}
}
} catch (err) {
this.logger.info({err, evt}, 'TaskLlmVoiceAgent - error calling function');
this.logger.info({err, evt}, 'TaskLlmVoiceAgent_S2S:_onServerEvent - error calling function');
this.results = {
completionReason: 'client error calling function',
error: err

View File

@@ -1,7 +1,6 @@
const Task = require('./task');
const {TaskName, TaskPreconditions} = require('../utils/constants');
const { PlayFileNotFoundError } = require('../utils/error');
class TaskPlay extends Task {
constructor(logger, opts) {
super(logger, opts);
@@ -27,6 +26,7 @@ class TaskPlay extends Task {
let playbackSeconds = 0;
let playbackMilliseconds = 0;
let completed = !(this.timeoutSecs > 0 || this.loop);
cs.playingAudio = true;
if (this.timeoutSecs > 0) {
timeout = setTimeout(async() => {
completed = true;
@@ -40,6 +40,22 @@ class TaskPlay extends Task {
try {
this.notifyStatus({event: 'start-playback'});
while (!this.killed && (this.loop === 'forever' || this.loop--) && this.ep.connected) {
/* Listen for playback-start event and set up a one-time listener for uuid_break
* that will kill the audio playback if the taskIds match. This ensures that
* we only kill the currently playing audio and not audio from other tasks.
* As we are using stickyEventEmitter, even if the event is emitted before the listener is registered,
* the listener will receive the most recent event.
*/
ep.once('playback-start', (evt) => {
this.logger.debug({evt}, 'Play got playback-start');
this.cs.stickyEventEmitter.once('uuid_break', (t) => {
if (t?.taskId === this.taskId) {
this.logger.debug(`Play got kill-playback, executing uuid_break, taskId: ${t?.taskId}`);
this.ep.api('uuid_break', this.ep.uuid).catch((err) => this.logger.info(err, 'Error killing audio'));
this.notifyStatus({event: 'kill-playback'});
}
});
});
if (cs.isInConference) {
const {memberId, confName, confUuid} = cs;
if (Array.isArray(this.url)) {
@@ -87,15 +103,15 @@ class TaskPlay extends Task {
async kill(cs) {
super.kill(cs);
if (this.ep.connected && !this.playComplete) {
if (this.ep?.connected && !this.playComplete) {
this.logger.debug('TaskPlay:kill - killing audio');
if (cs.isInConference) {
const {memberId, confName} = cs;
this.killPlayToConfMember(this.ep, memberId, confName);
}
else {
this.notifyStatus({event: 'kill-playback'});
this.ep.api('uuid_break', this.ep.uuid).catch((err) => this.logger.info(err, 'Error killing audio'));
//this.ep.api('uuid_break', this.ep.uuid).catch((err) => this.logger.info(err, 'Error killing audio'));
cs.stickyEventEmitter.emit('uuid_break', this);
}
}
}

View File

@@ -107,7 +107,7 @@ class TaskSay extends TtsTask {
throw new SpeechCredentialError(
`No text-to-speech service credentials for ${vendor} with labels: ${label} have been configured`);
}
this.ep = ep;
try {
await this.setTtsStreamingChannelVars(vendor, language, voice, credentials, ep);
@@ -223,7 +223,19 @@ class TaskSay extends TtsTask {
});
ep.once('playback-stop', (evt) => {
this.logger.debug({evt}, 'Say got playback-stop');
if (evt.variable_tts_error) {
this.notifyStatus({event: 'stop-playback'});
this.notifiedPlayBackStop = true;
const tts_error = evt.variable_tts_error;
let response_code = 200;
// Check if any property ends with _response_code
for (const [key, value] of Object.entries(evt)) {
if (key.endsWith('_response_code')) {
response_code = parseInt(value, 10) || 200;
break;
}
}
if (tts_error) {
writeAlerts({
account_sid,
alert_type: AlertType.TTS_FAILURE,
@@ -232,7 +244,7 @@ class TaskSay extends TtsTask {
target_sid
}).catch((err) => this.logger.info({err}, 'Error generating alert for no tts'));
}
if (evt.variable_tts_cache_filename && !this.killed) {
if (!tts_error && response_code < 300 && evt.variable_tts_cache_filename && !this.killed) {
const text = parseTextFromSayString(this.text[segment]);
addFileToCache(evt.variable_tts_cache_filename, {
account_sid,
@@ -241,12 +253,14 @@ class TaskSay extends TtsTask {
voice,
engine,
model: this.model || this.model_id,
text
text,
instructions: this.instructions
}).catch((err) => this.logger.info({err}, 'Error adding file to cache'));
}
if (this._playResolve) {
evt.variable_tts_error ? this._playReject(new Error(evt.variable_tts_error)) : this._playResolve();
(tts_error || response_code >= 300) ? this._playReject(new Error(evt.variable_tts_error)) :
this._playResolve();
}
});
// wait for playback-stop event received to confirm if the playback is successful
@@ -292,8 +306,13 @@ class TaskSay extends TtsTask {
if (cs.isInConference) {
const {memberId, confName} = cs;
this.killPlayToConfMember(this.ep, memberId, confName);
}
else {
} else if (this.isStreamingTts) {
this.logger.debug('TaskSay:kill - clearing TTS stream for streaming audio');
cs.clearTtsStream();
} else {
if (!this.notifiedPlayBackStop) {
this.notifyStatus({event: 'stop-playback'});
}
this.notifyStatus({event: 'kill-playback'});
this.ep.api('uuid_break', this.ep.uuid);
}
@@ -314,6 +333,7 @@ class TaskSay extends TtsTask {
if (key.startsWith('variable_tts_')) {
let newKey = key.substring('variable_tts_'.length)
.replace('whisper_', 'whisper.')
.replace('nvidia_', 'nvidia.')
.replace('deepgram_', 'deepgram.')
.replace('playht_', 'playht.')
.replace('cartesia_', 'cartesia.')

