ben.carrasco/jambonz-feature-server
1
0
Fork 0
mirror of https://github.com/jambonz/jambonz-feature-server.git synced 2025-12-20 16:50:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

Feat/speechmatics (#932)

Browse Source 
* wip

* initial working version of speechmatics

* linting
This commit is contained in:
Dave Horton
2024-10-12 18:42:53 -04:00
committed by GitHub
parent 84b7456c2d
commit f84b3793e1
7 changed files with 175 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
lib/tasks/gather.js
View File

@@ -12,7 +12,8 @@ const {
JambonzTranscriptionEvents,
AssemblyAiTranscriptionEvents,
VadDetection,
VerbioTranscriptionEvents
VerbioTranscriptionEvents,
SpeechmaticsTranscriptionEvents
} = require('../utils/constants.json');
const {
JAMBONES_GATHER_EARLY_HINTS_MATCH,
@@ -514,6 +515,24 @@ class TaskGather extends SttTask {
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
case 'speechmatics':
this.bugname = `${this.bugname_prefix}speechmatics_transcribe`;
this.addCustomEventListener(
ep, SpeechmaticsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Info,
this._onSpeechmaticsInfo.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.RecognitionStarted,
this._onSpeechmaticsRecognitionStarted.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Error,
this._onSpeechmaticsErrror.bind(this, cs, ep));
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
@@ -752,7 +771,7 @@ class TaskGather extends SttTask {
evt = this.normalizeTranscription(evt, this.vendor, 1, this.language,
this.shortUtterance, this.data.recognizer.punctuation);
//this.logger.debug({evt, bugname, finished, vendor: this.vendor}, 'Gather:_onTranscription normalized transcript');
this.logger.debug({evt, bugname, finished, vendor: this.vendor}, 'Gather:_onTranscription normalized transcript');
if (evt.alternatives.length === 0) {
this.logger.info({evt}, 'TaskGather:_onTranscription - got empty transcript, continue listening');
@@ -1006,12 +1025,25 @@ class TaskGather extends SttTask {
}
}
async _onSpeechmaticsErrror(cs, _ep, evt) {
// eslint-disable-next-line no-unused-vars
const {message, ...e} = evt;
this._onVendorError(cs, _ep, {error: JSON.stringify(e)});
}
async _onVendorError(cs, _ep, evt) {
super._onVendorError(cs, _ep, evt);
if (!(await this._startFallback(cs, _ep, evt))) {
this._resolve('stt-error', evt);
}
}
async _onSpeechmaticsRecognitionStarted(_cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onSpeechmaticsRecognitionStarted');
}
async _onSpeechmaticsInfo(_cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onSpeechmaticsInfo');
}
_onVadDetected(cs, ep) {
if (this.bargein && this.minBargeinWordCount === 0) {

4
lib/tasks/stt-task.js
View File

@@ -17,12 +17,14 @@ class SttTask extends Task {
normalizeTranscription,
setSpeechCredentialsAtRuntime,
compileSonioxTranscripts,
consolidateTranscripts
consolidateTranscripts,
updateSpeechmaticsPayload
} = require('../utils/transcription-utils')(logger);
this.setChannelVarsForStt = setChannelVarsForStt;
this.normalizeTranscription = normalizeTranscription;
this.compileSonioxTranscripts = compileSonioxTranscripts;
this.consolidateTranscripts = consolidateTranscripts;
this.updateSpeechmaticsPayload = updateSpeechmaticsPayload;
this.eventHandlers = [];
this.isHandledByPrimaryProvider = true;
/**

41
lib/tasks/transcribe.js
View File

@@ -12,7 +12,9 @@ const {
NvidiaTranscriptionEvents,
JambonzTranscriptionEvents,
TranscribeStatus,
AssemblyAiTranscriptionEvents
AssemblyAiTranscriptionEvents,
VerbioTranscriptionEvents,
SpeechmaticsTranscriptionEvents
} = require('../utils/constants.json');
const { normalizeJambones } = require('@jambonz/verb-specifications');
const SttTask = require('./stt-task');
@@ -237,6 +239,13 @@ class TaskTranscribe extends SttTask {
this.addCustomEventListener(ep, SonioxTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep, channel));
break;
case 'verbio':
this.bugname = `${this.bugname_prefix}verbio_transcribe`;
this.addCustomEventListener(
ep, VerbioTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
break;
case 'cobalt':
this.bugname = `${this.bugname_prefix}cobalt_transcribe`;
this.addCustomEventListener(ep, CobaltTranscriptionEvents.Transcription,
@@ -294,6 +303,22 @@ class TaskTranscribe extends SttTask {
this._onVendorConnectFailure.bind(this, cs, ep, channel));
break;
case 'speechmatics':
this.bugname = `${this.bugname_prefix}speechmatics_transcribe`;
this.addCustomEventListener(
ep, SpeechmaticsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Info,
this._onSpeechmaticsInfo.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.RecognitionStarted,
this._onSpeechmaticsRecognitionStarted.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Connect,
this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
this.addCustomEventListener(ep, SpeechmaticsTranscriptionEvents.Error,
this._onSpeechmaticsError.bind(this, cs, ep));
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${this.vendor}_transcribe`;
@@ -644,6 +669,20 @@ class TaskTranscribe extends SttTask {
}
}
async _onSpeechmaticsRecognitionStarted(_cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onSpeechmaticsRecognitionStarted');
}
async _onSpeechmaticsInfo(_cs, _ep, evt) {
this.logger.debug({evt}, 'TaskGather:_onSpeechmaticsInfo');
}
async _onSpeechmaticsErrror(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);