@mastra/voice-deepgram 0.12.0-beta.0 → 0.12.0-beta.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +9 -0
- package/README.md +10 -7
- package/dist/index.cjs +54 -21
- package/dist/index.cjs.map +1 -1
- package/dist/index.d.ts +16 -1
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +54 -21
- package/dist/index.js.map +1 -1
- package/package.json +7 -5
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,14 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 0.12.0-beta.1
|
|
4
|
+
|
|
5
|
+
### Patch Changes
|
|
6
|
+
|
|
7
|
+
- feat(voice-deepgram): add speaker diarization support for STT ([#10206](https://github.com/mastra-ai/mastra/pull/10206))
|
|
8
|
+
|
|
9
|
+
- Updated dependencies [[`21a15de`](https://github.com/mastra-ai/mastra/commit/21a15de369fe82aac26bb642ed7be73505475e8b), [`feb7ee4`](https://github.com/mastra-ai/mastra/commit/feb7ee4d09a75edb46c6669a3beaceec78811747), [`b0e2ea5`](https://github.com/mastra-ai/mastra/commit/b0e2ea5b52c40fae438b9e2f7baee6f0f89c5442), [`c456e01`](https://github.com/mastra-ai/mastra/commit/c456e0149e3c176afcefdbd9bb1d2c5917723725), [`ab035c2`](https://github.com/mastra-ai/mastra/commit/ab035c2ef6d8cc7bb25f06f1a38508bd9e6f126b), [`1a46a56`](https://github.com/mastra-ai/mastra/commit/1a46a566f45a3fcbadc1cf36bf86d351f264bfa3), [`3cf540b`](https://github.com/mastra-ai/mastra/commit/3cf540b9fbfea8f4fc8d3a2319a4e6c0b0cbfd52), [`1c6ce51`](https://github.com/mastra-ai/mastra/commit/1c6ce51f875915ab57fd36873623013699a2a65d), [`898a972`](https://github.com/mastra-ai/mastra/commit/898a9727d286c2510d6b702dfd367e6aaf5c6b0f), [`a97003a`](https://github.com/mastra-ai/mastra/commit/a97003aa1cf2f4022a41912324a1e77263b326b8), [`ccc141e`](https://github.com/mastra-ai/mastra/commit/ccc141ed27da0abc3a3fc28e9e5128152e8e37f4), [`fe3b897`](https://github.com/mastra-ai/mastra/commit/fe3b897c2ccbcd2b10e81b099438c7337feddf89), [`00123ba`](https://github.com/mastra-ai/mastra/commit/00123ba96dc9e5cd0b110420ebdba56d8f237b25), [`29c4309`](https://github.com/mastra-ai/mastra/commit/29c4309f818b24304c041bcb4a8f19b5f13f6b62), [`16785ce`](https://github.com/mastra-ai/mastra/commit/16785ced928f6f22638f4488cf8a125d99211799), [`de8239b`](https://github.com/mastra-ai/mastra/commit/de8239bdcb1d8c0cfa06da21f1569912a66bbc8a), [`b5e6cd7`](https://github.com/mastra-ai/mastra/commit/b5e6cd77fc8c8e64e0494c1d06cee3d84e795d1e), [`3759cb0`](https://github.com/mastra-ai/mastra/commit/3759cb064935b5f74c65ac2f52a1145f7352899d), [`651e772`](https://github.com/mastra-ai/mastra/commit/651e772eb1475fb13e126d3fcc01751297a88214), [`b61b93f`](https://github.com/mastra-ai/mastra/commit/b61b93f9e058b11dd2eec169853175d31dbdd567), [`bae33d9`](https://github.com/mastra-ai/mastra/commit/bae33d91a63fbb64d1e80519e1fc1acaed1e9013), [`c0b731f`](https://github.com/mastra-ai/mastra/commit/c0b731fb27d712dc8582e846df5c0332a6a0c5ba), [`43ca8f2`](https://github.com/mastra-ai/mastra/commit/43ca8f2c7334851cc7b4d3d2f037d8784bfbdd5f), [`2ca67cc`](https://github.com/mastra-ai/mastra/commit/2ca67cc3bb1f6a617353fdcab197d9efebe60d6f), [`9e67002`](https://github.com/mastra-ai/mastra/commit/9e67002b52c9be19936c420a489dbee9c5fd6a78), [`35edc49`](https://github.com/mastra-ai/mastra/commit/35edc49ac0556db609189641d6341e76771b81fc)]:
|
|
10
|
+
- @mastra/core@1.0.0-beta.5
|
|
11
|
+
|
|
3
12
|
## 0.12.0-beta.0
|
|
4
13
|
|
|
5
14
|
### Minor Changes
|
package/README.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
# @mastra/voice-deepgram
|
|
2
2
|
|
|
3
|
-
Deepgram voice integration for Mastra, providing both Text-to-Speech (TTS) and Speech-to-Text (STT)
|
|
3
|
+
Deepgram voice integration for Mastra, providing both Text-to-Speech (TTS) and Speech-to-Text (STT) using Deepgram's Aura (TTS) and Nova (STT) families.
|
|
4
4
|
|
|
5
5
|
## Installation
|
|
6
6
|
|
|
@@ -24,14 +24,14 @@ import { DeepgramVoice } from '@mastra/voice-deepgram';
|
|
|
24
24
|
// Create voice with both speech and listening capabilities
|
|
25
25
|
const voice = new DeepgramVoice({
|
|
26
26
|
speechModel: {
|
|
27
|
-
name: 'aura
|
|
27
|
+
name: 'aura', // TTS family
|
|
28
28
|
apiKey: 'your-api-key', // Optional, can use DEEPGRAM_API_KEY env var
|
|
29
29
|
},
|
|
30
30
|
listeningModel: {
|
|
31
|
-
name: 'nova', //
|
|
31
|
+
name: 'nova', // STT family
|
|
32
32
|
apiKey: 'your-api-key', // Optional, can use DEEPGRAM_API_KEY env var
|
|
33
33
|
},
|
|
34
|
-
speaker: '
|
|
34
|
+
speaker: 'asteria-en', // default voiceId (see voice.ts)
|
|
35
35
|
});
|
|
36
36
|
|
|
37
37
|
// List available voices
|
|
@@ -39,12 +39,15 @@ const voices = await voice.getSpeakers();
|
|
|
39
39
|
|
|
40
40
|
// Generate speech
|
|
41
41
|
const audioStream = await voice.speak('Hello from Mastra!', {
|
|
42
|
-
speaker: '
|
|
43
|
-
speed: 1.0, // Optional: adjust speech speed
|
|
42
|
+
speaker: 'hera-en', // override speaker voice
|
|
44
43
|
});
|
|
45
44
|
|
|
46
45
|
// Convert speech to text
|
|
47
|
-
const
|
|
46
|
+
const result = await voice.listen(audioStream, {
|
|
47
|
+
diarize: true,
|
|
48
|
+
diarize_speaker_count: 2,
|
|
49
|
+
});
|
|
50
|
+
console.log(result.transcript);
|
|
48
51
|
```
|
|
49
52
|
|
|
50
53
|
## Features
|
package/dist/index.cjs
CHANGED
|
@@ -26,6 +26,9 @@ var DEEPGRAM_VOICES = [
|
|
|
26
26
|
var DeepgramVoice = class extends voice.MastraVoice {
|
|
27
27
|
speechClient;
|
|
28
28
|
listeningClient;
|
|
29
|
+
storedSpeechModel;
|
|
30
|
+
storedListeningModel;
|
|
31
|
+
storedSpeaker;
|
|
29
32
|
constructor({
|
|
30
33
|
speechModel,
|
|
31
34
|
listeningModel,
|
|
@@ -51,6 +54,14 @@ var DeepgramVoice = class extends voice.MastraVoice {
|
|
|
51
54
|
},
|
|
52
55
|
speaker
|
|
53
56
|
});
|
|
57
|
+
this.storedSpeechModel = {
|
|
58
|
+
name: speechModel?.name ?? defaultSpeechModel.name,
|
|
59
|
+
apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey
|
|
60
|
+
};
|
|
61
|
+
this.storedListeningModel = {
|
|
62
|
+
name: listeningModel?.name ?? defaultListeningModel.name,
|
|
63
|
+
apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey
|
|
64
|
+
};
|
|
54
65
|
const speechApiKey = speechModel?.apiKey || defaultApiKey;
|
|
55
66
|
const listeningApiKey = listeningModel?.apiKey || defaultApiKey;
|
|
56
67
|
if (!speechApiKey && !listeningApiKey) {
|
|
@@ -62,7 +73,7 @@ var DeepgramVoice = class extends voice.MastraVoice {
|
|
|
62
73
|
if (listeningApiKey) {
|
|
63
74
|
this.listeningClient = sdk.createClient(listeningApiKey);
|
|
64
75
|
}
|
|
65
|
-
this.