View File

@@ -18,6 +18,11 @@ class TaskSipDecline extends Task {
super.exec(cs);
res.send(this.data.status, this.data.reason, {
headers: this.headers
}, (err) => {
if (!err) {
// Call was successfully declined
cs._callReleased();
}
});
cs.emit('callStatusChange', {
callStatus: CallStatus.Failed,

View File

@@ -5,6 +5,30 @@ const { TaskPreconditions, CobaltTranscriptionEvents } = require('../utils/const
const { SpeechCredentialError } = require('../utils/error');
const {JAMBONES_AWS_TRANSCRIBE_USE_GRPC} = require('../config');
/**
* "Please insert turns here: {{turns:4}}"
// -> { processed: 'Please insert turns here: {{turns}}', turns: 4 }
processTurnString("Please insert turns here: {{turns}}"));
// -> { processed: 'Please insert turns here: {{turns}}', turns: null }
*/
const processTurnString = (input) => {
const regex = /\{\{turns(?::(\d+))?\}\}/;
const match = input.match(regex);
if (!match) {
return {
processed: input,
turns: null
};
}
const turns = match[1] ? parseInt(match[1], 10) : null;
const processed = input.replace(regex, '{{turns}}');
return { processed, turns };
};
class SttTask extends Task {
constructor(logger, data, parentTask) {
@@ -290,6 +314,57 @@ class SttTask extends Task {
});
}
formatOpenAIPrompt(cs, {prompt, hintsTemplate, conversationHistoryTemplate, hints}) {
let conversationHistoryPrompt, hintsPrompt;
/* generate conversation history from template */
if (conversationHistoryTemplate) {
const {processed, turns} = processTurnString(conversationHistoryTemplate);
this.logger.debug({processed, turns}, 'SttTask: processed conversation history template');
conversationHistoryPrompt = cs.getFormattedConversation(turns || 4);
//this.logger.debug({conversationHistoryPrompt}, 'SttTask: conversation history');
if (conversationHistoryPrompt) {
conversationHistoryPrompt = processed.replace('{{turns}}', `\n${conversationHistoryPrompt}\nuser: `);
}
}
/* generate hints from template */
if (hintsTemplate && Array.isArray(hints) && hints.length > 0) {
hintsPrompt = hintsTemplate.replace('{{hints}}', hints);
}
/* combine into final prompt */
let finalPrompt = prompt || '';
if (hintsPrompt) {
finalPrompt = `${finalPrompt}\n${hintsPrompt}`;
}
if (conversationHistoryPrompt) {
finalPrompt = `${finalPrompt}\n${conversationHistoryPrompt}`;
}
this.logger.debug({
finalPrompt,
hints,
hintsPrompt,
conversationHistoryTemplate,
conversationHistoryPrompt
}, 'SttTask: formatted OpenAI prompt');
return finalPrompt?.trimStart();
}
/* some STT engines will keep listening after a final response, so no need to restart */
doesVendorContinueListeningAfterFinalTranscript(vendor) {
return (vendor.startsWith('custom:') || [
'soniox',
'aws',
'microsoft',
'deepgram',
'google',
'speechmatics',
'openai',
].includes(vendor));
}
_onCompileContext(ep, key, evt) {
const {addKey} = this.cs.srf.locals.dbHelpers;
this.logger.debug({evt}, `received cobalt compile context event, will cache under ${key}`);

View File

@@ -19,6 +19,7 @@ class Task extends Emitter {
this.data = data;
this.actionHook = this.data.actionHook;
this.id = data.id;
this.taskId = uuidv4();
this._killInProgress = false;
this._completionPromise = new Promise((resolve) => this._completionResolver = resolve);

