Compare commits

..

12 Commits

Author SHA1 Message Date
Dave Horton
c71a58dcb5 wip 2023-12-01 11:06:34 -05:00
Dave Horton
8022f9d16c aws pii support 2023-12-01 10:32:19 -05:00
Dave Horton
c4dcb051be wip 2023-12-01 09:52:34 -05:00
Dave Horton
29dc2f7052 cleanup alerts 2023-12-01 09:52:34 -05:00
Dave Horton
7718f50877 report aws stt errors 2023-12-01 09:52:34 -05:00
Dave Horton
11283edf6f fix bug in prev commit 2023-12-01 09:52:34 -05:00
Dave Horton
6043921067 aws stt: calculate transcript-level confidence based on word confidence scores 2023-12-01 09:52:34 -05:00
Dave Horton
8ad947c0fd wip 2023-12-01 09:52:34 -05:00
Dave Horton
63c925c731 wip 2023-12-01 09:52:34 -05:00
Dave Horton
e8647b2b55 when using AWS for STT generate and use a session token 2023-12-01 09:52:34 -05:00
Dave Horton
dab83423cf update speech utils 2023-12-01 09:52:34 -05:00
Dave Horton
864a673ea0 get an aws security token for STT 2023-12-01 09:52:34 -05:00
27 changed files with 872 additions and 1440 deletions

2
app.js
View File

@@ -109,7 +109,7 @@ const disconnect = () => {
httpServer?.on('close', resolve);
httpServer?.close();
srf.disconnect();
srf.locals.mediaservers?.forEach((ms) => ms.disconnect());
srf.locals.mediaservers.forEach((ms) => ms.disconnect());
});
};

View File

@@ -45,7 +45,7 @@ The GCP credential is the JSON service key in stringified format.
#### Install Docker
The test suite also requires [Docker](https://www.docker.com/) and docker-compose to be installed on your laptop. Docker is used to set up a network with all of the elements required to test the jambonz-feature-server in a black-box type of fashion.
The test suite ralso equires [Docker](https://www.docker.com/) and docker-compose to be installed on your laptop. Docker is used to set up a network with all of the elements required to test the jambonz-feature-server in a black-box type of fashion.
Once you have docker installed, you can optionally make sure everything Docker-wise is working properly by running this command from the project folder:

View File

@@ -120,7 +120,6 @@ const HTTP_TIMEOUT = 10000;
const HTTP_PROXY_IP = process.env.JAMBONES_HTTP_PROXY_IP;
const HTTP_PROXY_PORT = process.env.JAMBONES_HTTP_PROXY_PORT;
const HTTP_PROXY_PROTOCOL = process.env.JAMBONES_HTTP_PROXY_PROTOCOL || 'http';
const HTTP_USER_AGENT_HEADER = process.env.JAMBONES_HTTP_USER_AGENT_HEADER || 'jambonz';
const OPTIONS_PING_INTERVAL = parseInt(process.env.OPTIONS_PING_INTERVAL, 10) || 30000;
@@ -130,8 +129,6 @@ const JAMBONZ_RECORD_WS_PASSWORD = process.env.JAMBONZ_RECORD_WS_PASSWORD || pro
const JAMBONZ_DISABLE_DIAL_PAI_HEADER = process.env.JAMBONZ_DISABLE_DIAL_PAI_HEADER || false;
const JAMBONES_DISABLE_DIRECT_P2P_CALL = process.env.JAMBONES_DISABLE_DIRECT_P2P_CALL || false;
const JAMBONES_EAGERLY_PRE_CACHE_AUDIO = process.env.JAMBONES_EAGERLY_PRE_CACHE_AUDIO;
module.exports = {
JAMBONES_MYSQL_HOST,
JAMBONES_MYSQL_USER,
@@ -154,7 +151,6 @@ module.exports = {
JAMBONES_API_BASE_URL,
JAMBONES_TIME_SERIES_HOST,
JAMBONES_INJECT_CONTENT,
JAMBONES_EAGERLY_PRE_CACHE_AUDIO,
JAMBONES_ESL_LISTEN_ADDRESS,
JAMBONES_SBCS,
JAMBONES_OTEL_ENABLED,
@@ -197,7 +193,6 @@ module.exports = {
HTTP_PROXY_IP,
HTTP_PROXY_PORT,
HTTP_PROXY_PROTOCOL,
HTTP_USER_AGENT_HEADER,
OPTIONS_PING_INTERVAL,
RESPONSE_TIMEOUT_MS,
JAMBONES_WS_HANDSHAKE_TIMEOUT_MS,

View File

@@ -116,9 +116,12 @@ const customSanitizeFunction = (value) => {
/* trims characters at the beginning and at the end of a string */
value = value.trim();
/* Verify strings including 'http' via new URL */
/* We don't escape URLs but verify them via new URL */
if (value.includes('http')) {
value = new URL(value).toString();
} else {
/* replaces <, >, &, ', " and / with their corresponding HTML entities */
value = escape(value);
}
}
} catch (error) {

View File

@@ -52,16 +52,10 @@ module.exports = function(srf, logger) {
let clientDb = null;
if (req.has('X-Authenticated-User')) {
req.locals.originatingUser = req.get('X-Authenticated-User');
let clientSettings;
const arr = /^(.*)@(.*)/.exec(req.locals.originatingUser);
if (arr) {
[clientSettings] = await lookupClientByAccountAndUsername(account_sid, arr[1]);
[clientDb] = await lookupClientByAccountAndUsername(account_sid, arr[1]);
}
clientDb = await registrar.query(req.locals.originatingUser);
clientDb = {
...clientDb,
...clientSettings,
};
}
// check for call to application
@@ -166,7 +160,7 @@ module.exports = function(srf, logger) {
};
logger.info({callId, metadata, sdp}, 'successfully parsed SIPREC payload');
} catch (err) {
logger.info({err, callId}, 'Error parsing multipart payload');
logger.info({callId}, 'Error parsing multipart payload');
return res.send(503);
}
}
@@ -382,7 +376,7 @@ module.exports = function(srf, logger) {
},
recognizer: {
vendor: app.speech_recognizer_vendor,
...(app.speech_recognizer_label && {label: app.speech_recognizer_label}),
...(app.speech_synthesis_label && {label: app.speech_synthesis_label}),
language: app.speech_recognizer_language,
...(app.fallback_speech_recognizer_vendor && {fallback_vendor: app.fallback_speech_recognizer_vendor}),
...(app.fallback_speech_recognizer_label && {fallback_label: app.fallback_speech_recognizer_label}),

View File

@@ -1,6 +1,4 @@
const CallSession = require('./call-session');
const {CallStatus} = require('../utils/constants');
const moment = require('moment');
/**
* @classdesc Subclass of CallSession. Represents a CallSession
@@ -28,7 +26,6 @@ class AdultingCallSession extends CallSession {
this._callReleased();
});
this.sd.emit('adulting');
this.on('callStatusChange', this._notifyCallStatusChange.bind(this));
}
get dlg() {
@@ -53,18 +50,6 @@ class AdultingCallSession extends CallSession {
}
_callerHungup() {
if (this.dlg.connectTime) {
const duration = moment().diff(this.dlg.connectTime, 'seconds');
this.rootSpan.setAttributes({'call.termination': 'hangup by caller'});
this.callInfo.callTerminationBy = 'caller';
this.emit('callStatusChange', {
callStatus: CallStatus.Completed,
duration
});
}
this.logger.info('InboundCallSession: caller hung up');
this._callReleased();
this.req.removeAllListeners('cancel');
}
}

