fix to be more precise about removing custom event handlers so that w… (#580)

* fix to be more precise about removing custom event handlers so that when we stop a gather we dont also inadvertently stop a background transcribe as well

* test fixes

* fix: endpointing=false was being ignored for Deepgram
This commit is contained in:
Dave Horton
2023-12-28 11:00:27 -05:00
committed by GitHub
parent 9d8291f892
commit 2c48083c26
8 changed files with 94 additions and 122 deletions

View File

@@ -221,7 +221,7 @@ class TaskGather extends SttTask {
} catch (err) {
this.logger.error(err, 'TaskGather:exec error');
}
this.removeSpeechListeners(ep);
this.removeCustomEventListeners();
}
kill(cs) {
@@ -261,7 +261,7 @@ class TaskGather extends SttTask {
else if (this.input.includes('digits')) {
if (this.digitBuffer.length === 0 && this.needsStt) {
// DTMF is higher priority than STT.
this.removeSpeechListeners(ep);
this.removeCustomEventListeners();
ep.stopTranscription({
vendor: this.vendor,
bugname: this.bugname,
@@ -305,33 +305,37 @@ class TaskGather extends SttTask {
switch (this.vendor) {
case 'google':
this.bugname = `${this.bugname_prefix}google_transcribe`;
ep.addCustomEventListener(GoogleTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(GoogleTranscriptionEvents.EndOfUtterance, this._onEndOfUtterance.bind(this, cs, ep));
ep.addCustomEventListener(GoogleTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.EndOfUtterance, this._onEndOfUtterance.bind(this, cs, ep));
this.addCustomEventListener(
ep, GoogleTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'aws':
case 'polly':
this.bugname = `${this.bugname_prefix}aws_transcribe`;
ep.addCustomEventListener(AwsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(AwsTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
this.addCustomEventListener(ep, AwsTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, AwsTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'microsoft':
this.bugname = `${this.bugname_prefix}azure_transcribe`;
ep.addCustomEventListener(AzureTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(AzureTranscriptionEvents.NoSpeechDetected,
this.addCustomEventListener(
ep, AzureTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, AzureTranscriptionEvents.NoSpeechDetected,
this._onNoSpeechDetected.bind(this, cs, ep));
ep.addCustomEventListener(AzureTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
this.addCustomEventListener(ep, AzureTranscriptionEvents.VadDetected, this._onVadDetected.bind(this, cs, ep));
break;
case 'nuance':
this.bugname = `${this.bugname_prefix}nuance_transcribe`;
ep.addCustomEventListener(NuanceTranscriptionEvents.Transcription,
this.addCustomEventListener(ep, NuanceTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(NuanceTranscriptionEvents.StartOfSpeech,
this.addCustomEventListener(ep, NuanceTranscriptionEvents.StartOfSpeech,
this._onStartOfSpeech.bind(this, cs, ep));
ep.addCustomEventListener(NuanceTranscriptionEvents.TranscriptionComplete,
this.addCustomEventListener(ep, NuanceTranscriptionEvents.TranscriptionComplete,
this._onTranscriptionComplete.bind(this, cs, ep));
ep.addCustomEventListener(NuanceTranscriptionEvents.VadDetected,
this.addCustomEventListener(ep, NuanceTranscriptionEvents.VadDetected,
this._onVadDetected.bind(this, cs, ep));
/* stall timers until prompt finishes playing */
@@ -342,9 +346,10 @@ class TaskGather extends SttTask {
case 'deepgram':
this.bugname = `${this.bugname_prefix}deepgram_transcribe`;
ep.addCustomEventListener(DeepgramTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(DeepgramTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
ep.addCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure,
this.addCustomEventListener(
ep, DeepgramTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
this.addCustomEventListener(ep, DeepgramTranscriptionEvents.ConnectFailure,
this._onVendorConnectFailure.bind(this, cs, ep));
/* if app sets deepgramOptions.utteranceEndMs they essentially want continuous asr */
@@ -353,12 +358,14 @@ class TaskGather extends SttTask {
case 'soniox':
this.bugname = `${this.bugname_prefix}soniox_transcribe`;
ep.addCustomEventListener(SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
break;
case 'cobalt':
this.bugname = `${this.bugname_prefix}cobalt_transcribe`;
ep.addCustomEventListener(CobaltTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
this.addCustomEventListener(
ep, CobaltTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
/* cobalt doesnt have language, it has model, which is required */
if (!this.data.recognizer.model) {
@@ -388,21 +395,21 @@ class TaskGather extends SttTask {
case 'ibm':
this.bugname = `${this.bugname_prefix}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.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._onVendorConnectFailure.bind(this, cs, ep));
break;
case 'nvidia':
this.bugname = `${this.bugname_prefix}nvidia_transcribe`;
ep.addCustomEventListener(NvidiaTranscriptionEvents.Transcription,
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
ep.addCustomEventListener(NvidiaTranscriptionEvents.StartOfSpeech,
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.StartOfSpeech,
this._onStartOfSpeech.bind(this, cs, ep));
ep.addCustomEventListener(NvidiaTranscriptionEvents.TranscriptionComplete,
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.TranscriptionComplete,
this._onTranscriptionComplete.bind(this, cs, ep));
ep.addCustomEventListener(NvidiaTranscriptionEvents.VadDetected,
this.addCustomEventListener(ep, NvidiaTranscriptionEvents.VadDetected,
this._onVadDetected.bind(this, cs, ep));
/* I think nvidia has this (??) - stall timers until prompt finishes playing */
@@ -413,19 +420,21 @@ class TaskGather extends SttTask {
case 'assemblyai':
this.bugname = `${this.bugname_prefix}assemblyai_transcribe`;
ep.addCustomEventListener(AssemblyAiTranscriptionEvents.Transcription,
this.addCustomEventListener(ep, AssemblyAiTranscriptionEvents.Transcription,
this._onTranscription.bind(this, cs, ep));
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.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,
this._onVendorConnectFailure.bind(this, cs, ep));
break;
default:
if (this.vendor.startsWith('custom:')) {
this.bugname = `${this.bugname_prefix}${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.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._onVendorConnectFailure.bind(this, cs, ep));
break;
}
@@ -437,7 +446,7 @@ class TaskGather extends SttTask {
}
/* common handler for all stt engine errors */
ep.addCustomEventListener(JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
this.addCustomEventListener(ep, JambonzTranscriptionEvents.Error, this._onJambonzError.bind(this, cs, ep));
await ep.set(opts)
.catch((err) => this.logger.info(err, 'Error setting channel variables'));
}