View File

@@ -14,6 +14,7 @@ const {
TranscribeStatus,
AssemblyAiTranscriptionEvents,
VoxistTranscriptionEvents,
OpenAITranscriptionEvents,
VerbioTranscriptionEvents,
SpeechmaticsTranscriptionEvents
} = require('../utils/constants.json');
@@ -30,7 +31,6 @@ class TaskTranscribe extends SttTask {
this.transcriptionHook = this.data.transcriptionHook;
this.translationHook = this.data.translationHook;
this.earlyMedia = this.data.earlyMedia === true || (parentTask && parentTask.earlyMedia);
if (this.data.recognizer) {
this.interim = !!this.data.recognizer.interim;
this.separateRecognitionPerChannel = this.data.recognizer.separateRecognitionPerChannel;
@@ -105,7 +105,7 @@ class TaskTranscribe extends SttTask {
if (cs.hasGlobalSttHints) {
const {hints, hintsBoost} = cs.globalSttHints;
this.data.recognizer.hints = this.data.recognizer.hints.concat(hints);
this.data.recognizer.hints = this.data.recognizer?.hints?.concat(hints);
if (!this.data.recognizer.hintsBoost && hintsBoost) this.data.recognizer.hintsBoost = hintsBoost;
this.logger.debug({hints: this.data.recognizer.hints, hintsBoost: this.data.recognizer.hintsBoost},
'Transcribe:exec - applying global sttHints');
@@ -330,6 +330,20 @@ class TaskTranscribe extends SttTask {
this._onSpeechmaticsError.bind(this, cs, ep));
break;
case 'openai':
this.bugname = `${this.bugname_prefix}openai_transcribe`;
this.addCustomEventListener(
ep, OpenAITranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
this.addCustomEventListener(ep, OpenAITranscriptionEvents.Error,
this._onOpenAIErrror.bind(this, cs, ep));
this.modelSupportsConversationTracking = opts.OPENAI_MODEL !== 'whisper-1';
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
@@ -365,6 +379,25 @@ class TaskTranscribe extends SttTask {
async _transcribe(ep) {
this.logger.debug(
`TaskTranscribe:_transcribe - starting transcription vendor ${this.vendor} bugname ${this.bugname}`);
/* special feature for openai: we can provide a prompt that includes recent conversation history */
let prompt;
if (this.vendor === 'openai') {
if (this.modelSupportsConversationTracking) {
prompt = this.formatOpenAIPrompt(this.cs, {
prompt: this.data.recognizer?.openaiOptions?.prompt,
hintsTemplate: this.data.recognizer?.openaiOptions?.promptTemplates?.hintsTemplate,
// eslint-disable-next-line max-len
conversationHistoryTemplate: this.data.recognizer?.openaiOptions?.promptTemplates?.conversationHistoryTemplate,
hints: this.data.recognizer?.hints,
});
this.logger.debug({prompt}, 'Gather:_startTranscribing - created an openai prompt');
}
else if (this.data.recognizer?.hints?.length > 0) {
prompt = this.data.recognizer?.hints.join(', ');
}
}
await ep.startTranscription({
vendor: this.vendor,
interim: this.interim ? true : false,
@@ -456,8 +489,9 @@ class TaskTranscribe extends SttTask {
this._startAsrTimer(channel);
/* some STT engines will keep listening after a final response, so no need to restart */
if (!['soniox', 'aws', 'microsoft', 'deepgram', 'google', 'speechmatics']
.includes(this.vendor)) this._startTranscribing(cs, ep, channel);
if (!this.doesVendorContinueListeningAfterFinalTranscript(this.vendor)) {
this._startTranscribing(cs, ep, channel);
}
}
else {
if (this.vendor === 'soniox') {
@@ -480,9 +514,7 @@ class TaskTranscribe extends SttTask {
this.logger.debug({evt}, 'TaskTranscribe:_onTranscription - sending final transcript');
this._resolve(channel, evt);
/* some STT engines will keep listening after a final response, so no need to restart */
if (!['soniox', 'aws', 'microsoft', 'deepgram', 'google', 'speechmatics'].includes(this.vendor) &&
!this.vendor.startsWith('custom:')) {
if (!this.doesVendorContinueListeningAfterFinalTranscript(this.vendor)) {
this.logger.debug('TaskTranscribe:_onTranscription - restarting transcribe');
this._startTranscribing(cs, ep, channel);
}
@@ -621,12 +653,21 @@ class TaskTranscribe extends SttTask {
}
_onMaxDurationExceeded(cs, ep, channel) {
this.logger.debug(`TaskTranscribe:_onMaxDurationExceeded on channel ${channel}`);
this.restartDueToError(ep, channel, 'Max duration exceeded');
}
_onMaxBufferExceeded(cs, ep, channel) {
this.restartDueToError(ep, channel, 'Max buffer exceeded');
}
restartDueToError(ep, channel, reason) {
this.logger.debug(`TaskTranscribe:${reason} on channel ${channel}`);
if (this.paused) return;
if (this.childSpan[channel - 1] && this.childSpan[channel - 1].span) {
this.childSpan[channel - 1].span.setAttributes({
channel,
'stt.resolve': 'max duration exceeded',
'stt.resolve': reason,
'stt.label': this.label || 'None',
});
this.childSpan[channel - 1].span.end();
@@ -683,6 +724,14 @@ class TaskTranscribe extends SttTask {
return;
}
this.logger.info({evt}, 'TaskTranscribe:_onJambonzError');
if (this.vendor === 'microsoft' &&
evt.error?.includes('Due to service inactivity, the client buffer exceeded maximum size. Resetting the buffer')) {
let channel = 1;
if (this.ep !== _ep) {
channel = 2;
}
return this._onMaxBufferExceeded(cs, _ep, channel);
}
if (this.paused) return;
const {writeAlerts, AlertType} = cs.srf.locals;
@@ -733,6 +782,12 @@ class TaskTranscribe extends SttTask {
this._onVendorError(cs, _ep, {error: JSON.stringify(e)});
}
async _onOpenAIErrror(cs, _ep, evt) {
// eslint-disable-next-line no-unused-vars
const {message, ...e} = evt;
this._onVendorError(cs, _ep, {error: JSON.stringify(e)});
}
_startAsrTimer(channel) {
if (this.vendor === 'deepgram') return; // no need
assert(this.isContinuousAsr);