View File

@@ -19,10 +19,11 @@ const HttpRequestor = require('../utils/http-requestor');
const WsRequestor = require('../utils/ws-requestor');
const {
JAMBONES_INJECT_CONTENT,
JAMBONES_EAGERLY_PRE_CACHE_AUDIO,
AWS_REGION,
JAMBONZ_RECORD_WS_BASE_URL,
JAMBONZ_RECORD_WS_USERNAME,
JAMBONZ_RECORD_WS_PASSWORD,
} = require('../config');
const BackgroundTaskManager = require('../utils/background-task-manager');
const BADPRECONDITIONS = 'preconditions not met';
const CALLER_CANCELLED_ERR_MSG = 'Response not sent due to unknown transaction';
@@ -65,11 +66,6 @@ class CallSession extends Emitter {
this.callGone = false;
this.notifiedComplete = false;
this.rootSpan = rootSpan;
this.backgroundTaskManager = new BackgroundTaskManager({
cs: this,
logger,
rootSpan
});
this._origRecognizerSettings = {
vendor: this.application?.speech_recognizer_vendor,
@@ -140,7 +136,8 @@ class CallSession extends Emitter {
}
get isBackGroundListen() {
return this.backgroundTaskManager.isTaskRunning('listen');
return !(this.backgroundListenTask === null ||
this.backgroundListenTask === undefined);
}
/**
@@ -173,18 +170,6 @@ class CallSession extends Emitter {
return this.application.notifier;
}
/**
* syntheizer
*/
get synthesizer() {
return this._synthesizer;
}
set synthesizer(synth) {
this._synthesizer = synth;
}
/**
* default vendor to use for speech synthesis if not provided in the app
*/
@@ -267,16 +252,6 @@ class CallSession extends Emitter {
set fallbackSpeechRecognizerVendor(vendor) {
this.application.fallback_speech_recognizer_vendor = vendor;
}
/**
* recognizer
*/
get recognizer() {
return this._recognizer;
}
set recognizer(rec) {
this._recognizer = rec;
}
/**
* default vendor to use for speech recognition if not provided in the app
*/
@@ -382,11 +357,11 @@ class CallSession extends Emitter {
}
get isBotModeEnabled() {
return this.backgroundTaskManager.isTaskRunning('bargeIn');
return this.backgroundGatherTask;
}
get isListenEnabled() {
return this.backgroundTaskManager.isTaskRunning('listen');
return this.backgroundListenTask;
}
get b3() {
@@ -615,41 +590,117 @@ class CallSession extends Emitter {
}
}
async startBackgroundListen(opts, bugname) {
if (this.isListenEnabled) {
this.logger.info('CallSession:startBackgroundListen - listen is already enabled, ignoring request');
return;
}
try {
this.logger.debug({opts}, 'CallSession:startBackgroundListen');
const t = normalizeJambones(this.logger, [opts]);
this.backgroundListenTask = makeTask(this.logger, t[0]);
this.backgroundListenTask.bugname = bugname;
// Remove unneeded customer data to be sent to api server.
this.backgroundListenTask.ignoreCustomerData = true;
const resources = await this._evaluatePreconditions(this.backgroundListenTask);
const {span, ctx} = this.rootSpan.startChildSpan(`background-listen:${this.backgroundListenTask.summary}`);
this.backgroundListenTask.span = span;
this.backgroundListenTask.ctx = ctx;
this.backgroundListenTask.exec(this, resources)
.then(() => {
this.logger.info('CallSession:startBackgroundListen: listen completed');
this.backgroundListenTask && this.backgroundListenTask.removeAllListeners();
this.backgroundListenTask && this.backgroundListenTask.span.end();
this.backgroundListenTask = null;
return;
})
.catch((err) => {
this.logger.info({err}, 'CallSession:startBackgroundListen: listen threw error');
this.backgroundListenTask && this.backgroundListenTask.removeAllListeners();
this.backgroundListenTask && this.backgroundListenTask.span.end();
this.backgroundListenTask = null;
});
} catch (err) {
this.logger.info({err, opts}, 'CallSession:startBackgroundListen - Error creating listen task');
}
}
async stopBackgroundListen() {
this.logger.debug('CallSession:stopBackgroundListen');
try {
if (this.backgroundListenTask) {
this.backgroundListenTask.removeAllListeners();
this.backgroundListenTask.kill().catch(() => {});
}
} catch (err) {
this.logger.info({err}, 'CallSession:stopBackgroundListen - Error stopping listen task');
}
}
async enableBotMode(gather, autoEnable) {
try {
let task;
const t = normalizeJambones(this.logger, [gather]);
const task = makeTask(this.logger, t[0]);
if (this.isBotModeEnabled) {
task = this.backgroundTaskManager.getTask('bargeIn');
const currInput = task.input;
const t = normalizeJambones(this.logger, [gather]);
const tmpTask = makeTask(this.logger, t[0]);
const newInput = tmpTask.input;
const currInput = this.backgroundGatherTask.input;
const newInput = task.input;
if (JSON.stringify(currInput) === JSON.stringify(newInput)) {
this.logger.info('CallSession:enableBotMode - bot mode currently enabled, ignoring request to start again');
return;
} else {
}
else {
this.logger.info({currInput, newInput},
'CallSession:enableBotMode - restarting background bargeIn to apply new input type');
task.sticky = false;
'CallSession:enableBotMode - restarting background gather to apply new input type');
this.backgroundGatherTask.sticky = false;
await this.disableBotMode();
}
}
task = await this.backgroundTaskManager.newTask('bargeIn', gather);
task.sticky = autoEnable;
task.once('bargeIn-done', () => {
if (this.requestor instanceof WsRequestor) {
try {
this.kill(true);
} catch (err) {}
}
});
this.logger.info({gather}, 'CallSession:enableBotMode - starting background bargeIn');
this.backgroundGatherTask = task;
this._bargeInEnabled = true;
this.backgroundGatherTask
.once('dtmf', this._clearTasks.bind(this, this.backgroundGatherTask))
.once('vad', this._clearTasks.bind(this, this.backgroundGatherTask))
.once('transcription', this._clearTasks.bind(this, this.backgroundGatherTask))
.once('timeout', this._clearTasks.bind(this, this.backgroundGatherTask));
this.logger.info({gather}, 'CallSession:enableBotMode - starting background gather');
const resources = await this._evaluatePreconditions(this.backgroundGatherTask);
const {span, ctx} = this.rootSpan.startChildSpan(`background-gather:${this.backgroundGatherTask.summary}`);
this.backgroundGatherTask.span = span;
this.backgroundGatherTask.ctx = ctx;
this.backgroundGatherTask.sticky = autoEnable;
this.backgroundGatherTask.exec(this, resources)
.then(() => {
this.logger.info('CallSession:enableBotMode: gather completed');
this.backgroundGatherTask && this.backgroundGatherTask.removeAllListeners();
this.backgroundGatherTask && this.backgroundGatherTask.span.end();
const sticky = this.backgroundGatherTask?.sticky;
this.backgroundGatherTask = null;
if (sticky && !this.callGone && !this._stopping && this._bargeInEnabled) {
this.logger.info('CallSession:enableBotMode: restarting background gather');
setImmediate(() => this.enableBotMode(gather, true));
}
return;
})
.catch((err) => {
this.logger.info({err}, 'CallSession:enableBotMode: gather threw error');
this.backgroundGatherTask && this.backgroundGatherTask.removeAllListeners();
this.backgroundGatherTask && this.backgroundGatherTask.span.end();
this.backgroundGatherTask = null;
});
} catch (err) {
this.logger.info({err, gather}, 'CallSession:enableBotMode - Error creating bargeIn task');
this.logger.info({err, gather}, 'CallSession:enableBotMode - Error creating gather task');
}
}
async disableBotMode() {
this.backgroundTaskManager.stop('bargeIn');
this._bargeInEnabled = false;
if (this.backgroundGatherTask) {
try {
this.backgroundGatherTask.removeAllListeners();
await this.backgroundGatherTask.kill();
} catch (err) {}
this.backgroundGatherTask = null;
}
}
setConferenceDetails(memberId, confName, confUuid) {
@@ -833,7 +884,7 @@ class CallSession extends Emitter {
let skip = false;
this.currentTask = task;
if (TaskName.Gather === task.name && this.isBotModeEnabled) {
if (this.backgroundTaskManager.getTask('bargeIn').updateTaskInProgress(task) !== false) {
if (this.backgroundGatherTask.updateTaskInProgress(task) !== false) {
this.logger.info(`CallSession:exec skipping #${stackNum}:${taskNum}: ${task.name}`);
skip = true;
}
@@ -868,8 +919,7 @@ class CallSession extends Emitter {
if (0 === this.tasks.length &&
this.requestor instanceof WsRequestor &&
!this.requestor.closedGracefully &&
!this.callGone &&
!this.isConfirmCallSession
!this.callGone
) {
try {
await this._awaitCommandsOrHangup();
@@ -1321,6 +1371,7 @@ Duration=${duration} `
get the full transcription.
*/
delete t.bargeIn.enable;
this._bargeInEnabled = false;
this.logger.info('CallSession:kill - found bargein disabled in the stack, clearing to that point');
break;
}
@@ -1331,35 +1382,6 @@ Duration=${duration} `
this.taskIdx = 0;
}
_preCacheAudio(newTasks) {
for (const task of newTasks) {
if (task.name === TaskName.Config && task.hasSynthesizer) {
/* if they change synthesizer settings don't try to precache */
break;
}
if (task.name === TaskName.Say) {
/* identify vendor language, voice, and label */
const vendor = task.synthesizer.vendor && task.synthesizer.vendor !== 'default' ?
task.synthesizer.vendor :
this.speechSynthesisVendor;
const language = task.synthesizer.language && task.synthesizer.language !== 'default' ?
task.synthesizer.language :
this.speechSynthesisLanguage ;
const voice = task.synthesizer.voice && task.synthesizer.voice !== 'default' ?
task.synthesizer.voice :
this.speechSynthesisVoice;
const label = task.synthesizer.label && task.synthesizer.label !== 'default' ?
task.synthesizer.label :
this.speechSynthesisLabel;
this.logger.info({vendor, language, voice, label},
'CallSession:_preCacheAudio - precaching audio for future prompt');
task._synthesizeWithSpecificVendor(this, this.ep, {vendor, language, voice, label, preCache: true})
.catch((err) => this.logger.error(err, 'CallSession:_preCacheAudio - error precaching audio'));
}
}
}
/**
* Append tasks to the current execution stack UNLESS there is a gather in the stack.
* in that case, insert the tasks before the gather AND if the tasks include
@@ -1417,12 +1439,10 @@ Duration=${duration} `
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');
}
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');
}
@@ -1657,8 +1677,6 @@ Duration=${duration} `
this.notifier && this.notifier.close();
this.rootSpan && this.rootSpan.end();
// close all background tasks
this.backgroundTaskManager.stopAll();
}
/**
@@ -1915,7 +1933,7 @@ Duration=${duration} `
wrapDialog(dlg) {
dlg.connectTime = moment();
const origDestroy = dlg.destroy.bind(dlg);
dlg.destroy = (opts) => {
dlg.destroy = () => {
if (dlg.connected) {
dlg.connected = false;
dlg.destroy = origDestroy;
@@ -1924,7 +1942,7 @@ Duration=${duration} `
this.emit('callStatusChange', {callStatus: CallStatus.Completed, duration});
this.logger.debug('CallSession: call terminated by jambonz');
this.rootSpan.setAttributes({'call.termination': 'hangup by jambonz'});
origDestroy(opts).catch((err) => this.logger.info({err}, 'CallSession - error destroying dialog'));
origDestroy().catch((err) => this.logger.info({err}, 'CallSession - error destroying dialog'));
if (this.wakeupResolver) {
this.wakeupResolver({reason: 'session ended'});
this.wakeupResolver = null;
@@ -1975,14 +1993,12 @@ Duration=${duration} `
async _notifyCallStatusChange({callStatus, sipStatus, sipReason, duration}) {
if (this.callMoved) return;
// manage record all call.
if (callStatus === CallStatus.InProgress) {
if (this.accountInfo.account.record_all_calls ||
this.application.record_all_calls) {
this.backgroundTaskManager.newTask('record');
}
} else if (callStatus == CallStatus.Completed) {
this.backgroundTaskManager.stop('record');
// nice, call is in progress, good time to enable record
await this.enableRecordAllCall();
} else if (callStatus == CallStatus.Completed && this.isBackGroundListen) {
this.stopBackgroundListen().catch((err) => this.logger.error(
{err}, 'CallSession:_notifyCallStatusChange - error stopping background listen'));
}
/* race condition: we hang up at the same time as the caller */
@@ -2019,6 +2035,29 @@ Duration=${duration} `
}
}
async enableRecordAllCall() {
if (this.accountInfo.account.record_all_calls || this.application.record_all_calls) {
if (!JAMBONZ_RECORD_WS_BASE_URL || !this.accountInfo.account.bucket_credential) {
this.logger.error('Record all calls: invalid configuration');
return;
}
const listenOpts = {
url: `${JAMBONZ_RECORD_WS_BASE_URL}/record/${this.accountInfo.account.bucket_credential.vendor}`,
disableBidirectionalAudio: true,
mixType : 'stereo',
passDtmf: true
};
if (JAMBONZ_RECORD_WS_USERNAME && JAMBONZ_RECORD_WS_PASSWORD) {
listenOpts.wsAuth = {
username: JAMBONZ_RECORD_WS_USERNAME,
password: JAMBONZ_RECORD_WS_PASSWORD
};
}
this.logger.debug({listenOpts}, 'Record all calls: enabling listen');
await this.startBackgroundListen({verb: 'listen', ...listenOpts}, 'jambonz-session-record');
}
}
_configMsEndpoint() {
if (this.onHoldMusic) {
this.ep.set({hold_music: `shout://${this.onHoldMusic.replace(/^https?:\/\//, '')}`});
@@ -2061,18 +2100,6 @@ Duration=${duration} `
} catch (err) {}
}
}
/**
* startBackgroundTask - Start background task
*/
async startBackgroundTask(type, opts) {
await this.backgroundTaskManager.newTask(type, opts);
}
stopBackgroundTask(type) {
this.backgroundTaskManager.stop(type);
}
}
module.exports = CallSession;