|
|
76
|
+
this.storedSpeaker = speaker || "asteria-en";
|
|
66
77
|
}
|
|
67
78
|
async getSpeakers() {
|
|
68
79
|
return DEEPGRAM_VOICES.map((voice) => ({
|
|
@@ -90,21 +101,15 @@ var DeepgramVoice = class extends voice.MastraVoice {
|
|
|
90
101
|
if (text.trim().length === 0) {
|
|
91
102
|
throw new Error("Input text is empty");
|
|
92
103
|
}
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
}
|
|
96
|
-
let model;
|
|
97
|
-
if (options?.speaker) {
|
|
98
|
-
model = this.speechModel?.name + "-" + options.speaker;
|
|
99
|
-
} else if (this.speaker) {
|
|
100
|
-
model = this.speechModel?.name + "-" + this.speaker;
|
|
101
|
-
}
|
|
104
|
+
const baseModel = this.storedSpeechModel?.name;
|
|
105
|
+
const speakerId = options?.speaker || this.storedSpeaker;
|
|
106
|
+
const modelName = baseModel && speakerId ? speakerId.startsWith(`${baseModel}-`) ? speakerId : `${baseModel}-${speakerId}` : baseModel || speakerId;
|
|
102
107
|
const speakClient = this.speechClient.speak;
|
|
103
108
|
const response = await speakClient.request(
|
|
104
109
|
{ text },
|
|
105
110
|
{
|
|
106
|
-
model,
|
|
107
|
-
...options
|
|
111
|
+
model: modelName,
|
|
112
|
+
...Object.fromEntries(Object.entries(options ?? {}).filter(([k]) => k !== "speaker"))
|
|
108
113
|
}
|
|
109
114
|
);
|
|
110
115
|
const webStream = await response.getStream();
|
|
@@ -139,6 +144,17 @@ var DeepgramVoice = class extends voice.MastraVoice {
|
|
|
139
144
|
async getListener() {
|
|
140
145
|
return { enabled: true };
|
|
141
146
|
}
|
|
147
|
+
/**
|
|
148
|
+
* Transcribes audio with optional speaker diarization.
|
|
149
|
+
*
|
|
150
|
+
* @param audioStream - Audio input stream
|
|
151
|
+
* @param options - Transcription options (diarize, language, etc.)
|
|
152
|
+
* @returns Promise resolving to:
|
|
153
|
+
* - transcript: Full transcript string
|
|
154
|
+
* - words: Array of word objects with timing and confidence
|
|
155
|
+
* - raw: Complete Deepgram API response
|
|
156
|
+
* - speakerSegments: (when diarize=true) Array of {word, speaker, start, end}
|
|
157
|
+
*/
|
|
142
158
|
async listen(audioStream, options) {
|
|
143
159
|
if (!this.listeningClient) {
|
|
144
160
|
throw new Error("Deepgram listening client not configured");
|
|
@@ -152,21 +168,38 @@ var DeepgramVoice = class extends voice.MastraVoice {
|
|
|
152
168
|
}
|
|
153
169
|
}
|
|
154
170
|
const buffer = Buffer.concat(chunks);
|
|
155
|
-
|
|
156
|
-
throw new Error("No listening client configured");
|
|
157
|
-
}
|
|
171
|
+
const { diarize, diarize_speaker_count: _, ...restOptions } = options ?? {};
|
|
158
172
|
const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {
|
|
159
|
-
|
|
160
|
-
|
|
173
|
+
...restOptions,
|
|
174
|
+
model: this.storedListeningModel?.name,
|
|
175
|
+
diarize
|
|
161
176
|
});
|
|
162
177
|
if (error) {
|
|
163
178
|
throw error;
|
|
164
179
|
}
|
|
165
|
-
const
|
|
166
|
-
|
|
167
|
-
|
|
180
|
+
const channel = result.results?.channels?.[0];
|
|
181
|
+
const alt = channel?.alternatives?.[0];
|
|
182
|
+
if (!alt) {
|
|
183
|
+
return {
|
|
184
|
+
transcript: "",
|
|
185
|
+
words: [],
|
|
186
|
+
raw: result
|
|
187
|
+
};
|
|
188
|
+
}
|
|
189
|
+
const response = {
|
|
190
|
+
transcript: alt.transcript,
|
|
191
|
+
words: alt.words,
|
|
192
|
+
raw: result
|
|
193
|
+
};
|
|
194
|
+
if (diarize && alt.words) {
|
|
195
|
+
response.speakerSegments = alt.words.map((w) => ({
|
|
196
|
+
word: w.word,
|
|
197
|
+
speaker: w.speaker,
|
|
198
|
+
start: w.start,
|
|
199
|
+
end: w.end
|
|
200
|
+
}));
|
|
168
201
|
}
|
|
169
|
-
return
|
|
202
|
+
return response;
|
|
170
203
|
}
|
|
171
204
|
};
|
|
172
205
|
|
package/dist/index.cjs.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/voices.ts","../src/index.ts"],"names":["MastraVoice","createClient","PassThrough"],"mappings":";;;;;;;;;AAKO,IAAM,eAAA,GAAkB;AAAA,EAC7B,YAAA;AAAA,EACA,SAAA;AAAA,EACA,WAAA;AAAA,EACA,WAAA;AAAA,EACA,SAAA;AAAA,EACA,UAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,WAAA;AAAA,EACA;AACF,CAAA;;;ACHO,IAAM,aAAA,GAAN,cAA4BA,iBAAA,CAAY;AAAA,EACrC,YAAA;AAAA,EACA,eAAA;AAAA,EAER,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA;AAAA,GACF,GAA4G,EAAC,EAAG;AAC9G,IAAA,MAAM,aAAA,GAAgB,QAAQ,GAAA,CAAI,gBAAA;AAElC,IAAA,MAAM,kBAAA,GAAqB;AAAA,MACzB,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,MAAM,qBAAA,GAAwB;AAAA,MAC5B,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,QAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,OACpD;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,QACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,OAC1D;AAAA,MACA;AAAA,KACD,CAAA;AAED,IAAA,MAAM,YAAA,GAAe,aAAa,MAAA,IAAU,aAAA;AAC5C,IAAA,MAAM,eAAA,GAAkB,gBAAgB,MAAA,IAAU,aAAA;AAElD,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,MAAM,4FAA4F,CAAA;AAAA,IAC9G;AAEA,IAAA,IAAI,YAAA,EAAc;AAChB,MAAA,IAAA,CAAK,YAAA,GAAeC,iBAAa,YAAY,CAAA;AAAA,IAC/C;AACA,IAAA,IAAI,eAAA,EAAiB;AACnB,MAAA,IAAA,CAAK,eAAA,GAAkBA,iBAAa,eAAe,CAAA;AAAA,IACrD;AAEA,IAAA,IAAA,CAAK,UAAU,OAAA,IAAW,YAAA;AAAA,EAC5B;AAAA,EAEA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,eAAA,CAAgB,IAAI,CAAA,KAAA,MAAU;AAAA,MACnC,OAAA,EAAS;AAAA,KACX,CAAE,CAAA;AAAA,EACJ;AAAA,EAEA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EAIgC;AAChC,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,uCAAuC,CAAA;AAAA,IACzD;AAEA,IAAA,IAAI,IAAA;AACJ,IAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,MAAA,MAAM,SAAmB,EAAC;AAC1B,MAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,QAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,UAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,QAChC,CAAA,MAAO;AACL,UAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,QACnB;AAAA,MACF;AACA,MAAA,IAAA,GAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,IAC/C,CAAA,MAAO;AACL,MAAA,IAAA,GAAO,KAAA;AAAA,IACT;AAEA,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,6BAA6B,CAAA;AAAA,IAC/C;AAEA,IAAA,IAAI,KAAA;AACJ,IAAA,IAAI,SAAS,OAAA,EAAS;AACpB,MAAA,KAAA,GAAQ,IAAA,CAAK,WAAA,EAAa,IAAA,GAAO,GAAA,GAAM,OAAA,CAAQ,OAAA;AAAA,IACjD,CAAA,MAAA,IAAW,KAAK,OAAA,EAAS;AACvB,MAAA,KAAA,GAAQ,IAAA,CAAK,WAAA,EAAa,IAAA,GAAO,GAAA,GAAM,IAAA,CAAK,OAAA;AAAA,IAC9C;AAEA,IAAA,MAAM,WAAA,GAAc,KAAK,YAAA,CAAa,KAAA;AACtC,IAAA,MAAM,QAAA,GAAW,MAAM,WAAA,CAAY,OAAA;AAAA,MACjC,EAAE,IAAA,EAAK;AAAA,MACP;AAAA,QACE,KAAA;AAAA,QACA,GAAG;AAAA;AACL,KACF;AAEA,IAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,SAAA,EAAU;AAC3C,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,IAAI,