View File

@@ -21,6 +21,7 @@ class TtsTask extends Task {
this.synthesizer = this.data.synthesizer || {};
this.disableTtsCache = this.data.disableTtsCache;
this.options = this.synthesizer.options || {};
this.instructions = this.data.instructions;
}
async exec(cs) {
@@ -42,6 +43,11 @@ class TtsTask extends Task {
}
}
}
const fullText = Array.isArray(this.text) ? this.text.join(' ') : this.text;
if (fullText.length > 0) {
cs.emit('botSaid', fullText);
}
}
getTtsVendorData(cs) {
@@ -60,7 +66,6 @@ class TtsTask extends Task {
async setTtsStreamingChannelVars(vendor, language, voice, credentials, ep) {
const {api_key, model_id, custom_tts_streaming_url, auth_token} = credentials;
const {stability, similarity_boost, use_speaker_boost, style} = this.options;
let obj;
this.logger.debug({credentials},
@@ -82,6 +87,7 @@ class TtsTask extends Task {
};
break;
case 'elevenlabs':
const {stability, similarity_boost, use_speaker_boost, style, speed} = this.options.voice_settings || {};
obj = {
ELEVENLABS_API_KEY: api_key,
ELEVENLABS_TTS_STREAMING_MODEL_ID: model_id,
@@ -91,7 +97,14 @@ class TtsTask extends Task {
...(stability && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_STABILITY: stability}),
...(similarity_boost && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_SIMILARITY_BOOST: similarity_boost}),
...(use_speaker_boost && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_USE_SPEAKER_BOOST: use_speaker_boost}),
...(style && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_STYLE: style})
...(style && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_STYLE: style}),
// speed has value 0.7 to 1.2, 1.0 is default, make sure we send the value event it's 0
...(speed !== null && speed !== undefined && {ELEVENLABS_TTS_STREAMING_VOICE_SETTINGS_SPEED: `${speed}`}),
...(this.options.pronunciation_dictionary_locators &&
Array.isArray(this.options.pronunciation_dictionary_locators) && {
ELEVENLABS_TTS_STREAMING_PRONUNCIATION_DICTIONARY_LOCATORS:
JSON.stringify(this.options.pronunciation_dictionary_locators)
}),
};
break;
case 'rimelabs':
@@ -125,7 +138,7 @@ class TtsTask extends Task {
throw new Error(`vendor ${vendor} is not supported for tts streaming yet`);
}
}
this.logger.info({vendor, credentials, obj}, 'setTtsStreamingChannelVars');
this.logger.debug({vendor, credentials, obj}, 'setTtsStreamingChannelVars');
await ep.set(obj);
}
@@ -152,7 +165,6 @@ class TtsTask extends Task {
} else if (vendor === 'deepgram') {
this.model = voice;
}
this.model_id = credentials.model_id;
/* allow for microsoft custom region voice and api_key to be specified as an override */
if (vendor === 'microsoft' && this.options.deploymentId) {
@@ -193,8 +205,12 @@ class TtsTask extends Task {
};
}
}
} else if (vendor === 'cartesia') {
credentials.model_id = this.options.model_id || credentials.model_id;
}
this.model_id = credentials.model_id;
/**
* note on cache_speech_handles. This was found to be risky.
* It can cause a crash in the following sequence on a single call:
@@ -216,7 +232,7 @@ class TtsTask extends Task {
//cs.currentTtsVendor = vendor;
if (!preCache && !this._disableTracing)
this.logger.info({vendor, language, voice, model: this.model}, 'TaskSay:exec');
this.logger.debug({vendor, language, voice, model: this.model}, 'TaskSay:exec');
try {
if (!credentials) {
writeAlerts({
@@ -247,6 +263,7 @@ class TtsTask extends Task {
const {filePath, servedFromCache, rtt} = await synthAudio(stats, {
account_sid,
text,
instructions: this.instructions,
vendor,
language,
voice,

View File

@@ -51,7 +51,7 @@ class ActionHookDelayProcessor extends Emitter {
this.logger.debug({opts}, 'ActionHookDelayProcessor#init');
this.actions = opts.actions;
this.retries = opts.retries || 0;
this.retries = Math.max((opts.retries || 1), opts.actions.length);
this.noResponseTimeout = opts.noResponseTimeout;
this.noResponseGiveUpTimeout = opts.noResponseGiveUpTimeout;
this.giveUpActions = opts.giveUpActions;

View File

@@ -32,7 +32,7 @@
"Tag": "tag",
"Transcribe": "transcribe"
},
"AllowedSipRecVerbs": ["answer", "config", "gather", "transcribe", "listen", "tag"],
"AllowedSipRecVerbs": ["answer", "config", "gather", "transcribe", "listen", "tag", "hangup", "sip:decline"],
"AllowedConfirmSessionVerbs": ["config", "gather", "plays", "say", "tag"],
"CallStatus": {
"Trying": "trying",
@@ -137,6 +137,18 @@
"Connect": "speechmatics_transcribe::connect",
"Error": "speechmatics_transcribe::error"
},
"OpenAITranscriptionEvents": {
"Transcription": "openai_transcribe::transcription",
"Translation": "openai_transcribe::translation",
"SpeechStarted": "openai_transcribe::speech_started",
"SpeechStopped": "openai_transcribe::speech_stopped",
"PartialTranscript": "openai_transcribe::partial_transcript",
"Info": "openai_transcribe::info",
"RecognitionStarted": "openai_transcribe::recognition_started",
"ConnectFailure": "openai_transcribe::connect_failed",
"Connect": "openai_transcribe::connect",
"Error": "openai_transcribe::error"
},
"JambonzTranscriptionEvents": {
"Transcription": "jambonz_transcribe::transcription",
"ConnectFailure": "jambonz_transcribe::connect_failed",
@@ -182,6 +194,13 @@
"Disconnect": "openai_s2s::disconnect",
"ServerEvent": "openai_s2s::server_event"
},
"LlmEvents_Google": {
"Error": "error",
"Connect": "google_s2s::connect",
"ConnectFailure": "google_s2s::connect_failed",
"Disconnect": "google_s2s::disconnect",
"ServerEvent": "google_s2s::server_event"
},
"LlmEvents_Elevenlabs": {
"Error": "error",
"Connect": "elevenlabs_s2s::connect",