View File

@@ -9,8 +9,7 @@ class TaskConfig extends Task {
'recognizer',
'bargeIn',
'record',
'listen',
'transcribe'
'listen'
].forEach((k) => this[k] = this.data[k] || {});
if ('notifyEvents' in this.data) {
@@ -31,13 +30,6 @@ class TaskConfig extends Task {
if (this.bargeIn[k]) this.gatherOpts[k] = this.bargeIn[k];
});
}
if (this.transcribe?.enable) {
this.transcribeOpts = {
verb: 'transcribe',
...this.transcribe
};
delete this.transcribeOpts.enable;
}
if (this.data.reset) {
if (typeof this.data.reset === 'string') this.data.reset = [this.data.reset];
@@ -45,11 +37,7 @@ class TaskConfig extends Task {
else this.data.reset = [];
if (this.bargeIn.sticky) this.autoEnable = true;
this.preconditions = (this.bargeIn.enable ||
this.record?.action ||
this.listen?.url ||
this.data.amd ||
this.transcribe?.enable) ?
this.preconditions = (this.bargeIn.enable || this.record?.action || this.listen?.url || this.data.amd) ?
TaskPreconditions.Endpoint :
TaskPreconditions.None;
@@ -62,7 +50,6 @@ class TaskConfig extends Task {
get hasRecognizer() { return Object.keys(this.recognizer).length; }
get hasRecording() { return Object.keys(this.record).length; }
get hasListen() { return Object.keys(this.listen).length; }
get hasTranscribe() { return Object.keys(this.transcribe).length; }
get summary() {
const phrase = [];
@@ -85,9 +72,6 @@ class TaskConfig extends Task {
if (this.hasListen) {
phrase.push(this.listen.enable ? `listen ${this.listen.url}` : 'stop listen');
}
if (this.hasTranscribe) {
phrase.push(this.transcribe.enable ? `transcribe ${this.transcribe.transcriptionHook}` : 'stop transcribe');
}
if (this.data.amd) phrase.push('enable amd');
if (this.notifyEvents) phrase.push(`event notification ${this.notifyEvents ? 'on' : 'off'}`);
if (this.onHoldMusic) phrase.push(`onHoldMusic: ${this.onHoldMusic}`);
@@ -125,7 +109,6 @@ class TaskConfig extends Task {
});
if (this.hasSynthesizer) {
cs.synthesizer = this.synthesizer;
cs.speechSynthesisVendor = this.synthesizer.vendor !== 'default'
? this.synthesizer.vendor
: cs.speechSynthesisVendor;
@@ -155,7 +138,6 @@ class TaskConfig extends Task {
this.logger.info({synthesizer: this.synthesizer}, 'Config: updated synthesizer');
}
if (this.hasRecognizer) {
cs.recognizer = this.recognizer;
cs.speechRecognizerVendor = this.recognizer.vendor !== 'default'
? this.recognizer.vendor
: cs.speechRecognizerVendor;
@@ -228,25 +210,10 @@ class TaskConfig extends Task {
const {enable, ...opts} = this.listen;
if (enable) {
this.logger.debug({opts}, 'Config: enabling listen');
cs.startBackgroundTask('listen', {verb: 'listen', ...opts});
cs.startBackgroundListen({verb: 'listen', ...opts});
} else {
this.logger.info('Config: disabling listen');
cs.stopBackgroundTask('listen');
}
}
if (this.hasTranscribe) {
if (this.transcribe.enable) {
this.transcribeOpts.recognizer = this.hasRecognizer ?
this.recognizer :
{
vendor: cs.speechRecognizerVendor,
language: cs.speechRecognizerLanguage
};
this.logger.debug(this.transcribeOpts, 'Config: enabling transcribe');
cs.startBackgroundTask('transcribe', this.transcribeOpts);
} else {
this.logger.info('Config: disabling transcribe');
cs.stopBackgroundTask('transcribe');
cs.stopBackgroundListen();
}
}
if (this.data.sipRequestWithinDialogHook) {

View File

@@ -323,7 +323,7 @@ class TaskDial extends Task {
const by = parseUri(req.getParsedHeader('Referred-By').uri);
this.logger.info({to}, 'refer to parsed');
const json = await cs.requestor.request('verb:hook', this.referHook, {
...(callInfo.toJSON()),
...callInfo,
refer_details: {
sip_refer_to: req.get('Refer-To'),
sip_referred_by: req.get('Referred-By'),

View File

@@ -71,7 +71,6 @@ class TaskGather extends SttTask {
/* buffer speech for continuous asr */
this._bufferedTranscripts = [];
this.partialTranscriptsCount = 0;
this.bugname_prefix = 'gather_';
}
get name() { return TaskName.Gather; }
@@ -104,7 +103,7 @@ class TaskGather extends SttTask {
async exec(cs, {ep}) {
this.logger.debug({options: this.data}, 'Gather:exec');
await super.exec(cs, {ep});
await super.exec(cs);
const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, cs.srf);
if (cs.hasGlobalSttHints && !this.maskGlobalSttHints) {
@@ -117,6 +116,14 @@ class TaskGather extends SttTask {
this.logger.debug({hints: this.data.recognizer.hints, hintsBoost: this.data.recognizer.hintsBoost},
'Gather:exec - applying global sttHints');
}
if (cs.hasAltLanguages) {
this.data.recognizer.altLanguages = this.data.recognizer.altLanguages.concat(cs.altLanguages);
this.logger.debug({altLanguages: this.data.recognizer?.altLanguages},
'Gather:exec - applying altLanguages');
}
if (cs.hasGlobalSttPunctuation && !this.data.recognizer.punctuation) {
this.data.recognizer.punctuation = cs.globalSttPunctuation;
}
if (!this.isContinuousAsr && cs.isContinuousAsr) {
this.isContinuousAsr = true;
this.asrTimeout = cs.asrTimeout * 1000;
@@ -134,6 +141,59 @@ class TaskGather extends SttTask {
this.interim = true;
this.logger.debug('Gather:exec - early hints match enabled');
}
this.ep = ep;
if ('default' === this.vendor || !this.vendor) {
this.vendor = cs.speechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.vendor = this.vendor;
}
if ('default' === this.language || !this.language) {
this.language = cs.speechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.language = this.language;
}
if ('default' === this.label || !this.label) {
this.label = cs.speechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.label = this.label;
}
// Fallback options
if ('default' === this.fallbackVendor || !this.fallbackVendor) {
this.fallbackVendor = cs.fallbackSpeechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.fallbackVendor = this.fallbackVendor;
}
if ('default' === this.fallbackLanguage || !this.fallbackLanguage) {
this.fallbackLanguage = cs.fallbackSpeechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.fallbackLanguage = this.fallbackLanguage;
}
if ('default' === this.fallbackLabel || !this.fallbackLabel) {
this.fallbackLabel = cs.fallbackSpeechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.fallbackLabel = this.fallbackLabel;
}
if (!this.data.recognizer.vendor) {
this.data.recognizer.vendor = this.vendor;
}
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
// By default, application saves cobalt model in language
this.data.recognizer.model = cs.speechRecognizerLanguage;
}
if (this.needsStt && !this.sttCredentials) {
try {
this.sttCredentials = await this._initSpeechCredentials(cs, this.vendor, this.label);
} catch (error) {
if (this.fallbackVendor && this.isHandledByPrimaryProvider) {
await this._fallback();
} else {
throw error;
}
}
}
/* when using cobalt model is required */
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
this.notifyError({ msg: 'ASR error', details:'Cobalt requires a model to be specified'});
throw new Error('Cobalt requires a model to be specified');
}
const startListening = async(cs, ep) => {
this._startTimer();
if (this.isContinuousAsr && 0 === this.timeout) this._startAsrTimer();
@@ -221,7 +281,7 @@ class TaskGather extends SttTask {
} catch (err) {
this.logger.error(err, 'TaskGather:exec error');
}
this.removeCustomEventListeners();
this.removeSpeechListeners(ep);
}
kill(cs) {
@@ -259,16 +319,6 @@ class TaskGather extends SttTask {
this._resolve('dtmf-terminator-key');
}
else if (this.input.includes('digits')) {
if (this.digitBuffer.length === 0 && this.needsStt) {
// DTMF is higher priority than STT.
this.removeCustomEventListeners();
ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname,
})
.catch((err) => this.logger.error({err},
` Received DTMF, Error stopping transcription for vendor ${this.vendor}`));
}
this.digitBuffer += evt.dtmf;
const len = this.digitBuffer.length;
if (len === this.numDigits || len === this.maxDigits) {
@@ -304,38 +354,34 @@ class TaskGather extends SttTask {
const opts = this.setChannelVarsForStt(this, this.sttCredentials, this.data.recognizer);
switch (this.vendor) {
case 'google':
this.bugname = `${this.bugname_prefix}google_transcribe`;
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.EndOfUtterance, this._onEndOfUtterance.bind(this, cs, ep));
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
this.bugname = 'google_transcribe';
ep.addCustomEventListener(GoogleTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(GoogleTranscriptionEvents.EndOfUtterance, this._onEndOfUtterance.bind(this, cs, ep));
ep.addCustomEventListener(GoogleTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'aws':
case 'polly':
this.bugname = `${this.bugname_prefix}aws_transcribe`;
this.addCustomEventListener(ep, AwsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, AwsTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
this.bugname = 'aws_transcribe';
ep.addCustomEventListener(AwsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(AwsTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'microsoft':
this.bugname = `${this.bugname_prefix}azure_transcribe`;
this.addCustomEventListener(
ep, AzureTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, AzureTranscriptionEvents.NoSpeechDetected,
this.bugname = 'azure_transcribe';
ep.addCustomEventListener(AzureTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(AzureTranscriptionEvents.NoSpeechDetected,
this._onNoSpeechDetected.bind(this, cs, ep));
this.addCustomEventListener(ep, AzureTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
ep.addCustomEventListener(AzureTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'nuance':
this.bugname = `${this.bugname_prefix}nuance_transcribe`;
this.addCustomEventListener(ep, NuanceTranscriptionEvents.Transcription,
this.bugname = 'nuance_transcribe';
ep.addCustomEventListener(NuanceTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, NuanceTranscriptionEvents.StartOfSpeech,
ep.addCustomEventListener(NuanceTranscriptionEvents.StartOfSpeech,
this._onStartOfSpeech.bind(this, cs, ep));
this.addCustomEventListener(ep, NuanceTranscriptionEvents.TranscriptionComplete,
ep.addCustomEventListener(NuanceTranscriptionEvents.TranscriptionComplete,
this._onTranscriptionComplete.bind(this, cs, ep));
this.addCustomEventListener(ep, NuanceTranscriptionEvents.VadDetected,
ep.addCustomEventListener(NuanceTranscriptionEvents.VadDetected,
this._onVadDetected.bind(this, cs, ep));
/* stall timers until prompt finishes playing */
@@ -345,11 +391,10 @@ class TaskGather extends SttTask {
break;
case 'deepgram':
this.bugname = `${this.bugname_prefix}deepgram_transcribe`;
this.addCustomEventListener(
ep, DeepgramTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.ConnectFailure,
this.bugname = 'deepgram_transcribe';
ep.addCustomEventListener(DeepgramTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(DeepgramTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
/* if app sets deepgramOptions.utteranceEndMs they essentially want continuous asr */
@@ -357,15 +402,13 @@ class TaskGather extends SttTask {
break;
case 'soniox':
this.bugname = `${this.bugname_prefix}soniox_transcribe`;
this.addCustomEventListener(
ep, SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.bugname = 'soniox_transcribe';
ep.addCustomEventListener(SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
break;
case 'cobalt':
this.bugname = `${this.bugname_prefix}cobalt_transcribe`;
this.addCustomEventListener(
ep, CobaltTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.bugname = 'cobalt_transcribe';
ep.addCustomEventListener(CobaltTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
/* cobalt doesnt have language, it has model, which is required */
if (!this.data.recognizer.model) {
@@ -394,22 +437,22 @@ class TaskGather extends SttTask {
break;
case 'ibm':
this.bugname = `${this.bugname_prefix}ibm_transcribe`;
this.addCustomEventListener(ep, IbmTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, IbmTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, IbmTranscriptionEvents.ConnectFailure,
this.bugname = 'ibm_transcribe';
ep.addCustomEventListener(IbmTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(IbmTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(IbmTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
case 'nvidia':
this.bugname = `${this.bugname_prefix}nvidia_transcribe`;
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.Transcription,
this.bugname = 'nvidia_transcribe';
ep.addCustomEventListener(NvidiaTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.StartOfSpeech,
ep.addCustomEventListener(NvidiaTranscriptionEvents.StartOfSpeech,
this._onStartOfSpeech.bind(this, cs, ep));
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.TranscriptionComplete,
ep.addCustomEventListener(NvidiaTranscriptionEvents.TranscriptionComplete,
this._onTranscriptionComplete.bind(this, cs, ep));
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.VadDetected,
ep.addCustomEventListener(NvidiaTranscriptionEvents.VadDetected,
this._onVadDetected.bind(this, cs, ep));
/* I think nvidia has this (??) - stall timers until prompt finishes playing */
@@ -419,22 +462,20 @@ class TaskGather extends SttTask {
break;
case 'assemblyai':
this.bugname = `${this.bugname_prefix}assemblyai_transcribe`;
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.Transcription,
this.bugname = 'assemblyai_transcribe';
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, AssemblyAiTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.ConnectFailure,
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
this.addCustomEventListener(
ep, JambonzTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, JambonzTranscriptionEvents.ConnectFailure,
this.bugname = `${this.vendor}_transcribe`;
ep.addCustomEventListener(JambonzTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(JambonzTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(JambonzTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
}
@@ -446,7 +487,7 @@ class TaskGather extends SttTask {
}
/* common handler for all stt engine errors */
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
ep.addCustomEventListener(JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
await ep.set(opts)
.catch((err) => this.logger.info(err, 'Error setting channel variables'));
}
@@ -487,9 +528,7 @@ class TaskGather extends SttTask {
this._clearTimer();
this._timeoutTimer = setTimeout(() => {
if (this.isContinuousAsr) this._startAsrTimer();
else if (this.interDigitTimeout <= 0 ||
this.digitBuffer.length < this.minDigits ||
this.needsStt && this.digitBuffer.length === 0) {
else if (this.interDigitTimeout <= 0 || this.digitBuffer.length < this.minDigits || this.needsStt) {
this._resolve(this.digitBuffer.length >= this.minDigits ? 'dtmf-num-digits' : 'timeout');
}
}, this.timeout);
@@ -595,8 +634,7 @@ class TaskGather extends SttTask {
return;
}
evt = this.normalizeTranscription(evt, this.vendor, 1, this.language,
this.shortUtterance, this.data.recognizer.punctuation);
evt = this.normalizeTranscription(evt, this.vendor, 1, this.language, this.shortUtterance);
if (evt.alternatives.length === 0) {
this.logger.info({evt}, 'TaskGather:_onTranscription - got empty transcript, continue listening');
return;
@@ -765,10 +803,7 @@ class TaskGather extends SttTask {
async _onJambonzError(cs, ep, evt) {
this.logger.info({evt}, 'TaskGather:_onJambonzError');
if (this.isHandledByPrimaryProvider && this.fallbackVendor) {
ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname
})
ep.stopTranscription({vendor: this.vendor})
.catch((err) => this.logger.error({err}, `Error stopping transcription for primary vendor ${this.vendor}`));
const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, cs.srf);
try {
@@ -789,13 +824,14 @@ class TaskGather extends SttTask {
if (code === 413 && error === 'Too much speech') return this._resolve('timeout');
}
this.logger.info({evt}, 'TaskGather:_onJambonzError');
const errMessage = evt.error || evt.Message;
writeAlerts({
account_sid: cs.accountSid,
alert_type: AlertType.STT_FAILURE,
message: `Custom speech vendor ${this.vendor} error: ${evt.error}`,
message: `Speech vendor ${this.vendor} error: ${errMessage}`,
vendor: this.vendor,
}).catch((err) => this.logger.info({err}, 'Error generating alert for jambonz custom connection failure'));
this.notifyError({msg: 'ASR error', details:`Custom speech vendor ${this.vendor} error: ${evt.error}`});
this.notifyError({msg: 'ASR error', details:`Speech vendor ${this.vendor} error: ${evt.error}`});
}
_onVendorConnectFailure(cs, _ep, evt) {
@@ -849,10 +885,7 @@ class TaskGather extends SttTask {
'stt.result': JSON.stringify(evt)
});
if (this.needsStt && this.ep && this.ep.connected) {
this.ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname
})
this.ep.stopTranscription({vendor: this.vendor})
.catch((err) => this.logger.error({err}, 'Error stopping transcription'));
}