MAAM,kCAAkC,CAAA;AAAA,IACpD;AAEA,IAAA,MAAM,MAAA,GAAS,UAAU,SAAA,EAAU;AACnC,IAAA,MAAM,UAAA,GAAa,IAAIC,kBAAA,EAAY;AAGnC,IAAA,CAAC,YAAY;AACX,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AACR,YAAA,UAAA,CAAW,GAAA,EAAI;AACf,YAAA;AAAA,UACF;AACA,UAAA,UAAA,CAAW,MAAM,KAAK,CAAA;AAAA,QACxB;AAAA,MACF,SAAS,KAAA,EAAO;AACd,QAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,MACnC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,CAAA,KAAA,KAAS;AAClB,MAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,IACnC,CAAC,CAAA;AAED,IAAA,OAAO,UAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,SAAS,IAAA,EAAK;AAAA,EACzB;AAAA,EAEA,MAAM,MAAA,CACJ,WAAA,EACA,OAAA,EAGiB;AACjB,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,WAAA,EAAa;AACrC,MAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,QAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,MAChC,CAAA,MAAO;AACL,QAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,MACnB;AAAA,IACF;AACA,IAAA,MAAM,MAAA,GAAS,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA;AAEnC,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,gCAAgC,CAAA;AAAA,IAClD;AACA,IAAA,MAAM,EAAE,MAAA,EAAQ,KAAA,EAAM,GAAI,MAAM,KAAK,eAAA,CAAgB,MAAA,CAAO,WAAA,CAAY,cAAA,CAAe,MAAA,EAAQ;AAAA,MAC7F,KAAA,EAAO,KAAK,cAAA,EAAgB,IAAA;AAAA,MAC5B,GAAG;AAAA,KACJ,CAAA;AAED,IAAA,IAAI,KAAA,EAAO;AACT,MAAA,MAAM,KAAA;AAAA,IACR;AAEA,IAAA,MAAM,UAAA,GAAa,OAAO,OAAA,EAAS,QAAA,GAAW,CAAC,CAAA,EAAG,YAAA,GAAe,CAAC,CAAA,EAAG,UAAA;AACrE,IAAA,IAAI,CAAC,UAAA,EAAY;AACf,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,OAAO,UAAA;AAAA,EACT;AACF","file":"index.cjs","sourcesContent":["/**\n * List of available Deepgram voice models for text-to-speech\n * Each voice is designed for specific use cases and languages\n * Format: {name}-{language} (e.g. asteria-en)\n */\nexport const DEEPGRAM_VOICES = [\n 'asteria-en',\n 'luna-en',\n 'stella-en',\n 'athena-en',\n 'hera-en',\n 'orion-en',\n 'arcas-en',\n 'perseus-en',\n 'angus-en',\n 'orpheus-en',\n 'helios-en',\n 'zeus-en',\n] as const;\n\nexport type DeepgramVoiceId = (typeof DEEPGRAM_VOICES)[number];\n\n/**\n * List of available Deepgram models for text-to-speech and speech-to-text\n */\nexport const DEEPGRAM_MODELS = ['aura', 'whisper', 'base', 'enhanced', 'nova', 'nova-2', 'nova-3'] as const;\n\nexport type DeepgramModel = (typeof DEEPGRAM_MODELS)[number];\n","import { PassThrough } from 'stream';\n\nimport { createClient } from '@deepgram/sdk';\nimport { MastraVoice } from '@mastra/core/voice';\n\nimport { DEEPGRAM_VOICES } from './voices';\nimport type { DeepgramVoiceId, DeepgramModel } from './voices';\n\ninterface DeepgramVoiceConfig {\n name?: DeepgramModel;\n apiKey?: string;\n properties?: Record<string, any>;\n language?: string;\n}\n\nexport class DeepgramVoice extends MastraVoice {\n private speechClient?: ReturnType<typeof createClient>;\n private listeningClient?: ReturnType<typeof createClient>;\n\n constructor({\n speechModel,\n listeningModel,\n speaker,\n }: { speechModel?: DeepgramVoiceConfig; listeningModel?: DeepgramVoiceConfig; speaker?: DeepgramVoiceId } = {}) {\n const defaultApiKey = process.env.DEEPGRAM_API_KEY;\n\n const defaultSpeechModel = {\n name: 'aura',\n apiKey: defaultApiKey,\n };\n\n const defaultListeningModel = {\n name: 'nova',\n apiKey: defaultApiKey,\n };\n\n super({\n speechModel: {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n },\n listeningModel: {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n },\n speaker,\n });\n\n const speechApiKey = speechModel?.apiKey || defaultApiKey;\n const listeningApiKey = listeningModel?.apiKey || defaultApiKey;\n\n if (!speechApiKey && !listeningApiKey) {\n throw new Error('At least one of DEEPGRAM_API_KEY, speechModel.apiKey, or listeningModel.apiKey must be set');\n }\n\n if (speechApiKey) {\n this.speechClient = createClient(speechApiKey);\n }\n if (listeningApiKey) {\n this.listeningClient = createClient(listeningApiKey);\n }\n\n this.speaker = speaker || 'asteria-en';\n }\n\n async getSpeakers() {\n return DEEPGRAM_VOICES.map(voice => ({\n voiceId: voice,\n }));\n }\n\n async speak(\n input: string | NodeJS.ReadableStream,\n options?: {\n speaker?: string;\n [key: string]: any;\n },\n ): Promise<NodeJS.ReadableStream> {\n if (!this.speechClient) {\n throw new Error('Deepgram speech client not configured');\n }\n\n let text: string;\n if (typeof input !== 'string') {\n const chunks: Buffer[] = [];\n for await (const chunk of input) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n text = Buffer.concat(chunks).toString('utf-8');\n } else {\n text = input;\n }\n\n if (text.trim().length === 0) {\n throw new Error('Input text is empty');\n }\n\n if (!this.speechClient) {\n throw new Error('No speech client configured');\n }\n\n let model;\n if (options?.speaker) {\n model = this.speechModel?.name + '-' + options.speaker;\n } else if (this.speaker) {\n model = this.speechModel?.name + '-' + this.speaker;\n }\n\n const speakClient = this.speechClient.speak;\n const response = await speakClient.request(\n { text },\n {\n model,\n ...options,\n },\n );\n\n const webStream = await response.getStream();\n if (!webStream) {\n throw new Error('No stream returned from Deepgram');\n }\n\n const reader = webStream.getReader();\n const nodeStream = new PassThrough();\n\n // Add error handling for the stream processing\n (async () => {\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) {\n nodeStream.end();\n break;\n }\n nodeStream.write(value);\n }\n } catch (error) {\n nodeStream.destroy(error as Error);\n }\n })().catch(error => {\n nodeStream.destroy(error as Error);\n });\n\n return nodeStream;\n }\n\n /**\n * Checks if listening capabilities are enabled.\n *\n * @returns {Promise<{ enabled: boolean }>}\n */\n async getListener() {\n return { enabled: true };\n }\n\n async listen(\n audioStream: NodeJS.ReadableStream,\n options?: {\n [key: string]: any;\n },\n ): Promise<string> {\n if (!this.listeningClient) {\n throw new Error('Deepgram listening client not configured');\n }\n\n const chunks: Buffer[] = [];\n for await (const chunk of audioStream) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n const buffer = Buffer.concat(chunks);\n\n if (!this.listeningClient) {\n throw new Error('No listening client configured');\n }\n const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {\n model: this.listeningModel?.name,\n ...options,\n });\n\n if (error) {\n throw error;\n }\n\n const transcript = result.results?.channels?.[0]?.alternatives?.[0]?.transcript;\n if (!transcript) {\n throw new Error('No transcript found in Deepgram response');\n }\n\n return transcript;\n }\n}\n\nexport type { DeepgramVoiceConfig, DeepgramVoiceId, DeepgramModel };\n"]}