View File

@@ -142,6 +142,11 @@ const speechMapper = (cred) => {
obj.api_key = o.api_key;
obj.speechmatics_stt_uri = o.speechmatics_stt_uri;
}
else if ('openai' === obj.vendor) {
const o = JSON.parse(decrypt(credential));
obj.api_key = o.api_key;
obj.model_id = o.model_id;
}
else if (obj.vendor.startsWith('custom:')) {
const o = JSON.parse(decrypt(credential));
obj.auth_token = o.auth_token;

View File

@@ -108,7 +108,7 @@ class HttpRequestor extends BaseRequestor {
assert(HookMsgTypes.includes(type));
const payload = params ? snakeCaseKeys(params, ['customerData', 'sip']) : null;
const payload = params ? snakeCaseKeys(params, ['customerData', 'sip', 'env_vars', 'args']) : null;
const url = hook.url || hook;
const method = hook.method || 'POST';
let buf = '';
@@ -219,7 +219,7 @@ class HttpRequestor extends BaseRequestor {
const rtt = this._roundTrip(startAt);
if (buf) this.stats.histogram('app.hook.response_time', rtt, ['hook_type:app']);
if (buf && Array.isArray(buf)) {
if (buf && (Array.isArray(buf) || type == 'llm:tool-call')) {
this.logger.info({response: buf}, `HttpRequestor:request ${method} ${url} succeeded in ${rtt}ms`);
return buf;
}

View File

@@ -31,18 +31,26 @@ function getLocalIp() {
return '127.0.0.1'; // Fallback to localhost if no suitable interface found
}
function initMS(logger, wrapper, ms) {
function initMS(logger, wrapper, ms, {
onFreeswitchConnect,
onFreeswitchDisconnect
}) {
Object.assign(wrapper, {ms, active: true, connects: 1});
logger.info(`connected to freeswitch at ${ms.address}`);
onFreeswitchConnect(wrapper);
ms.conn
.on('esl::end', () => {
wrapper.active = false;
wrapper.connects = 0;
logger.info(`lost connection to freeswitch at ${ms.address}`);
onFreeswitchDisconnect(wrapper);
ms.removeAllListeners();
})
.on('esl::ready', () => {
if (wrapper.connects > 0) {
logger.info(`connected to freeswitch at ${ms.address}`);
logger.info(`esl::ready connected to freeswitch at ${ms.address}`);
}
wrapper.connects = 1;
wrapper.active = true;
@@ -56,7 +64,10 @@ function initMS(logger, wrapper, ms) {
});
}
function installSrfLocals(srf, logger) {
function installSrfLocals(srf, logger, {
onFreeswitchConnect = () => {},
onFreeswitchDisconnect = () => {}
}) {
logger.debug('installing srf locals');
assert(!srf.locals.dbHelpers);
const {tracer} = srf.locals.otel;
@@ -91,7 +102,10 @@ function installSrfLocals(srf, logger) {
mediaservers.push(val);
try {
const ms = await mrf.connect(fs);
initMS(logger, val, ms);
initMS(logger, val, ms, {
onFreeswitchConnect,
onFreeswitchDisconnect
});
}
catch (err) {
logger.info({err}, `failed connecting to freeswitch at ${fs.address}, will retry shortly: ${err.message}`);
@@ -102,9 +116,15 @@ function installSrfLocals(srf, logger) {
for (const val of mediaservers) {
if (val.connects === 0) {
try {
// make sure all listeners are removed before reconnecting
val.ms?.disconnect();
val.ms = null;
logger.info({mediaserver: val.opts}, 'Retrying initial connection to media server');
const ms = await mrf.connect(val.opts);
initMS(logger, val, ms);
initMS(logger, val, ms, {
onFreeswitchConnect,
onFreeswitchDisconnect
});
} catch (err) {
logger.info({err}, `failed connecting to freeswitch at ${val.opts.address}, will retry shortly`);
}