View File

@@ -59,7 +59,7 @@ class TaskSay extends Task {
}
}
async _synthesizeWithSpecificVendor(cs, ep, {vendor, language, voice, label, preCache = false}) {
async _synthesizeWithSpecificVendor(cs, ep, {vendor, language, voice, label}) {
const {srf} = cs;
const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, srf);
const {writeAlerts, AlertType, stats} = srf.locals;
@@ -76,8 +76,6 @@ class TaskSay extends Task {
voice = arr[1];
model = arr[2];
}
} else if (vendor === 'deepgram') {
model = voice;
}
/* allow for microsoft custom region voice and api_key to be specified as an override */
@@ -97,7 +95,7 @@ class TaskSay extends Task {
voice = this.options.voice_id || voice;
}
if (!preCache) this.logger.info({vendor, language, voice, model}, 'TaskSay:exec');
this.logger.info({vendor, language, voice, model}, 'TaskSay:exec');
try {
if (!credentials) {
writeAlerts({
@@ -120,15 +118,11 @@ class TaskSay extends Task {
if (text.startsWith('silence_stream://')) return text;
/* otel: trace time for tts */
let otelSpan;
if (!preCache) {
const {span} = this.startChildSpan('tts-generation', {
'tts.vendor': vendor,
'tts.language': language,
'tts.voice': voice
});
otelSpan = span;
}
const {span} = this.startChildSpan('tts-generation', {
'tts.vendor': vendor,
'tts.language': language,
'tts.voice': voice
});
try {
const {filePath, servedFromCache, rtt} = await synthAudio(stats, {
account_sid: cs.accountSid,
@@ -150,9 +144,9 @@ class TaskSay extends Task {
updateSpeechCredentialLastUsed(credentials.speech_credential_sid)
.catch(() => {/*already logged error */});
}
if (otelSpan) otelSpan.setAttributes({'tts.cached': servedFromCache});
if (otelSpan) otelSpan.end();
if (!servedFromCache && rtt && !preCache) {
span.setAttributes({'tts.cached': servedFromCache});
span.end();
if (!servedFromCache && rtt) {
this.notifyStatus({
event: 'synthesized-audio',
vendor,
@@ -164,7 +158,7 @@ class TaskSay extends Task {
return filePath;
} catch (err) {
this.logger.info({err}, 'Error synthesizing tts');
if (otelSpan) otelSpan.end();
span.end();
writeAlerts({
account_sid: cs.accountSid,
alert_type: AlertType.TTS_FAILURE,

View File

@@ -14,15 +14,17 @@ class SttTask extends Task {
const {
setChannelVarsForStt,
normalizeTranscription,
removeSpeechListeners,
setSpeechCredentialsAtRuntime,
compileSonioxTranscripts,
consolidateTranscripts
} = require('../utils/transcription-utils')(logger);
this.setChannelVarsForStt = setChannelVarsForStt;
this.normalizeTranscription = normalizeTranscription;
this.removeSpeechListeners = removeSpeechListeners;
this.compileSonioxTranscripts = compileSonioxTranscripts;
this.consolidateTranscripts = consolidateTranscripts;
this.eventHandlers = [];
this.isHandledByPrimaryProvider = true;
if (this.data.recognizer) {
const recognizer = this.data.recognizer;
@@ -47,108 +49,11 @@ class SttTask extends Task {
/* buffer for soniox transcripts */
this._sonioxTranscripts = [];
/*bug name prefix */
this.bugname_prefix = '';
}
async exec(cs, {ep, ep2}) {
super.exec(cs);
this.ep = ep;
this.ep2 = ep2;
// copy all value from config verb to this object.
if (cs.recognizer) {
for (const k in cs.recognizer) {
if (Array.isArray(this.data.recognizer[k]) ||
Array.isArray(cs.recognizer[k])) {
this.data.recognizer[k] = [
...this.data.recognizer[k],
...cs.recognizer[k]
];
} else if (typeof this.data.recognizer[k] === 'object' ||
typeof cs.recognizer[k] === 'object'
) {
this.data.recognizer[k] = {
...this.data.recognizer[k],
...cs.recognizer[k]
};
} else {
this.data.recognizer[k] = cs.recognizer[k] || this.data.recognizer[k];
}
}
}
if ('default' === this.vendor || !this.vendor) {
this.vendor = cs.speechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.vendor = this.vendor;
}
if ('default' === this.language || !this.language) {
this.language = cs.speechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.language = this.language;
}
if ('default' === this.label || !this.label) {
this.label = cs.speechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.label = this.label;
}
// Fallback options
if ('default' === this.fallbackVendor || !this.fallbackVendor) {
this.fallbackVendor = cs.fallbackSpeechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.fallbackVendor = this.fallbackVendor;
}
if ('default' === this.fallbackLanguage || !this.fallbackLanguage) {
this.fallbackLanguage = cs.fallbackSpeechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.fallbackLanguage = this.fallbackLanguage;
}
if ('default' === this.fallbackLabel || !this.fallbackLabel) {
this.fallbackLabel = cs.fallbackSpeechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.fallbackLabel = this.fallbackLabel;
}
if (!this.data.recognizer.vendor) {
this.data.recognizer.vendor = this.vendor;
}
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
// By default, application saves cobalt model in language
this.data.recognizer.model = cs.speechRecognizerLanguage;
}
if (!this.sttCredentials) {
try {
this.sttCredentials = await this._initSpeechCredentials(this.cs, this.vendor, this.label);
} catch (error) {
if (this.fallbackVendor && this.isHandledByPrimaryProvider) {
await this._fallback();
} else {
throw error;
}
}
}
/* when using cobalt model is required */
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
this.notifyError({ msg: 'ASR error', details:'Cobalt requires a model to be specified'});
throw new Error('Cobalt requires a model to be specified');
}
if (cs.hasAltLanguages) {
this.data.recognizer.altLanguages = this.data.recognizer.altLanguages.concat(cs.altLanguages);
this.logger.debug({altLanguages: this.altLanguages},
'STT:exec - applying altLanguages');
}
if (cs.hasGlobalSttPunctuation && !this.data.recognizer.punctuation) {
this.data.recognizer.punctuation = cs.globalSttPunctuation;
}
}
addCustomEventListener(ep, event, handler) {
this.eventHandlers.push({ep, event, handler});
ep.addCustomEventListener(event, handler);
}
removeCustomEventListeners() {
this.eventHandlers.forEach((h) => h.ep.removeCustomEventListener(h.event, h.handler));
}
async _initSpeechCredentials(cs, vendor, label) {
const {getNuanceAccessToken, getIbmAccessToken} = cs.srf.locals.dbHelpers;
const {getNuanceAccessToken, getIbmAccessToken, getAwsAuthToken} = this.cs.srf.locals.dbHelpers;
let credentials = cs.getSpeechCredentials(vendor, 'stt', label);
if (!credentials) {
@@ -182,6 +87,15 @@ class SttTask extends Task {
this.logger.debug({stt_api_key}, `got ibm access token ${servedFromCache ? 'from cache' : ''}`);
credentials = {...credentials, access_token, stt_region};
}
else if (vendor == 'aws') {
/* get AWS access token */
const {accessKeyId, secretAccessKey, securityToken, region } = credentials;
if (!securityToken) {
const { servedFromCache, ...newCredentials} = await getAwsAuthToken(accessKeyId, secretAccessKey, region);
this.logger.debug({newCredentials}, `got aws security token ${servedFromCache ? 'from cache' : ''}`);
credentials = {...newCredentials, region};
}
}
return credentials;
}