|
|
1
|
+
{"version":3,"sources":["../src/voices.ts","../src/index.ts"],"names":["MastraVoice","createClient","PassThrough"],"mappings":";;;;;;;;;AAKO,IAAM,eAAA,GAAkB;AAAA,EAC7B,YAAA;AAAA,EACA,SAAA;AAAA,EACA,WAAA;AAAA,EACA,WAAA;AAAA,EACA,SAAA;AAAA,EACA,UAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,WAAA;AAAA,EACA;AACF,CAAA;;;ACKO,IAAM,aAAA,GAAN,cAA4BA,iBAAA,CAAY;AAAA,EACrC,YAAA;AAAA,EACA,eAAA;AAAA,EACA,iBAAA;AAAA,EACA,oBAAA;AAAA,EACA,aAAA;AAAA,EAER,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA;AAAA,GACF,GAA4G,EAAC,EAAG;AAC9G,IAAA,MAAM,aAAA,GAAgB,QAAQ,GAAA,CAAI,gBAAA;AAElC,IAAA,MAAM,kBAAA,GAA+D;AAAA,MACnE,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,MAAM,qBAAA,GAAkE;AAAA,MACtE,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,QAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,OACpD;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,QACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,OAC1D;AAAA,MACA;AAAA,KACD,CAAA;AAED,IAAA,IAAA,CAAK,iBAAA,GAAoB;AAAA,MACvB,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,MAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,KACpD;AACA,IAAA,IAAA,CAAK,oBAAA,GAAuB;AAAA,MAC1B,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,MACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,KAC1D;AAEA,IAAA,MAAM,YAAA,GAAe,aAAa,MAAA,IAAU,aAAA;AAC5C,IAAA,MAAM,eAAA,GAAkB,gBAAgB,MAAA,IAAU,aAAA;AAElD,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,MAAM,4FAA4F,CAAA;AAAA,IAC9G;AAEA,IAAA,IAAI,YAAA,EAAc;AAChB,MAAA,IAAA,CAAK,YAAA,GAAeC,iBAAa,YAAY,CAAA;AAAA,IAC/C;AACA,IAAA,IAAI,eAAA,EAAiB;AACnB,MAAA,IAAA,CAAK,eAAA,GAAkBA,iBAAa,eAAe,CAAA;AAAA,IACrD;AAEA,IAAA,IAAA,CAAK,gBAAgB,OAAA,IAAW,YAAA;AAAA,EAClC;AAAA,EAEA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,eAAA,CAAgB,IAAI,CAAA,KAAA,MAAU;AAAA,MACnC,OAAA,EAAS;AAAA,KACX,CAAE,CAAA;AAAA,EACJ;AAAA,EAEA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EAIgC;AAChC,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,uCAAuC,CAAA;AAAA,IACzD;AAEA,IAAA,IAAI,IAAA;AACJ,IAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,MAAA,MAAM,SAAmB,EAAC;AAC1B,MAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,QAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,UAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,QAChC,CAAA,MAAO;AACL,UAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,QACnB;AAAA,MACF;AACA,MAAA,IAAA,GAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,IAC/C,CAAA,MAAO;AACL,MAAA,IAAA,GAAO,KAAA;AAAA,IACT;AAEA,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,MAAM,SAAA,GAAY,KAAK,iBAAA,EAAmB,IAAA;AAC1C,IAAA,MAAM,SAAA,GAAY,OAAA,EAAS,OAAA,IAAW,IAAA,CAAK,aAAA;AAE3C,IAAA,MAAM,SAAA,GACJ,SAAA,IAAa,SAAA,GACT,SAAA,CAAU,WAAW,CAAA,EAAG,SAAS,CAAA,CAAA,CAAG,CAAA,GAClC,YACA,CAAA,EAAG,SAAS,CAAA,CAAA,EAAI,SAAS,KAC3B,SAAA,IAAa,SAAA;AAEnB,IAAA,MAAM,WAAA,GAAc,KAAK,YAAA,CAAa,KAAA;AACtC,IAAA,MAAM,QAAA,GAAW,MAAM,WAAA,CAAY,OAAA;AAAA,MACjC,EAAE,IAAA,EAAK;AAAA,MACP;AAAA,QACE,KAAA,EAAO,SAAA;AAAA,QACP,GAAG,MAAA,CAAO,WAAA,CAAY,MAAA,CAAO,OAAA,CAAQ,WAAW,EAAE,CAAA,CAAE,MAAA,CAAO,CAAC,CAAC,CAAC,CAAA,KAAM,CAAA,KAAM,SAAS,CAAC;AAAA;AACtF,KACF;AAEA,IAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,SAAA,EAAU;AAC3C,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,IAAI,MAAM,kCAAkC,CAAA;AAAA,IACpD;AAEA,IAAA,MAAM,MAAA,GAAS,UAAU,SAAA,EAAU;AACnC,IAAA,MAAM,UAAA,GAAa,IAAIC,kBAAA,EAAY;AAGnC,IAAA,CAAC,YAAY;AACX,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AACR,YAAA,UAAA,CAAW,GAAA,EAAI;AACf,YAAA;AAAA,UACF;AACA,UAAA,UAAA,CAAW,MAAM,KAAK,CAAA;AAAA,QACxB;AAAA,MACF,SAAS,KAAA,EAAO;AACd,QAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,MACnC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,CAAA,KAAA,KAAS;AAClB,MAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,IACnC,CAAC,CAAA;AAED,IAAA,OAAO,UAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,SAAS,IAAA,EAAK;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAaA,MAAM,MAAA,CACJ,WAAA,EACA,OAAA,EAIc;AACd,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,WAAA,EAAa;AACrC,MAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,QAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,MAChC,CAAA,MAAO;AACL,QAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,MACnB;AAAA,IACF;AACA,IAAA,MAAM,MAAA,GAAS,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA;AACnC,IAAA,MAAM,EAAE,SAAS,qBAAA,EAAuB,CAAA,EAAG,GAAG,WAAA,EAAY,GAAI,WAAW,EAAC;AAC1E,IAAA,MAAM,EAAE,MAAA,EAAQ,KAAA,EAAM,GAAI,MAAM,KAAK,eAAA,CAAgB,MAAA,CAAO,WAAA,CAAY,cAAA,CAAe,MAAA,EAAQ;AAAA,MAC7F,GAAG,WAAA;AAAA,MACH,KAAA,EAAO,KAAK,oBAAA,EAAsB,IAAA;AAAA,MAClC;AAAA,KACD,CAAA;AACD,IAAA,IAAI,KAAA,EAAO;AACT,MAAA,MAAM,KAAA;AAAA,IACR;AAEA,IAAA,MAAM,OAAA,GAAU,MAAA,CAAO,OAAA,EAAS,QAAA,GAAW,CAAC,CAAA;AAC5C,IAAA,MAAM,GAAA,GAKU,OAAA,EAAS,YAAA,GAAe,CAAC,CAAA;AAEzC,IAAA,IAAI,CAAC,GAAA,EAAK;AACR,MAAA,OAAO;AAAA,QACL,UAAA,EAAY,EAAA;AAAA,QACZ,OAAO,EAAC;AAAA,QACR,GAAA,EAAK;AAAA,OACP;AAAA,IACF;AAEA,IAAA,MAAM,QAAA,GAAgB;AAAA,MACpB,YAAY,GAAA,CAAI,UAAA;AAAA,MAChB,OAAO,GAAA,CAAI,KAAA;AAAA,MACX,GAAA,EAAK;AAAA,KACP;AAEA,IAAA,IAAI,OAAA,IAAW,IAAI,KAAA,EAAO;AACxB,MAAA,QAAA,CAAS,eAAA,GAAkB,GAAA,CAAI,KAAA,CAAM,GAAA,CAAI,CAAC,CAAA,MAAqB;AAAA,QAC7D,MAAM,CAAA,CAAE,IAAA;AAAA,QACR,SAAS,CAAA,CAAE,OAAA;AAAA,QACX,OAAO,CAAA,CAAE,KAAA;AAAA,QACT,KAAK,CAAA,CAAE;AAAA,OACT,CAAE,CAAA;AAAA,IACJ;AAEA,IAAA,OAAO,QAAA;AAAA,EACT;AACF","file":"index.cjs","sourcesContent":["/**\n * List of available Deepgram voice models for text-to-speech\n * Each voice is designed for specific use cases and languages\n * Format: {name}-{language} (e.g. asteria-en)\n */\nexport const DEEPGRAM_VOICES = [\n 'asteria-en',\n 'luna-en',\n 'stella-en',\n 'athena-en',\n 'hera-en',\n 'orion-en',\n 'arcas-en',\n 'perseus-en',\n 'angus-en',\n 'orpheus-en',\n 'helios-en',\n 'zeus-en',\n] as const;\n\nexport type DeepgramVoiceId = (typeof DEEPGRAM_VOICES)[number];\n\n/**\n * List of available Deepgram models for text-to-speech and speech-to-text\n */\nexport const DEEPGRAM_MODELS = ['aura', 'whisper', 'base', 'enhanced', 'nova', 'nova-2', 'nova-3'] as const;\n\nexport type DeepgramModel = (typeof DEEPGRAM_MODELS)[number];\n","import { PassThrough } from 'stream';\n\nimport { createClient } from '@deepgram/sdk';\nimport { MastraVoice } from '@mastra/core/voice';\n\nimport { DEEPGRAM_VOICES } from './voices';\nimport type { DeepgramVoiceId, DeepgramModel } from './voices';\n\ninterface DeepgramVoiceConfig {\n name?: DeepgramModel;\n apiKey?: string;\n properties?: Record<string, any>;\n language?: string;\n}\n\ninterface DeepgramWord {\n word: string;\n start?: number;\n end?: number;\n confidence?: number;\n speaker?: number;\n}\n\nexport class DeepgramVoice extends MastraVoice {\n private speechClient?: ReturnType<typeof createClient>;\n private listeningClient?: ReturnType<typeof createClient>;\n private storedSpeechModel?: { name: DeepgramModel; apiKey?: string };\n private storedListeningModel?: { name: DeepgramModel; apiKey?: string };\n private storedSpeaker?: DeepgramVoiceId;\n\n constructor({\n speechModel,\n listeningModel,\n speaker,\n }: { speechModel?: DeepgramVoiceConfig; listeningModel?: DeepgramVoiceConfig; speaker?: DeepgramVoiceId } = {}) {\n const defaultApiKey = process.env.DEEPGRAM_API_KEY;\n\n const defaultSpeechModel: { name: DeepgramModel; apiKey?: string } = {\n name: 'aura',\n apiKey: defaultApiKey,\n };\n\n const defaultListeningModel: { name: DeepgramModel; apiKey?: string } = {\n name: 'nova',\n apiKey: defaultApiKey,\n };\n\n super({\n speechModel: {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n },\n listeningModel: {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n },\n speaker,\n });\n\n this.storedSpeechModel = {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n };\n this.storedListeningModel = {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n };\n\n const speechApiKey = speechModel?.apiKey || defaultApiKey;\n const listeningApiKey = listeningModel?.apiKey || defaultApiKey;\n\n if (!speechApiKey && !listeningApiKey) {\n throw new Error('At least one of DEEPGRAM_API_KEY, speechModel.apiKey, or listeningModel.apiKey must be set');\n }\n\n if (speechApiKey) {\n this.speechClient = createClient(speechApiKey);\n }\n if (listeningApiKey) {\n this.listeningClient = createClient(listeningApiKey);\n }\n\n this.storedSpeaker = speaker || 'asteria-en';\n }\n\n async getSpeakers() {\n return DEEPGRAM_VOICES.map(voice => ({\n voiceId: voice,\n }));\n }\n\n async speak(\n input: string | NodeJS.ReadableStream,\n options?: {\n speaker?: string;\n [key: string]: any;\n },\n ): Promise<NodeJS.ReadableStream> {\n if (!this.speechClient) {\n throw new Error('Deepgram speech client not configured');\n }\n\n let text: string;\n if (typeof input !== 'string') {\n const chunks: Buffer[] = [];\n for await (const chunk of input) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n text = Buffer.concat(chunks).toString('utf-8');\n } else {\n text = input;\n }\n\n if (text.trim().length === 0) {\n throw new Error('Input text is empty');\n }\n\n const baseModel = this.storedSpeechModel?.name;\n const speakerId = options?.speaker || this.storedSpeaker;\n\n const modelName =\n baseModel && speakerId\n ? speakerId.startsWith(`${baseModel}-`)\n ? speakerId\n : `${baseModel}-${speakerId}`\n : baseModel || speakerId;\n\n const speakClient = this.speechClient.speak;\n const response = await speakClient.request(\n { text },\n {\n model: modelName,\n ...Object.fromEntries(Object.entries(options ?? {}).filter(([k]) => k !== 'speaker')),\n },\n );\n\n const webStream = await response.getStream();\n if (!webStream) {\n throw new Error('No stream returned from Deepgram');\n }\n\n const reader = webStream.getReader();\n const nodeStream = new PassThrough();\n\n // Add error handling for the stream processing\n (async () => {\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) {\n nodeStream.end();\n break;\n }\n nodeStream.write(value);\n }\n } catch (error) {\n nodeStream.destroy(error as Error);\n }\n })().catch(error => {\n nodeStream.destroy(error as Error);\n });\n\n return nodeStream;\n }\n\n /**\n * Checks if listening capabilities are enabled.\n *\n * @returns {Promise<{ enabled: boolean }>}\n */\n async getListener() {\n return { enabled: true };\n }\n\n /**\n * Transcribes audio with optional speaker diarization.\n *\n * @param audioStream - Audio input stream\n * @param options - Transcription options (diarize, language, etc.)\n * @returns Promise resolving to:\n * - transcript: Full transcript string\n * - words: Array of word objects with timing and confidence\n * - raw: Complete Deepgram API response\n * - speakerSegments: (when diarize=true) Array of {word, speaker, start, end}\n */\n async listen(\n audioStream: NodeJS.ReadableStream,\n options?: {\n diarize?: boolean;\n [key: string]: any;\n },\n ): Promise<any> {\n if (!this.listeningClient) {\n throw new Error('Deepgram listening client not configured');\n }\n\n const chunks: Buffer[] = [];\n for await (const chunk of audioStream) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n const buffer = Buffer.concat(chunks);\n const { diarize, diarize_speaker_count: _, ...restOptions } = options ?? {};\n const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {\n ...restOptions,\n model: this.storedListeningModel?.name,\n diarize,\n });\n if (error) {\n throw error;\n }\n\n const channel = result.results?.channels?.[0];\n const alt:\n | {\n transcript?: string;\n words?: DeepgramWord[];\n }\n | undefined = channel?.alternatives?.[0];\n\n if (!alt) {\n return {\n transcript: '',\n words: [],\n raw: result,\n };\n }\n\n const response: any = {\n transcript: alt.transcript,\n words: alt.words,\n raw: result,\n };\n\n if (diarize && alt.words) {\n response.speakerSegments = alt.words.map((w: DeepgramWord) => ({\n word: w.word,\n speaker: w.speaker,\n start: w.start,\n end: w.end,\n }));\n }\n\n return response;\n }\n}\n\nexport type { DeepgramVoiceConfig, DeepgramVoiceId, DeepgramModel };\n"]}
|
package/dist/index.d.ts
CHANGED
|
@@ -9,6 +9,9 @@ interface DeepgramVoiceConfig {
|
|
|
9
9
|
export declare class DeepgramVoice extends MastraVoice {
|
|
10
10
|
private speechClient?;
|
|
11
11
|
private listeningClient?;
|
|
12
|
+
private storedSpeechModel?;
|
|
13
|
+
private storedListeningModel?;
|
|
14
|
+
private storedSpeaker?;
|
|
12
15
|
constructor({ speechModel, listeningModel, speaker, }?: {
|
|
13
16
|
speechModel?: DeepgramVoiceConfig;
|
|
14
17
|
listeningModel?: DeepgramVoiceConfig;
|
|
@@ -29,9 +32,21 @@ export declare class DeepgramVoice extends MastraVoice {
|
|
|
29
32
|
getListener(): Promise<{
|
|
30
33
|
enabled: boolean;
|
|
31
34
|
}>;
|
|
35
|
+
/**
|
|
36
|
+
* Transcribes audio with optional speaker diarization.