103
lib/utils/llm-mcp.js Normal file
View File

@@ -0,0 +1,103 @@
const { Client } = require('@modelcontextprotocol/sdk/client/index.js');
class LlmMcpService {
constructor(logger, mcpServers) {
this.logger = logger;
this.mcpServers = mcpServers || [];
this.mcpClients = [];
}
// make sure we call init() before using any of the mcp clients
// this is to ensure that we have a valid connection to the MCP server
// and that we have collected the available tools.
async init() {
if (this.mcpClients.length > 0) {
return;
}
const { SSEClientTransport } = await import('@modelcontextprotocol/sdk/client/sse.js');
for (const server of this.mcpServers) {
const { url } = server;
if (url) {
try {
const transport = new SSEClientTransport(new URL(url), {});
const client = new Client({ name: 'Jambonz MCP Client', version: '1.0.0' });
await client.connect(transport);
// collect available tools
const { tools } = await client.listTools();
this.mcpClients.push({
url,
client,
tools
});
} catch (err) {
this.logger.error(`LlmMcpService: Failed to connect to MCP server at ${url}: ${err.message}`);
}
}
}
}
async getAvailableMcpTools() {
// returns a list of available tools from all MCP clients
const tools = [];
for (const mcpClient of this.mcpClients) {
const {tools: availableTools} = mcpClient;
if (availableTools) {
tools.push(...availableTools);
}
}
return tools;
}
async getMcpClientByToolName(name) {
for (const mcpClient of this.mcpClients) {
const { tools } = mcpClient;
if (tools && tools.some((tool) => tool.name === name)) {
return mcpClient.client;
}
}
return null;
}
async getMcpClientByToolId(id) {
for (const mcpClient of this.mcpClients) {
const { tools } = mcpClient;
if (tools && tools.some((tool) => tool.id === id)) {
return mcpClient.client;
}
}
return null;
}
async callMcpTool(name, input) {
const client = await this.getMcpClientByToolName(name);
if (client) {
try {
const result = await client.callTool({
name,
arguments: input,
});
this.logger.debug({result}, 'LlmMcpService - result');
return result;
} catch (err) {
this.logger.error({err}, 'LlmMcpService - error calling tool');
throw err;
}
}
}
async close() {
for (const mcpClient of this.mcpClients) {
const { client } = mcpClient;
if (client) {
await client.close();
this.logger.debug({url: mcpClient.url}, 'LlmMcpService - mcp client closed');
}
}
this.mcpClients = [];
}
}
module.exports = LlmMcpService;

View File

@@ -15,7 +15,7 @@ const RootSpan = require('./call-tracer');
const uuidv4 = require('uuid-random');
const HttpRequestor = require('./http-requestor');
const WsRequestor = require('./ws-requestor');
const {makeOpusFirst} = require('./sdp-utils');
const {makeOpusFirst, removeVideoSdp} = require('./sdp-utils');
const {
JAMBONES_USE_FREESWITCH_TIMER_FD,
JAMBONES_MEDIA_TIMEOUT_MS,
@@ -154,15 +154,21 @@ class SingleDialer extends Emitter {
return;
}
let lastSdp;
const connectStream = async(remoteSdp) => {
const connectStream = async(remoteSdp, isVideoCall) => {
if (remoteSdp === lastSdp) return;
if (process.env.JAMBONES_VIDEO_CALLS_ENABLED_IN_FS && !isVideoCall) {
remoteSdp = removeVideoSdp(remoteSdp);
}
lastSdp = remoteSdp;
return this.ep.modify(remoteSdp);
};
let localSdp = this.ep.local.sdp;
if (process.env.JAMBONES_VIDEO_CALLS_ENABLED_IN_FS && !opts.isVideoCall) {
localSdp = removeVideoSdp(localSdp);
}
Object.assign(opts, {
proxy: `sip:${this.sbcAddress}`,
localSdp: opts.opusFirst ? makeOpusFirst(this.ep.local.sdp) : this.ep.local.sdp
localSdp: opts.opusFirst ? makeOpusFirst(localSdp) : localSdp
});
if (this.target.auth) opts.auth = this.target.auth;
inviteSpan = this.startSpan('invite', {
@@ -224,13 +230,13 @@ class SingleDialer extends Emitter {
status.callStatus = CallStatus.EarlyMedia;
this.emit('earlyMedia');
}
connectStream(prov.body);
connectStream(prov.body, opts.isVideoCall);
}
else status.callStatus = CallStatus.Ringing;
this.emit('callStatusChange', status);
}
});
await connectStream(this.dlg.remote.sdp);
await connectStream(this.dlg.remote.sdp, opts.isVideoCall);
this.dlg.callSid = this.callSid;
this.inviteInProgress = null;
this.emit('callStatusChange', {
@@ -273,7 +279,12 @@ class SingleDialer extends Emitter {
this.logger.info('dial is onhold, emit event');
this.emit('reinvite', req, res);
} else {
const newSdp = await this.ep.modify(req.body);
let newSdp = await this.ep.modify(req.body);
// in case of reINVITE if video call is enabled in FS and the call is not a video call,
// remove video media from the SDP
if (process.env.JAMBONES_VIDEO_CALLS_ENABLED_IN_FS && !this.opts?.isVideoCall) {
newSdp = removeVideoSdp(newSdp);
}
res.send(200, {body: newSdp});
this.logger.info({offer: req.body, answer: newSdp}, 'SingleDialer:exec: handling reINVITE');
}
@@ -559,7 +570,8 @@ function placeOutdial({
}) {
const myOpts = deepcopy(opts);
const sd = new SingleDialer({
logger, sbcAddress, target, myOpts, application, callInfo, accountInfo, rootSpan, startSpan, dialTask, onHoldMusic
logger, sbcAddress, target, opts: myOpts, application, callInfo,
accountInfo, rootSpan, startSpan, dialTask, onHoldMusic
});
sd.exec(srf, ms, myOpts);
return sd;

View File

@@ -35,6 +35,12 @@ const makeOpusFirst = (sdp) => {
}
return sdpTransform.write(parsedSdp);
};
const removeVideoSdp = (sdp) => {
const parsedSdp = sdpTransform.parse(sdp);
// Filter out video media sections, keeping only non-video media
parsedSdp.media = parsedSdp.media.filter((media) => media.type !== 'video');
return sdpTransform.write(parsedSdp);
};
const extractSdpMedia = (sdp) => {
const parsedSdp1 = sdpTransform.parse(sdp);
@@ -54,5 +60,6 @@ module.exports = {
mergeSdpMedia,
extractSdpMedia,
isOpusFirst,
makeOpusFirst
makeOpusFirst,
removeVideoSdp
};