View File

@@ -33,29 +33,19 @@ class TaskTranscribe extends SttTask {
this.childSpan = [null, null];
// Continuous asr timeout
// Continuos asr timeout
this.asrTimeout = typeof this.data.recognizer.asrTimeout === 'number' ? this.data.recognizer.asrTimeout * 1000 : 0;
if (this.asrTimeout > 0) {
this.isContinuousAsr = true;
}
/* buffer speech for continuous asr */
this._bufferedTranscripts = [];
this.bugname_prefix = 'transcribe_';
}
get name() { return TaskName.Transcribe; }
async exec(cs, {ep, ep2}) {
await super.exec(cs, {ep, ep2});
if (this.data.recognizer.vendor === 'nuance') {
this.data.recognizer.nuanceOptions = {
// by default, nuance STT will recognize only 1st utterance.
// enable multiple allow nuance detact all utterances
utteranceDetectionMode: 'multiple',
...this.data.recognizer.nuanceOptions
};
}
super.exec(cs);
const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, cs.srf);
if (cs.hasGlobalSttHints) {
@@ -65,6 +55,67 @@ class TaskTranscribe extends SttTask {
this.logger.debug({hints: this.data.recognizer.hints, hintsBoost: this.data.recognizer.hintsBoost},
'Transcribe:exec - applying global sttHints');
}
if (cs.hasAltLanguages) {
this.data.recognizer.altLanguages = this.data.recognizer.altLanguages.concat(cs.altLanguages);
this.logger.debug({altLanguages: this.altLanguages},
'Transcribe:exec - applying altLanguages');
}
if (cs.hasGlobalSttPunctuation && !this.data.recognizer.punctuation) {
this.data.recognizer.punctuation = cs.globalSttPunctuation;
}
this.ep = ep;
this.ep2 = ep2;
if ('default' === this.vendor || !this.vendor) {
this.vendor = cs.speechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.vendor = this.vendor;
}
if ('default' === this.language || !this.language) {
this.language = cs.speechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.language = this.language;
}
if ('default' === this.label || !this.label) {
this.label = cs.speechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.label = this.label;
}
// Fallback options
if ('default' === this.fallbackVendor || !this.fallbackVendor) {
this.fallbackVendor = cs.fallbackSpeechRecognizerVendor;
if (this.data.recognizer) this.data.recognizer.fallbackVendor = this.fallbackVendor;
}
if ('default' === this.fallbackLanguage || !this.fallbackLanguage) {
this.fallbackLanguage = cs.fallbackSpeechRecognizerLanguage;
if (this.data.recognizer) this.data.recognizer.fallbackLanguage = this.fallbackLanguage;
}
if ('default' === this.fallbackLabel || !this.fallbackLabel) {
this.fallbackLabel = cs.fallbackSpeechRecognizerLabel;
if (this.data.recognizer) this.data.recognizer.fallbackLabel = this.fallbackLabel;
}
if (!this.data.recognizer.vendor) {
this.data.recognizer.vendor = this.vendor;
}
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
// By default, application saves cobalt model in language
this.data.recognizer.model = cs.speechRecognizerLanguage;
}
if (!this.sttCredentials) {
try {
this.sttCredentials = await this._initSpeechCredentials(cs, this.vendor, this.label);
} catch (error) {
if (this.fallbackVendor && this.isHandledByPrimaryProvider) {
await this._fallback();
} else {
throw error;
}
}
}
/* when using cobalt model is required */
if (this.vendor === 'cobalt' && !this.data.recognizer.model) {
this.notifyError({ msg: 'ASR error', details:'Cobalt requires a model to be specified'});
throw new Error('Cobalt requires a model to be specified');
}
try {
await this._startTranscribing(cs, ep, 1);
@@ -80,17 +131,14 @@ class TaskTranscribe extends SttTask {
this.logger.info(err, 'TaskTranscribe:exec - error');
this.parentTask && this.parentTask.emit('error', err);
}
this.removeCustomEventListeners();
this.removeSpeechListeners(ep);
}
async _stopTranscription() {
let stopTranscription = false;
if (this.ep?.connected) {
stopTranscription = true;
this.ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname
})
this.ep.stopTranscription({vendor: this.vendor})
.catch((err) => this.logger.info(err, 'Error TaskTranscribe:kill'));
}
if (this.separateRecognitionPerChannel && this.ep2 && this.ep2.connected) {
@@ -141,44 +189,44 @@ class TaskTranscribe extends SttTask {
const opts = this.setChannelVarsForStt(this, this.sttCredentials, this.data.recognizer);
switch (this.vendor) {
case 'google':
this.bugname = `${this.bugname_prefix}google_transcribe`;
this.addCustomEventListener(ep, GoogleTranscriptionEvents.Transcription,
this.bugname = 'google_transcribe';
ep.addCustomEventListener(GoogleTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, GoogleTranscriptionEvents.NoAudioDetected,
ep.addCustomEventListener(GoogleTranscriptionEvents.NoAudioDetected,
this._onNoAudio.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, GoogleTranscriptionEvents.MaxDurationExceeded,
ep.addCustomEventListener(GoogleTranscriptionEvents.MaxDurationExceeded,
this._onMaxDurationExceeded.bind(this, cs, ep, channel));
break;
case 'aws':
case 'polly':
this.bugname = `${this.bugname_prefix}aws_transcribe`;
this.addCustomEventListener(ep, AwsTranscriptionEvents.Transcription,
this.bugname = 'aws_transcribe';
ep.addCustomEventListener(AwsTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, AwsTranscriptionEvents.NoAudioDetected,
ep.addCustomEventListener(AwsTranscriptionEvents.NoAudioDetected,
this._onNoAudio.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, AwsTranscriptionEvents.MaxDurationExceeded,
ep.addCustomEventListener(AwsTranscriptionEvents.MaxDurationExceeded,
this._onMaxDurationExceeded.bind(this, cs, ep, channel));
break;
case 'microsoft':
this.bugname = `${this.bugname_prefix}azure_transcribe`;
this.addCustomEventListener(ep, AzureTranscriptionEvents.Transcription,
this.bugname = 'azure_transcribe';
ep.addCustomEventListener(AzureTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, AzureTranscriptionEvents.NoSpeechDetected,
ep.addCustomEventListener(AzureTranscriptionEvents.NoSpeechDetected,
this._onNoAudio.bind(this, cs, ep, channel));
break;
case 'nuance':
this.bugname = `${this.bugname_prefix}nuance_transcribe`;
this.addCustomEventListener(ep, NuanceTranscriptionEvents.Transcription,
this.bugname = 'nuance_transcribe';
ep.addCustomEventListener(NuanceTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
break;
case 'deepgram':
this.bugname = `${this.bugname_prefix}deepgram_transcribe`;
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.Transcription,
this.bugname = 'deepgram_transcribe';
ep.addCustomEventListener(DeepgramTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.Connect,
ep.addCustomEventListener(DeepgramTranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.ConnectFailure,
ep.addCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep, channel));
/* if app sets deepgramOptions.utteranceEndMs they essentially want continuous asr */
@@ -186,13 +234,13 @@ class TaskTranscribe extends SttTask {
break;
case 'soniox':
this.bugname = `${this.bugname_prefix}soniox_transcribe`;
this.addCustomEventListener(ep, SonioxTranscriptionEvents.Transcription,
this.bugname = 'soniox_transcribe';
ep.addCustomEventListener(SonioxTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
break;
case 'cobalt':
this.bugname = `${this.bugname_prefix}cobalt_transcribe`;
this.addCustomEventListener(ep, CobaltTranscriptionEvents.Transcription,
this.bugname = 'cobalt_transcribe';
ep.addCustomEventListener(CobaltTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
/* cobalt doesnt have language, it has model, which is required */
@@ -221,39 +269,38 @@ class TaskTranscribe extends SttTask {
break;
case 'ibm':
this.bugname = `${this.bugname_prefix}ibm_transcribe`;
this.addCustomEventListener(ep, IbmTranscriptionEvents.Transcription,
this.bugname = 'ibm_transcribe';
ep.addCustomEventListener(IbmTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, IbmTranscriptionEvents.Connect,
ep.addCustomEventListener(IbmTranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, IbmTranscriptionEvents.ConnectFailure,
ep.addCustomEventListener(IbmTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep, channel));
break;
case 'nvidia':
this.bugname = `${this.bugname_prefix}nvidia_transcribe`;
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.bugname = 'nvidia_transcribe';
ep.addCustomEventListener(NvidiaTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
break;
case 'assemblyai':
this.bugname = `${this.bugname_prefix}assemblyai_transcribe`;
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.Transcription,
this.bugname = 'assemblyai_transcribe';
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep,
AssemblyAiTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.ConnectFailure,
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep, channel));
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Transcription,
this.bugname = `${this.vendor}_transcribe`;
ep.addCustomEventListener(JambonzTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, JambonzTranscriptionEvents.ConnectFailure,
ep.addCustomEventListener(JambonzTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(JambonzTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
}
@@ -265,7 +312,7 @@ class TaskTranscribe extends SttTask {
}
/* common handler for all stt engine errors */
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
ep.addCustomEventListener(JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
await ep.set(opts)
.catch((err) => this.logger.info(err, 'Error setting channel variables'));
}
@@ -280,8 +327,6 @@ class TaskTranscribe extends SttTask {
}
async _transcribe(ep) {
this.logger.debug(
`TaskTranscribe:_transcribe - starting transcription vendor ${this.vendor} bugname ${this.bugname}`);
await ep.startTranscription({
vendor: this.vendor,
interim: this.interim ? true : false,
@@ -314,8 +359,7 @@ class TaskTranscribe extends SttTask {
}
this.logger.debug({evt}, 'TaskTranscribe:_onTranscription - before normalization');
evt = this.normalizeTranscription(evt, this.vendor, channel, this.language, undefined,
this.data.recognizer.punctuation);
evt = this.normalizeTranscription(evt, this.vendor, channel, this.language);
this.logger.debug({evt}, 'TaskTranscribe:_onTranscription');
if (evt.alternatives.length === 0) {
this.logger.info({evt}, 'TaskTranscribe:_onTranscription - got empty transcript, continue listening');
@@ -441,10 +485,7 @@ class TaskTranscribe extends SttTask {
async _onJambonzError(cs, _ep, evt) {
this.logger.info({evt}, 'TaskTranscribe:_onJambonzError');
if (this.isHandledByPrimaryProvider && this.fallbackVendor) {
_ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname
})
_ep.stopTranscription({vendor: this.vendor})
.catch((err) => this.logger.error({err}, `Error stopping transcription for primary vendor ${this.vendor}`));
const {updateSpeechCredentialLastUsed} = require('../utils/db-utils')(this.logger, cs.srf);
try {