|
|
37
|
+
*
|
|
38
|
+
* @param audioStream - Audio input stream
|
|
39
|
+
* @param options - Transcription options (diarize, language, etc.)
|
|
40
|
+
* @returns Promise resolving to:
|
|
41
|
+
* - transcript: Full transcript string
|
|
42
|
+
* - words: Array of word objects with timing and confidence
|
|
43
|
+
* - raw: Complete Deepgram API response
|
|
44
|
+
* - speakerSegments: (when diarize=true) Array of {word, speaker, start, end}
|
|
45
|
+
*/
|
|
32
46
|
listen(audioStream: NodeJS.ReadableStream, options?: {
|
|
47
|
+
diarize?: boolean;
|
|
33
48
|
[key: string]: any;
|
|
34
|
-
}): Promise<
|
|
49
|
+
}): Promise<any>;
|
|
35
50
|
}
|
|
36
51
|
export type { DeepgramVoiceConfig, DeepgramVoiceId, DeepgramModel };
|
|
37
52
|
//# sourceMappingURL=index.d.ts.map
|
package/dist/index.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAGjD,OAAO,KAAK,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAE/D,UAAU,mBAAmB;IAC3B,IAAI,CAAC,EAAE,aAAa,CAAC;IACrB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;IACjC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAGjD,OAAO,KAAK,EAAE,eAAe,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AAE/D,UAAU,mBAAmB;IAC3B,IAAI,CAAC,EAAE,aAAa,CAAC;IACrB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;IACjC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB;AAUD,qBAAa,aAAc,SAAQ,WAAW;IAC5C,OAAO,CAAC,YAAY,CAAC,CAAkC;IACvD,OAAO,CAAC,eAAe,CAAC,CAAkC;IAC1D,OAAO,CAAC,iBAAiB,CAAC,CAA2C;IACrE,OAAO,CAAC,oBAAoB,CAAC,CAA2C;IACxE,OAAO,CAAC,aAAa,CAAC,CAAkB;gBAE5B,EACV,WAAW,EACX,cAAc,EACd,OAAO,GACR,GAAE;QAAE,WAAW,CAAC,EAAE,mBAAmB,CAAC;QAAC,cAAc,CAAC,EAAE,mBAAmB,CAAC;QAAC,OAAO,CAAC,EAAE,eAAe,CAAA;KAAO;IAmDxG,WAAW;;;IAMX,KAAK,CACT,KAAK,EAAE,MAAM,GAAG,MAAM,CAAC,cAAc,EACrC,OAAO,CAAC,EAAE;QACR,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;KACpB,GACA,OAAO,CAAC,MAAM,CAAC,cAAc,CAAC;IAwEjC;;;;OAIG;IACG,WAAW;;;IAIjB;;;;;;;;;;OAUG;IACG,MAAM,CACV,WAAW,EAAE,MAAM,CAAC,cAAc,EAClC,OAAO,CAAC,EAAE;QACR,OAAO,CAAC,EAAE,OAAO,CAAC;QAClB,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;KACpB,GACA,OAAO,CAAC,GAAG,CAAC;CAyDhB;AAED,YAAY,EAAE,mBAAmB,EAAE,eAAe,EAAE,aAAa,EAAE,CAAC"}
|
package/dist/index.js
CHANGED
|
@@ -24,6 +24,9 @@ var DEEPGRAM_VOICES = [
|
|
|
24
24
|
var DeepgramVoice = class extends MastraVoice {
|
|
25
25
|
speechClient;
|
|
26
26
|
listeningClient;
|
|
27
|
+
storedSpeechModel;
|
|
28
|
+
storedListeningModel;
|
|
29
|
+
storedSpeaker;
|
|
27
30
|
constructor({
|
|
28
31
|
speechModel,
|
|
29
32
|
listeningModel,
|
|
@@ -49,6 +52,14 @@ var DeepgramVoice = class extends MastraVoice {
|
|
|
49
52
|
},
|
|
50
53
|
speaker
|
|
51
54
|
});
|
|
55
|
+
this.storedSpeechModel = {
|
|
56
|
+
name: speechModel?.name ?? defaultSpeechModel.name,
|
|
57
|
+
apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey
|
|
58
|
+
};
|
|
59
|
+
this.storedListeningModel = {
|
|
60
|
+
name: listeningModel?.name ?? defaultListeningModel.name,
|
|
61
|
+
apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey
|
|
62
|
+
};
|
|
52
63
|
const speechApiKey = speechModel?.apiKey || defaultApiKey;
|
|
53
64
|
const listeningApiKey = listeningModel?.apiKey || defaultApiKey;
|
|
54
65
|
if (!speechApiKey && !listeningApiKey) {
|
|
@@ -60,7 +71,7 @@ var DeepgramVoice = class extends MastraVoice {
|
|
|
60
71
|
if (listeningApiKey) {
|
|
61
72
|
this.listeningClient = createClient(listeningApiKey);
|
|
62
73
|
}
|
|
63
|
-
this.