View File

@@ -0,0 +1,70 @@
const EventEmitter = require('events');
/**
* A specialized EventEmitter that caches the most recent event emissions.
* When new listeners are added, they immediately receive the most recent
* event if it was previously emitted. This is useful for handling state
* changes where late subscribers need to know the current state.
*
* Features:
* - Caches the most recent emission for each event type
* - New listeners immediately receive the cached event if available
* - Supports both regular (on) and one-time (once) listeners
* - Maintains compatibility with Node's EventEmitter interface
*/
class StickyEventEmitter extends EventEmitter {
constructor() {
super();
this._eventCache = new Map();
this._onceListeners = new Map(); // For storing once listeners if needed
}
destroy() {
this._eventCache.clear();
this._onceListeners.clear();
this.removeAllListeners();
}
emit(event, ...args) {
// Store the event and its args
this._eventCache.set(event, args);
// If there are any 'once' listeners waiting, call them
if (this._onceListeners.has(event)) {
const listeners = this._onceListeners.get(event);
for (const listener of listeners) {
listener(...args);
}
if (this.onSuccess) {
this.onSuccess();
}
this._onceListeners.delete(event);
}
return super.emit(event, ...args);
}
on(event, listener) {
if (this._eventCache.has(event)) {
listener(...this._eventCache.get(event));
}
return super.on(event, listener);
}
once(event, listener) {
if (this._eventCache.has(event)) {
listener(...this._eventCache.get(event));
if (this.onSuccess) {
this.onSuccess();
}
} else {
// Store listener in case emit comes before
if (!this._onceListeners.has(event)) {
this._onceListeners.set(event, []);
}
this._onceListeners.get(event).push(listener);
super.once(event, listener); // Also attach to native once
}
return this;
}
}
module.exports = StickyEventEmitter;

View File

@@ -117,7 +117,16 @@ const stickyVars = {
'SPEECHMATICS_SPEECH_HINTS',
'SPEECHMATICS_TRANSLATION_LANGUAGES',
'SPEECHMATICS_TRANSLATION_PARTIALS'
]
],
openai: [
'OPENAI_API_KEY',
'OPENAI_MODEL',
'OPENAI_INPUT_AUDIO_NOISE_REDUCTION',
'OPENAI_TURN_DETECTION_TYPE',
'OPENAI_TURN_DETECTION_THRESHOLD',
'OPENAI_TURN_DETECTION_PREFIX_PADDING_MS',
'OPENAI_TURN_DETECTION_SILENCE_DURATION_MS',
],
};
/**
@@ -304,13 +313,18 @@ const normalizeDeepgram = (evt, channel, language, shortUtterance) => {
confidence: alt.confidence,
transcript: alt.transcript,
}));
/**
* Some models (nova-2-general) return the detected language in the
* alternatives.languages array if the language is set as multi.
* If the language is detected, we use it as the language_code.
*/
const detectedLanguage = evt.channel?.alternatives?.[0]?.languages?.[0];
/**
* note difference between is_final and speech_final in Deepgram:
* https://developers.deepgram.com/docs/understand-endpointing-interim-results
*/
return {
language_code: language,
language_code: detectedLanguage || language,
channel_tag: channel,
is_final: shortUtterance ? evt.is_final : evt.speech_final,
alternatives: alternatives.length ? [alternatives[0]] : [],
@@ -566,6 +580,35 @@ const normalizeSpeechmatics = (evt, channel, language) => {
return obj;
};
const calculateConfidence = (logprobsArray) => {
// Sum the individual log probabilities
const totalLogProb = logprobsArray.reduce((sum, tokenInfo) => sum + tokenInfo.logprob, 0);
// Convert the total log probability back to a regular probability
const confidence = Math.exp(totalLogProb);
return confidence;
};
const normalizeOpenAI = (evt, channel, language) => {
const copy = JSON.parse(JSON.stringify(evt));
const obj = {
language_code: language,
channel_tag: channel,
is_final: true,
alternatives: [
{
transcript: evt.transcript,
confidence: evt.logprobs ? calculateConfidence(evt.logprobs) : 1.0,
}
],
vendor: {
name: 'openai',
evt: copy
}
};
return obj;
};
module.exports = (logger) => {
const normalizeTranscription = (evt, vendor, channel, language, shortUtterance, punctuation) => {
@@ -597,6 +640,8 @@ module.exports = (logger) => {
return normalizeVerbio(evt, channel, language);
case 'speechmatics':
return normalizeSpeechmatics(evt, channel, language);
case 'openai':
return normalizeOpenAI(evt, channel, language);
default:
if (vendor.startsWith('custom:')) {
return normalizeCustom(evt, channel, language, vendor);
@@ -784,7 +829,7 @@ module.exports = (logger) => {
};
}
else if ('deepgram' === vendor) {
let {model} = rOpts;
let model = rOpts.deepgramOptions?.model || rOpts.model;
const {deepgramOptions = {}} = rOpts;
const deepgramUri = deepgramOptions.deepgramSttUri || sttCredentials.deepgram_stt_uri;
const useTls = deepgramOptions.deepgramSttUseTls || sttCredentials.deepgram_stt_use_tls;
@@ -963,6 +1008,36 @@ module.exports = (logger) => {
{VOXIST_API_KEY: sttCredentials.api_key},
};
}
else if ('openai' === vendor) {
const {openaiOptions = {}} = rOpts;
const model = openaiOptions.model || rOpts.model || sttCredentials.model_id || 'whisper-1';
const apiKey = openaiOptions.apiKey || sttCredentials.api_key;
opts = {
OPENAI_MODEL: model,
OPENAI_API_KEY: apiKey,
...opts,
...(openaiOptions.prompt && {OPENAI_PROMPT: openaiOptions.prompt}),
...(openaiOptions.input_audio_noise_reduction &&
{OPENAI_INPUT_AUDIO_NOISE_REDUCTION: openaiOptions.input_audio_noise_reduction}),
};
if (openaiOptions.turn_detection) {
opts = {
...opts,
OPENAI_TURN_DETECTION_TYPE: openaiOptions.turn_detection.type,
...(openaiOptions.turn_detection.threshold && {
OPENAI_TURN_DETECTION_THRESHOLD: openaiOptions.turn_detection.threshold
}),
...(openaiOptions.turn_detection.prefix_padding_ms && {
OPENAI_TURN_DETECTION_PREFIX_PADDING_MS: openaiOptions.turn_detection.prefix_padding_ms
}),
...(openaiOptions.turn_detection.silence_duration_ms && {
OPENAI_TURN_DETECTION_SILENCE_DURATION_MS: openaiOptions.turn_detection.silence_duration_ms
}),
};
}
}
else if ('verbio' === vendor) {
const {verbioOptions = {}} = rOpts;
opts = {