View File

@@ -1,197 +0,0 @@
const { normalizeJambones } = require('@jambonz/verb-specifications');
const makeTask = require('../tasks/make_task');
const { JAMBONZ_RECORD_WS_BASE_URL, JAMBONZ_RECORD_WS_USERNAME, JAMBONZ_RECORD_WS_PASSWORD } = require('../config');
const Emitter = require('events');
class BackgroundTaskManager extends Emitter {
constructor({cs, logger, rootSpan}) {
super();
this.tasks = new Map();
this.cs = cs;
this.logger = logger;
this.rootSpan = rootSpan;
}
isTaskRunning(type) {
return this.tasks.has(type);
}
getTask(type) {
if (this.tasks.has(type)) {
return this.tasks.get(type);
}
}
count() {
return this.tasks.size;
}
async newTask(type, taskOpts) {
this.logger.info({taskOpts}, `initiating Background task ${type}`);
if (this.tasks.has(type)) {
this.logger.info(`Background task ${type} is running, skiped`);
return;
}
let task;
switch (type) {
case 'listen':
task = await this._initListen(taskOpts);
break;
case 'bargeIn':
task = await this._initBargeIn(taskOpts);
break;
case 'record':
task = await this._initRecord();
break;
case 'transcribe':
task = await this._initTranscribe(taskOpts);
break;
default:
break;
}
if (task) {
this.tasks.set(type, task);
}
return task;
}
stop(type) {
const task = this.getTask(type);
if (task) {
this.logger.info(`stopping background task: ${type}`);
task.removeAllListeners();
task.span.end();
task.kill();
// Remove task from managed List
this.tasks.delete(type);
} else {
this.logger.info(`stopping background task, ${type} is not running, skipped`);
}
}
stopAll() {
this.logger.info('BackgroundTaskManager:stopAll');
for (const key of this.tasks.keys()) {
this.stop(key);
}
}
// Initiate Task
// Initiate Listen
async _initListen(opts, bugname = 'jambonz-background-listen', ignoreCustomerData = false, type = 'listen') {
let task;
try {
const t = normalizeJambones(this.logger, [opts]);
task = makeTask(this.logger, t[0]);
task.bugname = bugname;
task.ignoreCustomerData = ignoreCustomerData;
const resources = await this.cs._evaluatePreconditions(task);
const {span, ctx} = this.rootSpan.startChildSpan(`background-${type}:${task.summary}`);
task.span = span;
task.ctx = ctx;
task.exec(this.cs, resources)
.then(this._taskCompleted.bind(this, type, task))
.catch(this._taskError.bind(this, type, task));
} catch (err) {
this.logger.info({err, opts}, `BackgroundTaskManager:_initListen - Error creating ${bugname} task`);
}
return task;
}
// Initiate Gather
async _initBargeIn(opts) {
let task;
try {
const t = normalizeJambones(this.logger, [opts]);
task = makeTask(this.logger, t[0]);
task
.once('dtmf', this._bargeInTaskCompleted.bind(this))
.once('vad', this._bargeInTaskCompleted.bind(this))
.once('transcription', this._bargeInTaskCompleted.bind(this))
.once('timeout', this._bargeInTaskCompleted.bind(this));
const resources = await this.cs._evaluatePreconditions(task);
const {span, ctx} = this.rootSpan.startChildSpan(`background-bargeIn:${task.summary}`);
task.span = span;
task.ctx = ctx;
task.bugname_prefix = 'background_bargeIn_';
task.exec(this.cs, resources)
.then(() => {
this._taskCompleted('bargeIn', task);
if (task.sticky && !this.cs.callGone && !this.cs._stopping) {
this.logger.info('BackgroundTaskManager:_initBargeIn: restarting background bargeIn');
this.newTask('bargeIn', opts);
}
return;
})
.catch(this._taskError.bind(this, 'bargeIn', task));
} catch (err) {
this.logger.info(err, 'BackgroundTaskManager:_initGather - Error creating bargeIn task');
}
return task;
}
// Initiate Record
async _initRecord() {
if (this.cs.accountInfo.account.record_all_calls || this.cs.application.record_all_calls) {
if (!JAMBONZ_RECORD_WS_BASE_URL || !this.cs.accountInfo.account.bucket_credential) {
this.logger.error(`_initRecord: invalid configuration,
missing JAMBONZ_RECORD_WS_BASE_URL or bucket configuration`);
return undefined;
}
const listenOpts = {
url: `${JAMBONZ_RECORD_WS_BASE_URL}/record/${this.cs.accountInfo.account.bucket_credential.vendor}`,
disableBidirectionalAudio: true,
mixType : 'stereo',
passDtmf: true
};
if (JAMBONZ_RECORD_WS_USERNAME && JAMBONZ_RECORD_WS_PASSWORD) {
listenOpts.wsAuth = {
username: JAMBONZ_RECORD_WS_USERNAME,
password: JAMBONZ_RECORD_WS_PASSWORD
};
}
this.logger.debug({listenOpts}, '_initRecord: enabling listen');
return await this._initListen({verb: 'listen', ...listenOpts}, 'jambonz-session-record', true, 'record');
}
}
// Initiate Transcribe
async _initTranscribe(opts) {
let task;
try {
const t = normalizeJambones(this.logger, [opts]);
task = makeTask(this.logger, t[0]);
const resources = await this.cs._evaluatePreconditions(task);
const {span, ctx} = this.rootSpan.startChildSpan(`background-transcribe:${task.summary}`);
task.span = span;
task.ctx = ctx;
task.bugname_prefix = 'background_transcribe_';
task.exec(this.cs, resources)
.then(this._taskCompleted.bind(this, 'transcribe', task))
.catch(this._taskError.bind(this, 'transcribe', task));
} catch (err) {
this.logger.info(err, 'BackgroundTaskManager:_initTranscribe - Error creating transcribe task');
}
return task;
}
_taskCompleted(type, task) {
this.logger.info({type, task}, 'BackgroundTaskManager:_taskCompleted: task completed');
task.removeAllListeners();
task.span.end();
this.tasks.delete(type);
}
_taskError(type, task, error) {
this.logger.info({type, task, error}, 'BackgroundTaskManager:_taskError: task Error');
task.removeAllListeners();
task.span.end();
this.tasks.delete(type);
}
_bargeInTaskCompleted(evt) {
this.logger.info({evt}, 'BackgroundTaskManager:_bargeInTaskCompleted on event from background bargeIn');
this.emit('bargeIn-done', evt);
}
}
module.exports = BackgroundTaskManager;

View File

@@ -14,7 +14,6 @@ const {
HTTP_PROXY_PORT,
HTTP_PROXY_PROTOCOL,
NODE_ENV,
HTTP_USER_AGENT_HEADER,
} = require('../config');
const toBase64 = (str) => Buffer.from(str || '', 'utf8').toString('base64');
@@ -117,10 +116,6 @@ class HttpRequestor extends BaseRequestor {
const url = hook.url || hook;
const method = hook.method || 'POST';
let buf = '';
httpHeaders = {
...httpHeaders,
...(HTTP_USER_AGENT_HEADER && {'user-agent' : HTTP_USER_AGENT_HEADER})
};
assert.ok(url, 'HttpRequestor:request url was not provided');
assert.ok, (['GET', 'POST'].includes(method), `HttpRequestor:request method must be 'GET' or 'POST' not ${method}`);

View File

@@ -178,7 +178,8 @@ function installSrfLocals(srf, logger) {
synthAudio,
getNuanceAccessToken,
getIbmAccessToken,
} = require('@jambonz/speech-utils')({}, logger);
getAwsAuthToken
} = require('@jambonz/speech-utils')({redis_client: client}, logger);
const {
writeAlerts,
AlertType
@@ -231,6 +232,7 @@ function installSrfLocals(srf, logger) {
getListPosition,
getNuanceAccessToken,
getIbmAccessToken,
getAwsAuthToken,
addToSortedSet,
retrieveFromSortedSet,
retrieveByPatternSortedSet,

View File

@@ -97,12 +97,8 @@ const parseSiprecPayload = (req, logger) => {
obj[`${prefix}participantstreamassoc`].forEach((ps) => {
const part = participants[ps.$.participant_id];
if (part) {
if (ps.hasOwnProperty(`${prefix}send`)) {
part.send = ps[`${prefix}send`][0];
}
if (ps.hasOwnProperty(`${prefix}recv`)) {
part.recv = ps[`${prefix}recv`][0];
}
part.send = ps[`${prefix}send`][0];
part.recv = ps[`${prefix}recv`][0];
}
});
}
@@ -113,9 +109,9 @@ const parseSiprecPayload = (req, logger) => {
obj[`${prefix}stream`].forEach((s) => {
const streamId = s.$.stream_id;
let sender;
for (const v of Object.values(participants)) {
for (const [k, v] of Object.entries(participants)) {
if (v.send === streamId) {
sender = v;
sender = k;
break;
}
}
@@ -125,15 +121,9 @@ const parseSiprecPayload = (req, logger) => {
sender.label = s[`${prefix}label`][0];
if (-1 !== ['1', 'a_leg', 'inbound', '10'].indexOf(sender.label)) {
opts.caller.aor = sender.aor;
if (-1 !== ['1', 'a_leg', 'inbound'].indexOf(sender.label)) {
opts.caller.aor = sender.aor ;
if (sender.name) opts.caller.name = sender.name;
// Remap the sdp stream base on sender label
if (!opts.sdp1.includes(`a=label:${sender.label}`)) {
const tmp = opts.sdp1;
opts.sdp1 = opts.sdp2;
opts.sdp2 = tmp;
}
}
else {
opts.callee.aor = sender.aor ;

View File

@@ -1,5 +1,15 @@
const {
TaskName,
AzureTranscriptionEvents,
GoogleTranscriptionEvents,
AwsTranscriptionEvents,
NuanceTranscriptionEvents,
DeepgramTranscriptionEvents,
SonioxTranscriptionEvents,
NvidiaTranscriptionEvents,
CobaltTranscriptionEvents,
JambonzTranscriptionEvents,
AssemblyAiTranscriptionEvents
} = require('./constants.json');
const stickyVars = {
@@ -50,7 +60,13 @@ const stickyVars = {
aws: [
'AWS_VOCABULARY_NAME',
'AWS_VOCABULARY_FILTER_METHOD',
'AWS_VOCABULARY_FILTER_NAME'
'AWS_VOCABULARY_FILTER_NAME',
'AWS_LANGUAGE_MODEL_NAME',
'AWS_ACCESS_KEY_ID',
'AWS_SECRET_ACCESS_KEY',
'AWS_REGION',
'AWS_SECURITY_TOKEN',
'AWS_PII_ENTITY_TYPES'
],
nuance: [
'NUANCE_ACCESS_TOKEN',
@@ -328,20 +344,19 @@ const normalizeNuance = (evt, channel, language) => {
};
};
const normalizeMicrosoft = (evt, channel, language, punctuation = true) => {
const normalizeMicrosoft = (evt, channel, language) => {
const copy = JSON.parse(JSON.stringify(evt));
const nbest = evt.NBest;
const language_code = evt.PrimaryLanguage?.Language || language;
const alternatives = nbest ? nbest.map((n) => {
return {
confidence: n.Confidence,
// remove all puntuation if needed
transcript: punctuation ? n.Display : n.Display.replace(/\p{P}/gu, '')
transcript: n.Display
};
}) :
[
{
transcript: punctuation ? evt.DisplayText || evt.Text : (evt.DisplayText || evt.Text).replace(/\p{P}/gu, '')
transcript: evt.DisplayText || evt.Text
}
];
@@ -359,11 +374,19 @@ const normalizeMicrosoft = (evt, channel, language, punctuation = true) => {
const normalizeAws = (evt, channel, language) => {
const copy = JSON.parse(JSON.stringify(evt));
const alternatives = evt.Transcript?.Results[0]?.Alternatives.map((alt) => {
const items = alt.Items.filter((item) => item.Type === 'pronunciation' && 'Confidence' in item);
const confidence = items.reduce((acc, item) => acc + item.Confidence, 0) / items.length;
return {
transcript: alt.Transcript,
confidence
};
});
return {
language_code: language,
channel_tag: channel,
is_final: evt[0].is_final,
alternatives: evt[0].alternatives,
is_final: evt.Transcript?.Results[0].IsPartial === false,
alternatives,
vendor: {
name: 'aws',
evt: copy
@@ -391,14 +414,14 @@ const normalizeAssemblyAi = (evt, channel, language) => {
};
module.exports = (logger) => {
const normalizeTranscription = (evt, vendor, channel, language, shortUtterance, punctuation) => {
const normalizeTranscription = (evt, vendor, channel, language, shortUtterance) => {
//logger.debug({ evt, vendor, channel, language }, 'normalizeTranscription');
switch (vendor) {
case 'deepgram':
return normalizeDeepgram(evt, channel, language, shortUtterance);
case 'microsoft':
return normalizeMicrosoft(evt, channel, language, punctuation);
return normalizeMicrosoft(evt, channel, language);
case 'google':
return normalizeGoogle(evt, channel, language);
case 'aws':
@@ -464,8 +487,7 @@ module.exports = (logger) => {
...(rOpts.hints?.length > 0 && typeof rOpts.hints[0] === 'object' &&
{GOOGLE_SPEECH_HINTS: JSON.stringify(rOpts.hints)}),
...(typeof rOpts.hintsBoost === 'number' && {GOOGLE_SPEECH_HINTS_BOOST: rOpts.hintsBoost}),
// When altLanguages is emptylist, we have to send value to freeswitch to clear the previous settings
...(rOpts.altLanguages &&
...(rOpts.altLanguages?.length > 0 &&
{GOOGLE_SPEECH_ALTERNATIVE_LANGUAGE_CODES: [...new Set(rOpts.altLanguages)].join(',')}),
...(rOpts.interactionType &&
{GOOGLE_SPEECH_METADATA_INTERACTION_TYPE: rOpts.interactionType}),
@@ -475,16 +497,29 @@ module.exports = (logger) => {
};
}
else if (['aws', 'polly'].includes(vendor)) {
const {awsOptions = {}} = rOpts;
const vocabularyName = awsOptions.vocabularyName || rOpts.vocabularyName;
const vocabularyFilterName = awsOptions.vocabularyFilterName || rOpts.vocabularyFilterName;
const filterMethod = awsOptions.vocabularyFilterMethod || rOpts.filterMethod;
opts = {
...opts,
...(rOpts.vocabularyName && {AWS_VOCABULARY_NAME: rOpts.vocabularyName}),
...(rOpts.vocabularyFilterName && {AWS_VOCABULARY_FILTER_NAME: rOpts.vocabularyFilterName}),
...(rOpts.filterMethod && {AWS_VOCABULARY_FILTER_METHOD: rOpts.filterMethod}),
...(vocabularyName && {AWS_VOCABULARY_NAME: vocabularyName}),
...(vocabularyFilterName && {AWS_VOCABULARY_FILTER_NAME: vocabularyFilterName}),
...(filterMethod && {AWS_VOCABULARY_FILTER_METHOD: filterMethod}),
...(sttCredentials && {
AWS_ACCESS_KEY_ID: sttCredentials.accessKeyId,
AWS_SECRET_ACCESS_KEY: sttCredentials.secretAccessKey,
AWS_REGION: sttCredentials.region
AWS_REGION: sttCredentials.region,
AWS_SECURITY_TOKEN: sttCredentials.securityToken
}),
...(awsOptions.accessKey && {AWS_ACCESS_KEY_ID: awsOptions.accessKey}),
...(awsOptions.secretKey && {AWS_SECRET_ACCESS_KEY: awsOptions.secretKey}),
...(awsOptions.region && {AWS_REGION: awsOptions.region}),
...(awsOptions.securityToken && {AWS_SECURITY_TOKEN: awsOptions.securityToken}),
...(awsOptions.languageModelName && {AWS_LANGUAGE_MODEL_NAME: awsOptions.languageModelName}),
...(awsOptions.piiEntityTypes?.length && {AWS_PII_ENTITY_TYPES: awsOptions.piiEntityTypes.join(',')}),
...(awsOptions.piiIdentifyEntities && {AWS_PII_IDENTIFY_ENTITIES: true}),
...(awsOptions.languageModelName && {AWS_LANGUAGE_MODEL_NAME: awsOptions.languageModelName}),
};
}
else if ('microsoft' === vendor) {
@@ -495,8 +530,7 @@ module.exports = (logger) => {
{AZURE_SPEECH_HINTS: rOpts.hints.map((h) => h.trim()).join(',')}),
...(rOpts.hints?.length > 0 && typeof rOpts.hints[0] === 'object' &&
{AZURE_SPEECH_HINTS: rOpts.hints.map((h) => h.phrase).join(',')}),
// When altLanguages is emptylist, we have to send value to freeswitch to clear the previous settings
...(rOpts.altLanguages &&
...(rOpts.altLanguages && rOpts.altLanguages.length > 0 &&
{AZURE_SPEECH_ALTERNATIVE_LANGUAGE_CODES: [...new Set(rOpts.altLanguages)].join(',')}),
...(rOpts.requestSnr && {AZURE_REQUEST_SNR: 1}),
...(rOpts.profanityOption && {AZURE_PROFANITY_OPTION: rOpts.profanityOption}),
@@ -605,7 +639,7 @@ module.exports = (logger) => {
...(deepgramOptions.keywords) &&
{DEEPGRAM_SPEECH_KEYWORDS: deepgramOptions.keywords.join(',')},
...('endpointing' in deepgramOptions) &&
{DEEPGRAM_SPEECH_ENDPOINTING: deepgramOptions.endpointing === false ? 'false' : deepgramOptions.endpointing},
{DEEPGRAM_SPEECH_ENDPOINTING: deepgramOptions.endpointing},
...(deepgramOptions.utteranceEndMs) &&
{DEEPGRAM_SPEECH_UTTERANCE_END_MS: deepgramOptions.utteranceEndMs},
...(deepgramOptions.vadTurnoff) &&
@@ -733,7 +767,7 @@ module.exports = (logger) => {
opts = {
...opts,
...(auth_token && {JAMBONZ_STT_API_KEY: auth_token}),
JAMBONZ_STT_API_KEY: auth_token,
JAMBONZ_STT_URL: custom_stt_url,
...(Object.keys(options).length > 0 && {JAMBONZ_STT_OPTIONS: JSON.stringify(options)}),
};
@@ -745,6 +779,48 @@ module.exports = (logger) => {
return opts;
};
const removeSpeechListeners = (ep) => {
ep.removeCustomEventListener(GoogleTranscriptionEvents.Transcription);
ep.removeCustomEventListener(GoogleTranscriptionEvents.EndOfUtterance);
ep.removeCustomEventListener(GoogleTranscriptionEvents.VadDetected);
ep.removeCustomEventListener(AwsTranscriptionEvents.Transcription);
ep.removeCustomEventListener(AwsTranscriptionEvents.VadDetected);
ep.removeCustomEventListener(AzureTranscriptionEvents.Transcription);
ep.removeCustomEventListener(AzureTranscriptionEvents.NoSpeechDetected);
ep.removeCustomEventListener(AzureTranscriptionEvents.VadDetected);
ep.removeCustomEventListener(NuanceTranscriptionEvents.Transcription);
ep.removeCustomEventListener(NuanceTranscriptionEvents.TranscriptionComplete);
ep.removeCustomEventListener(NuanceTranscriptionEvents.StartOfSpeech);
ep.removeCustomEventListener(NuanceTranscriptionEvents.VadDetected);
ep.removeCustomEventListener(DeepgramTranscriptionEvents.Transcription);
ep.removeCustomEventListener(DeepgramTranscriptionEvents.Connect);
ep.removeCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure);
ep.removeCustomEventListener(SonioxTranscriptionEvents.Transcription);
ep.removeCustomEventListener(CobaltTranscriptionEvents.Transcription);
ep.removeCustomEventListener(CobaltTranscriptionEvents.CompileContext);
ep.removeCustomEventListener(NvidiaTranscriptionEvents.Transcription);
ep.removeCustomEventListener(NvidiaTranscriptionEvents.TranscriptionComplete);
ep.removeCustomEventListener(NvidiaTranscriptionEvents.StartOfSpeech);
ep.removeCustomEventListener(NvidiaTranscriptionEvents.VadDetected);
ep.removeCustomEventListener(JambonzTranscriptionEvents.Transcription);
ep.removeCustomEventListener(JambonzTranscriptionEvents.Connect);
ep.removeCustomEventListener(JambonzTranscriptionEvents.ConnectFailure);
ep.removeCustomEventListener(JambonzTranscriptionEvents.Error);
ep.removeCustomEventListener(AssemblyAiTranscriptionEvents.Transcription);
ep.removeCustomEventListener(AssemblyAiTranscriptionEvents.Connect);
ep.removeCustomEventListener(AssemblyAiTranscriptionEvents.ConnectFailure);
};
const setSpeechCredentialsAtRuntime = (recognizer) => {
if (!recognizer) return;
if (recognizer.vendor === 'nuance') {
@@ -752,6 +828,17 @@ module.exports = (logger) => {
if (clientId && secret) return {client_id: clientId, secret};
if (kryptonEndpoint) return {nuance_stt_uri: kryptonEndpoint};
}
else if (recognizer.vendor === 'aws') {
const {accessKey, secretKey, region, securityToken} = recognizer.awsOptions || {};
if (accessKey || secretKey || region || securityToken) {
return {
accessKeyId: accessKey,
secretAccessKey: secretKey,
region,
securityToken
};
}
}
else if (recognizer.vendor === 'nvidia') {
const {rivaUri} = recognizer.nvidiaOptions || {};
if (rivaUri) return {riva_uri: rivaUri};
@@ -783,6 +870,7 @@ module.exports = (logger) => {
return {
normalizeTranscription,
setChannelVarsForStt,
removeSpeechListeners,
setSpeechCredentialsAtRuntime,
compileSonioxTranscripts,
consolidateTranscripts