View File

@@ -8,7 +8,7 @@ const {
const MAX_CHUNK_SIZE = 1800;
const HIGH_WATER_BUFFER_SIZE = 1000;
const LOW_WATER_BUFFER_SIZE = 200;
const TIMEOUT_RETRY_MSECS = 3000;
const TIMEOUT_RETRY_MSECS = 1000; // 1 second
const isWhitespace = (str) => /^\s*$/.test(str);
@@ -377,6 +377,7 @@ class TtsStreamingBuffer extends Emitter {
_onTimeout() {
this.logger.debug('TtsStreamingBuffer:_onTimeout Timeout waiting for sentence boundary');
this.timer = null;
// Check if new text has been added since the timer was set.
const now = Date.now();
if (now - this.lastUpdateTime < TIMEOUT_RETRY_MSECS) {
@@ -384,7 +385,6 @@ class TtsStreamingBuffer extends Emitter {
this._setTimerIfNeeded();
return;
}
this.timer = null;
this._feedQueue(true);
}

View File

@@ -132,7 +132,7 @@ class WsRequestor extends BaseRequestor {
assert(this.ws);
/* prepare and send message */
let payload = params ? snakeCaseKeys(params, ['customerData', 'sip']) : null;
let payload = params ? snakeCaseKeys(params, ['customerData', 'sip', 'env_vars', 'args']) : null;
if (type === 'session:new' || type === 'session:adulting') this._sessionData = payload;
if (type === 'session:reconnect') payload = this._sessionData;
assert.ok(url, 'WsRequestor:request url was not provided');
@@ -146,7 +146,9 @@ class WsRequestor extends BaseRequestor {
type,
msgid,
call_sid: this.call_sid,
hook: ['verb:hook', 'session:redirect', 'llm:event', 'llm:tool-call'].includes(type) ? url : undefined,
hook: [
'verb:hook', 'dial:confirm', 'session:redirect', 'llm:event', 'llm:tool-call'
].includes(type) ? url : undefined,
data: {...payload},
...b3
};

982
package-lock.json generated

File diff suppressed because it is too large Load Diff

View File

@@ -27,14 +27,15 @@
"dependencies": {
"@aws-sdk/client-auto-scaling": "^3.549.0",
"@aws-sdk/client-sns": "^3.549.0",
"@jambonz/db-helpers": "^0.9.6",
"@jambonz/db-helpers": "^0.9.12",
"@jambonz/http-health-check": "^0.0.1",
"@jambonz/mw-registrar": "^0.2.7",
"@jambonz/realtimedb-helpers": "^0.8.13",
"@jambonz/speech-utils": "^0.2.3",
"@jambonz/speech-utils": "^0.2.10",
"@jambonz/stats-collector": "^0.1.10",
"@jambonz/verb-specifications": "^0.0.98",
"@jambonz/time-series": "^0.2.13",
"@jambonz/verb-specifications": "^0.0.104",
"@modelcontextprotocol/sdk": "^1.9.0",
"@opentelemetry/api": "^1.8.0",
"@opentelemetry/exporter-jaeger": "^1.23.0",
"@opentelemetry/exporter-trace-otlp-http": "^0.50.0",
@@ -47,8 +48,8 @@
"bent": "^7.3.12",
"debug": "^4.3.4",
"deepcopy": "^2.1.0",
"drachtio-fsmrf": "^4.0.1",
"drachtio-srf": "^5.0.2",
"drachtio-fsmrf": "^4.0.3",
"drachtio-srf": "^5.0.5",
"express": "^4.19.2",
"express-validator": "^7.0.1",
"moment": "^2.30.1",
@@ -60,7 +61,7 @@
"short-uuid": "^5.1.0",
"sinon": "^17.0.1",
"to-snake-case": "^1.0.0",
"undici": "^6.20.0",
"undici": "^7.5.0",
"uuid-random": "^1.3.2",
"verify-aws-sns-signature": "^0.1.0",
"ws": "^8.18.0",