View File

@@ -9,8 +9,7 @@ const {
JAMBONES_WS_PING_INTERVAL_MS,
MAX_RECONNECTS,
JAMBONES_WS_HANDSHAKE_TIMEOUT_MS,
JAMBONES_WS_MAX_PAYLOAD,
HTTP_USER_AGENT_HEADER
JAMBONES_WS_MAX_PAYLOAD
} = require('../config');
class WsRequestor extends BaseRequestor {
@@ -229,9 +228,6 @@ class WsRequestor extends BaseRequestor {
maxRedirects: 2,
handshakeTimeout,
maxPayload: JAMBONES_WS_MAX_PAYLOAD ? parseInt(JAMBONES_WS_MAX_PAYLOAD) : 24 * 1024,
headers: {
...(HTTP_USER_AGENT_HEADER && {'user-agent' : HTTP_USER_AGENT_HEADER})
}
};
if (this.username && this.password) opts = {...opts, auth: `${this.username}:${this.password}`};

838
package-lock.json generated

File diff suppressed because it is too large Load Diff

View File

@@ -31,23 +31,23 @@
"@jambonz/http-health-check": "^0.0.1",
"@jambonz/mw-registrar": "^0.2.4",
"@jambonz/realtimedb-helpers": "^0.8.7",
"@jambonz/speech-utils": "^0.0.33",
"@jambonz/speech-utils": "^0.0.30",
"@jambonz/stats-collector": "^0.1.9",
"@jambonz/time-series": "^0.2.8",
"@jambonz/verb-specifications": "^0.0.50",
"@opentelemetry/api": "^1.4.0",
"@opentelemetry/exporter-jaeger": "^1.9.0",
"@opentelemetry/exporter-trace-otlp-http": "^0.35.0",
"@opentelemetry/exporter-zipkin": "^1.9.0",
"@opentelemetry/instrumentation": "^0.35.0",
"@opentelemetry/resources": "^1.9.0",
"@opentelemetry/sdk-trace-base": "^1.9.0",
"@opentelemetry/sdk-trace-node": "^1.9.0",
"@opentelemetry/semantic-conventions": "^1.9.0",
"@jambonz/verb-specifications": "^0.0.49",
"@opentelemetry/api": "^1.7.0",
"@opentelemetry/exporter-jaeger": "^1.18.1",
"@opentelemetry/exporter-trace-otlp-http": "^0.45.1",
"@opentelemetry/exporter-zipkin": "^1.18.1",
"@opentelemetry/instrumentation": "^0.45.1",
"@opentelemetry/resources": "^1.18.1",
"@opentelemetry/sdk-trace-base": "^1.18.1",
"@opentelemetry/sdk-trace-node": "^1.18.1",
"@opentelemetry/semantic-conventions": "^1.18.1",
"bent": "^7.3.12",
"debug": "^4.3.4",
"deepcopy": "^2.1.0",
"drachtio-fsmrf": "^3.0.33",
"drachtio-fsmrf": "^3.0.28",
"drachtio-srf": "^4.5.31",
"express": "^4.18.2",
"express-validator": "^7.0.1",

View File

@@ -99,8 +99,6 @@ test('test create-call call-hook basic authentication', async(t) => {
let obj = await getJSON(`http:127.0.0.1:3100/lastRequest/${from}`)
t.ok(obj.headers.Authorization = 'Basic dXNlcm5hbWU6cGFzc3dvcmQ=',
'create-call: call-hook contains basic authentication header');
t.ok(obj.headers['user-agent'] = 'jambonz',
'create-call: call-hook contains user-agent header');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);

View File

@@ -42,7 +42,7 @@ services:
ipv4_address: 172.38.0.7
drachtio:
image: drachtio/drachtio-server:0.8.24
image: drachtio/drachtio-server:0.8.22
restart: always
command: drachtio --contact "sip:*;transport=udp" --mtu 4096 --address 0.0.0.0 --port 9022
ports:
@@ -57,7 +57,7 @@ services:
condition: service_healthy
freeswitch:
image: drachtio/drachtio-freeswitch-mrf:0.6.1
image: drachtio/drachtio-freeswitch-mrf:0.4.33
restart: always
command: freeswitch --rtp-range-start 20000 --rtp-range-end 20100
environment:

View File

@@ -1,53 +0,0 @@
const test = require('tape');
const { sippUac } = require('./sipp')('test_fs');
const clearModule = require('clear-module');
const {provisionCallHook, provisionCustomHook} = require('./utils')
const bent = require('bent');
const getJSON = bent('json')
process.on('unhandledRejection', (reason, p) => {
console.log('Unhandled Rejection at: Promise', p, 'reason:', reason);
});
function connect(connectable) {
return new Promise((resolve, reject) => {
connectable.on('connect', () => {
return resolve();
});
});
}
test('\'hangup\' custom headers', async(t) => {
clearModule.all();
const {srf, disconnect} = require('../app');
try {
await connect(srf);
// GIVEN
const verbs = [
{
verb: 'play',
url: 'https://example.com/example.mp3'
},
{
"verb": "hangup",
"headers": {
"X-Reason" : "maximum call duration exceeded"
}
}
];
const from = 'hangup_custom_headers';
await provisionCallHook(from, verbs)
// THEN
await sippUac('uac-success-received-bye.xml', '172.38.0.10', from);
t.pass('play: succeeds when using single link');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);
disconnect();
t.error(err);
}
});

View File

@@ -16,8 +16,7 @@ require('./listen-tests');
require('./config-test');
require('./queue-test');
require('./in-dialog-test');
require('./hangup-test');
require('./sdp-utils-test');
require('./http-proxy-test');
require('./sdp-utils-test');
require('./remove-test-db');
require('./docker_stop');

View File

@@ -188,7 +188,7 @@ test('\'play\' tests with seekOffset and actionHook', async(t) => {
const seconds = parseInt(obj.body.playback_seconds);
const milliseconds = parseInt(obj.body.playback_milliseconds);
const lastOffsetPos = parseInt(obj.body.playback_last_offset_pos);
console.log({obj}, 'lastRequest');
//console.log({obj}, 'lastRequest');
t.ok(obj.body.reason === "playCompleted", "play: actionHook success received");
t.ok(seconds === 2, "playback_seconds: actionHook success received");
t.ok(milliseconds === 2048, "playback_milliseconds: actionHook success received");

View File

@@ -52,7 +52,6 @@ test('\'transcribe\' test - google', async(t) => {
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
'transcribe: succeeds when using google credentials');
@@ -90,7 +89,6 @@ test('\'transcribe\' test - microsoft', async(t) => {
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
'transcribe: succeeds when using microsoft credentials');
@@ -128,7 +126,6 @@ test('\'transcribe\' test - aws', async(t) => {
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
'transcribe: succeeds when using aws credentials');
@@ -140,71 +137,6 @@ test('\'transcribe\' test - aws', async(t) => {
}
});
test('\'transcribe\' test - deepgram config options', async(t) => {
if (!DEEPGRAM_API_KEY ) {
t.pass('skipping deepgram tests');
return t.end();
}
clearModule.all();
const {srf, disconnect} = require('../app');
try {
await connect(srf);
// GIVEN
let verbs = [
{
"verb": "config",
"recognizer": {
"vendor": "deepgram",
"language": "en-US",
"altLanguages": [
"en-US"
],
"deepgramOptions": {
"model": "2-ea",
"tier": "nova",
"numerals": true,
"ner": true,
"vadTurnoff": 10,
"keywords": [
"CPT"
]
}
}
},
{
"verb": "transcribe",
"transcriptionHook": "/transcriptionHook",
"recognizer": {
"vendor": "deepgram",
"altLanguages": [
"en-AU"
],
"hints": ["customer support", "sales", "human resources", "HR"],
"deepgramOptions": {
"apiKey": DEEPGRAM_API_KEY,
}
}
}
];
let from = "gather_success";
await provisionCallHook(from, verbs);
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().includes('like to speak to customer support'),
'transcribe: succeeds when using deepgram credentials');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);
disconnect();
t.error(err);
}
});
test('\'transcribe\' test - deepgram', async(t) => {
if (!DEEPGRAM_API_KEY ) {
t.pass('skipping deepgram tests');
@@ -234,7 +166,6 @@ test('\'transcribe\' test - deepgram', async(t) => {
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().includes('like to speak to customer support'),
'transcribe: succeeds when using deepgram credentials');
@@ -314,133 +245,9 @@ test('\'transcribe\' test - google with asrTimeout', async(t) => {
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
'transcribe: succeeds when using google credentials');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);
disconnect();
t.error(err);
}
});
test('\'transcribe\' test - deepgram config options altLanguages', async(t) => {
if (!DEEPGRAM_API_KEY ) {
t.pass('skipping deepgram tests');
return t.end();
}
clearModule.all();
const {srf, disconnect} = require('../app');
try {
await connect(srf);
// GIVEN
let verbs = [
{
"verb": "config",
"recognizer": {
"vendor": "deepgram",
"language": "en-US",
"altLanguages": [
"en-US"
],
"deepgramOptions": {
"model": "2-ea",
"tier": "nova",
"numerals": true,
"ner": true,
"vadTurnoff": 10,
"keywords": [
"CPT"
]
}
}
},
{
"verb": "transcribe",
"transcriptionHook": "/transcriptionHook",
"recognizer": {
"vendor": "deepgram",
"hints": ["customer support", "sales", "human resources", "HR"],
"deepgramOptions": {
"apiKey": DEEPGRAM_API_KEY,
}
}
}
];
let from = "gather_success_no_altLanguages";
await provisionCallHook(from, verbs);
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().includes('like to speak to customer support'),
'transcribe: succeeds when using deepgram credentials');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);
disconnect();
t.error(err);
}
});
test('\'transcribe\' test - deepgram config options altLanguages', async(t) => {
if (!DEEPGRAM_API_KEY ) {
t.pass('skipping deepgram tests');
return t.end();
}
clearModule.all();
const {srf, disconnect} = require('../app');
try {
await connect(srf);
// GIVEN
let verbs = [
{
"verb": "config",
"recognizer": {
"vendor": "deepgram",
"language": "en-US",
"altLanguages": [
"en-US"
],
"deepgramOptions": {
"model": "2-ea",
"tier": "nova",
"numerals": true,
"ner": true,
"vadTurnoff": 10,
"keywords": [
"CPT"
]
}
}
},
{
"verb": "transcribe",
"transcriptionHook": "/transcriptionHook",
"recognizer": {
"vendor": "deepgram",
"hints": ["customer support", "sales", "human resources", "HR"],
"altLanguages": [],
"deepgramOptions": {
"apiKey": DEEPGRAM_API_KEY,
}
}
}
];
let from = "gather_success_has_altLanguages";
await provisionCallHook(from, verbs);
// THEN
await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
//console.log(JSON.stringify(obj));
t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().includes('like to speak to customer support'),
'transcribe: succeeds when using deepgram credentials');
disconnect();
} catch (err) {
console.log(`error received: ${err}`);