|
|
74
|
+
this.storedSpeaker = speaker || "asteria-en";
|
|
64
75
|
}
|
|
65
76
|
async getSpeakers() {
|
|
66
77
|
return DEEPGRAM_VOICES.map((voice) => ({
|
|
@@ -88,21 +99,15 @@ var DeepgramVoice = class extends MastraVoice {
|
|
|
88
99
|
if (text.trim().length === 0) {
|
|
89
100
|
throw new Error("Input text is empty");
|
|
90
101
|
}
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
}
|
|
94
|
-
let model;
|
|
95
|
-
if (options?.speaker) {
|
|
96
|
-
model = this.speechModel?.name + "-" + options.speaker;
|
|
97
|
-
} else if (this.speaker) {
|
|
98
|
-
model = this.speechModel?.name + "-" + this.speaker;
|
|
99
|
-
}
|
|
102
|
+
const baseModel = this.storedSpeechModel?.name;
|
|
103
|
+
const speakerId = options?.speaker || this.storedSpeaker;
|
|
104
|
+
const modelName = baseModel && speakerId ? speakerId.startsWith(`${baseModel}-`) ? speakerId : `${baseModel}-${speakerId}` : baseModel || speakerId;
|
|
100
105
|
const speakClient = this.speechClient.speak;
|
|
101
106
|
const response = await speakClient.request(
|
|
102
107
|
{ text },
|
|
103
108
|
{
|
|
104
|
-
model,
|
|
105
|
-
...options
|
|
109
|
+
model: modelName,
|
|
110
|
+
...Object.fromEntries(Object.entries(options ?? {}).filter(([k]) => k !== "speaker"))
|
|
106
111
|
}
|
|
107
112
|
);
|
|
108
113
|
const webStream = await response.getStream();
|
|
@@ -137,6 +142,17 @@ var DeepgramVoice = class extends MastraVoice {
|
|
|
137
142
|
async getListener() {
|
|
138
143
|
return { enabled: true };
|
|
139
144
|
}
|
|
145
|
+
/**
|
|
146
|
+
* Transcribes audio with optional speaker diarization.
|
|
147
|
+
*
|
|
148
|
+
* @param audioStream - Audio input stream
|
|
149
|
+
* @param options - Transcription options (diarize, language, etc.)
|
|
150
|
+
* @returns Promise resolving to:
|
|
151
|
+
* - transcript: Full transcript string
|
|
152
|
+
* - words: Array of word objects with timing and confidence
|
|
153
|
+
* - raw: Complete Deepgram API response
|
|
154
|
+
* - speakerSegments: (when diarize=true) Array of {word, speaker, start, end}
|
|
155
|
+
*/
|
|
140
156
|
async listen(audioStream, options) {
|
|
141
157
|
if (!this.listeningClient) {
|
|
142
158
|
throw new Error("Deepgram listening client not configured");
|
|
@@ -150,21 +166,38 @@ var DeepgramVoice = class extends MastraVoice {
|
|
|
150
166
|
}
|
|
151
167
|
}
|
|
152
168
|
const buffer = Buffer.concat(chunks);
|
|
153
|
-
|
|
154
|
-
throw new Error("No listening client configured");
|
|
155
|
-
}
|
|
169
|
+
const { diarize, diarize_speaker_count: _, ...restOptions } = options ?? {};
|
|
156
170
|
const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {
|
|
157
|
-
|
|
158
|
-
|
|
171
|
+
...restOptions,
|
|
172
|
+
model: this.storedListeningModel?.name,
|
|
173
|
+
diarize
|
|
159
174
|
});
|
|
160
175
|
if (error) {
|
|
161
176
|
throw error;
|
|
162
177
|
}
|
|
163
|
-
const
|
|
164
|
-
|
|
165
|
-
|
|
178
|
+
const channel = result.results?.channels?.[0];
|
|
179
|
+
const alt = channel?.alternatives?.[0];
|
|
180
|
+
if (!alt) {
|
|
181
|
+
return {
|
|
182
|
+
transcript: "",
|
|
183
|
+
words: [],
|
|
184
|
+
raw: result
|
|
185
|
+
};
|
|
186
|
+
}
|
|
187
|
+
const response = {
|
|
188
|
+
transcript: alt.transcript,
|
|
189
|
+
words: alt.words,
|
|
190
|
+
raw: result
|
|
191
|
+
};
|
|
192
|
+
if (diarize && alt.words) {
|
|
193
|
+
response.speakerSegments = alt.words.map((w) => ({
|
|
194
|
+
word: w.word,
|
|
195
|
+
speaker: w.speaker,
|
|
196
|
+
start: w.start,
|
|
197
|
+
end: w.end
|
|
198
|
+
}));
|
|
166
199
|
}
|
|
167
|
-
return
|
|
200
|
+
return response;
|
|
168
201
|
}
|
|
169
202
|
};
|
|
170
203
|
|
package/dist/index.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/voices.ts","../src/index.ts"],"names":[],"mappings":";;;;;;;AAKO,IAAM,eAAA,GAAkB;AAAA,EAC7B,YAAA;AAAA,EACA,SAAA;AAAA,EACA,WAAA;AAAA,EACA,WAAA;AAAA,EACA,SAAA;AAAA,EACA,UAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,WAAA;AAAA,EACA;AACF,CAAA;;;ACHO,IAAM,aAAA,GAAN,cAA4B,WAAA,CAAY;AAAA,EACrC,YAAA;AAAA,EACA,eAAA;AAAA,EAER,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA;AAAA,GACF,GAA4G,EAAC,EAAG;AAC9G,IAAA,MAAM,aAAA,GAAgB,QAAQ,GAAA,CAAI,gBAAA;AAElC,IAAA,MAAM,kBAAA,GAAqB;AAAA,MACzB,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,MAAM,qBAAA,GAAwB;AAAA,MAC5B,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,QAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,OACpD;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,QACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,OAC1D;AAAA,MACA;AAAA,KACD,CAAA;AAED,IAAA,MAAM,YAAA,GAAe,aAAa,MAAA,IAAU,aAAA;AAC5C,IAAA,MAAM,eAAA,GAAkB,gBAAgB,MAAA,IAAU,aAAA;AAElD,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,MAAM,4FAA4F,CAAA;AAAA,IAC9G;AAEA,IAAA,IAAI,YAAA,EAAc;AAChB,MAAA,IAAA,CAAK,YAAA,GAAe,aAAa,YAAY,CAAA;AAAA,IAC/C;AACA,IAAA,IAAI,eAAA,EAAiB;AACnB,MAAA,IAAA,CAAK,eAAA,GAAkB,aAAa,eAAe,CAAA;AAAA,IACrD;AAEA,IAAA,IAAA,CAAK,UAAU,OAAA,IAAW,YAAA;AAAA,EAC5B;AAAA,EAEA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,eAAA,CAAgB,IAAI,CAAA,KAAA,MAAU;AAAA,MACnC,OAAA,EAAS;AAAA,KACX,CAAE,CAAA;AAAA,EACJ;AAAA,EAEA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EAIgC;AAChC,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,uCAAuC,CAAA;AAAA,IACzD;AAEA,IAAA,IAAI,IAAA;AACJ,IAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,MAAA,MAAM,SAAmB,EAAC;AAC1B,MAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,QAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,UAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,QAChC,CAAA,MAAO;AACL,UAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,QACnB;AAAA,MACF;AACA,MAAA,IAAA,GAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,IAC/C,CAAA,MAAO;AACL,MAAA,IAAA,GAAO,KAAA;AAAA,IACT;AAEA,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,6BAA6B,CAAA;AAAA,IAC/C;AAEA,IAAA,IAAI,KAAA;AACJ,IAAA,IAAI,SAAS,OAAA,EAAS;AACpB,MAAA,KAAA,GAAQ,IAAA,CAAK,WAAA,EAAa,IAAA,GAAO,GAAA,GAAM,OAAA,CAAQ,OAAA;AAAA,IACjD,CAAA,MAAA,IAAW,KAAK,OAAA,EAAS;AACvB,MAAA,KAAA,GAAQ,IAAA,CAAK,WAAA,EAAa,IAAA,GAAO,GAAA,GAAM,IAAA,CAAK,OAAA;AAAA,IAC9C;AAEA,IAAA,MAAM,WAAA,GAAc,KAAK,YAAA,CAAa,KAAA;AACtC,IAAA,MAAM,QAAA,GAAW,MAAM,WAAA,CAAY,OAAA;AAAA,MACjC,EAAE,IAAA,EAAK;AAAA,MACP;AAAA,QACE,KAAA;AAAA,QACA,GAAG;AAAA;AACL,KACF;AAEA,IAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,SAAA,EAAU;AAC3C,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,IAAI,MAAM,kCAAkC,CAAA;AAAA,IACpD;AAEA,IAAA,MAAM,MAAA,GAAS,UAAU,SAAA,EAAU;AACnC,IAAA,MAAM,UAAA,GAAa,IAAI,WAAA,EAAY;AAGnC,IAAA,CAAC,YAAY;AACX,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AACR,YAAA,UAAA,CAAW,GAAA,EAAI;AACf,YAAA;AAAA,UACF;AACA,UAAA,UAAA,CAAW,MAAM,KAAK,CAAA;AAAA,QACxB;AAAA,MACF,SAAS,KAAA,EAAO;AACd,QAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,MACnC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,CAAA,KAAA,KAAS;AAClB,MAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,IACnC,CAAC,CAAA;AAED,IAAA,OAAO,UAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,SAAS,IAAA,EAAK;AAAA,EACzB;AAAA,EAEA,MAAM,MAAA,CACJ,WAAA,EACA,OAAA,EAGiB;AACjB,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,WAAA,EAAa;AACrC,MAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,QAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,MAChC,CAAA,MAAO;AACL,QAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,MACnB;AAAA,IACF;AACA,IAAA,MAAM,MAAA,GAAS,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA;AAEnC,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,gCAAgC,CAAA;AAAA,IAClD;AACA,IAAA,MAAM,EAAE,MAAA,EAAQ,KAAA,EAAM,GAAI,MAAM,KAAK,eAAA,CAAgB,MAAA,CAAO,WAAA,CAAY,cAAA,CAAe,MAAA,EAAQ;AAAA,MAC7F,KAAA,EAAO,KAAK,cAAA,EAAgB,IAAA;AAAA,MAC5B,GAAG;AAAA,KACJ,CAAA;AAED,IAAA,IAAI,KAAA,EAAO;AACT,MAAA,MAAM,KAAA;AAAA,IACR;AAEA,IAAA,MAAM,UAAA,GAAa,OAAO,OAAA,EAAS,QAAA,GAAW,CAAC,CAAA,EAAG,YAAA,GAAe,CAAC,CAAA,EAAG,UAAA;AACrE,IAAA,IAAI,CAAC,UAAA,EAAY;AACf,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,OAAO,UAAA;AAAA,EACT;AACF","file":"index.js","sourcesContent":["/**\n * List of available Deepgram voice models for text-to-speech\n * Each voice is designed for specific use cases and languages\n * Format: {name}-{language} (e.g. asteria-en)\n */\nexport const DEEPGRAM_VOICES = [\n 'asteria-en',\n 'luna-en',\n 'stella-en',\n 'athena-en',\n 'hera-en',\n 'orion-en',\n 'arcas-en',\n 'perseus-en',\n 'angus-en',\n 'orpheus-en',\n 'helios-en',\n 'zeus-en',\n] as const;\n\nexport type DeepgramVoiceId = (typeof DEEPGRAM_VOICES)[number];\n\n/**\n * List of available Deepgram models for text-to-speech and speech-to-text\n */\nexport const DEEPGRAM_MODELS = ['aura', 'whisper', 'base', 'enhanced', 'nova', 'nova-2', 'nova-3'] as const;\n\nexport type DeepgramModel = (typeof DEEPGRAM_MODELS)[number];\n","import { PassThrough } from 'stream';\n\nimport { createClient } from '@deepgram/sdk';\nimport { MastraVoice } from '@mastra/core/voice';\n\nimport { DEEPGRAM_VOICES } from './voices';\nimport type { DeepgramVoiceId, DeepgramModel } from './voices';\n\ninterface DeepgramVoiceConfig {\n name?: DeepgramModel;\n apiKey?: string;\n properties?: Record<string, any>;\n language?: string;\n}\n\nexport class DeepgramVoice extends MastraVoice {\n private speechClient?: ReturnType<typeof createClient>;\n private listeningClient?: ReturnType<typeof createClient>;\n\n constructor({\n speechModel,\n listeningModel,\n speaker,\n }: { speechModel?: DeepgramVoiceConfig; listeningModel?: DeepgramVoiceConfig; speaker?: DeepgramVoiceId } = {}) {\n const defaultApiKey = process.env.DEEPGRAM_API_KEY;\n\n const defaultSpeechModel = {\n name: 'aura',\n apiKey: defaultApiKey,\n };\n\n const defaultListeningModel = {\n name: 'nova',\n apiKey: defaultApiKey,\n };\n\n super({\n speechModel: {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n },\n listeningModel: {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n },\n speaker,\n });\n\n const speechApiKey = speechModel?.apiKey || defaultApiKey;\n const listeningApiKey = listeningModel?.apiKey || defaultApiKey;\n\n if (!speechApiKey && !listeningApiKey) {\n throw new Error('At least one of DEEPGRAM_API_KEY, speechModel.apiKey, or listeningModel.apiKey must be set');\n }\n\n if (speechApiKey) {\n this.speechClient = createClient(speechApiKey);\n }\n if (listeningApiKey) {\n this.listeningClient = createClient(listeningApiKey);\n }\n\n this.speaker = speaker || 'asteria-en';\n }\n\n async getSpeakers() {\n return DEEPGRAM_VOICES.map(voice => ({\n voiceId: voice,\n }));\n }\n\n async speak(\n input: string | NodeJS.ReadableStream,\n options?: {\n speaker?: string;\n [key: string]: any;\n },\n ): Promise<NodeJS.ReadableStream> {\n if (!this.speechClient) {\n throw new Error('Deepgram speech client not configured');\n }\n\n let text: string;\n if (typeof input !== 'string') {\n const chunks: Buffer[] = [];\n for await (const chunk of input) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n text = Buffer.concat(chunks).toString('utf-8');\n } else {\n text = input;\n }\n\n if (text.trim().length === 0) {\n throw new Error('Input text is empty');\n }\n\n if (!this.speechClient) {\n throw new Error('No speech client configured');\n }\n\n let model;\n if (options?.speaker) {\n model = this.speechModel?.name + '-' + options.speaker;\n } else if (this.speaker) {\n model = this.speechModel?.name + '-' + this.speaker;\n }\n\n const speakClient = this.speechClient.speak;\n const response = await speakClient.request(\n { text },\n {\n model,\n ...options,\n },\n );\n\n const webStream = await response.getStream();\n if (!webStream) {\n throw new Error('No stream returned from Deepgram');\n }\n\n const reader = webStream.getReader();\n const nodeStream = new PassThrough();\n\n // Add error handling for the stream processing\n (async () => {\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) {\n nodeStream.end();\n break;\n }\n nodeStream.write(value);\n }\n } catch (error) {\n nodeStream.destroy(error as Error);\n }\n })().catch(error => {\n nodeStream.destroy(error as Error);\n });\n\n return nodeStream;\n }\n\n /**\n * Checks if listening capabilities are enabled.\n *\n * @returns {Promise<{ enabled: boolean }>}\n */\n async getListener() {\n return { enabled: true };\n }\n\n async listen(\n audioStream: NodeJS.ReadableStream,\n options?: {\n [key: string]: any;\n },\n ): Promise<string> {\n if (!this.listeningClient) {\n throw new Error('Deepgram listening client not configured');\n }\n\n const chunks: Buffer[] = [];\n for await (const chunk of audioStream) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n const buffer = Buffer.concat(chunks);\n\n if (!this.listeningClient) {\n throw new Error('No listening client configured');\n }\n const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {\n model: this.listeningModel?.name,\n ...options,\n });\n\n if (error) {\n throw error;\n }\n\n const transcript = result.results?.channels?.[0]?.alternatives?.[0]?.transcript;\n if (!transcript) {\n throw new Error('No transcript found in Deepgram response');\n }\n\n return transcript;\n }\n}\n\nexport type { DeepgramVoiceConfig, DeepgramVoiceId, DeepgramModel };\n"]}
|
|
1
|
+
{"version":3,"sources":["../src/voices.ts","../src/index.ts"],"names":[],"mappings":";;;;;;;AAKO,IAAM,eAAA,GAAkB;AAAA,EAC7B,YAAA;AAAA,EACA,SAAA;AAAA,EACA,WAAA;AAAA,EACA,WAAA;AAAA,EACA,SAAA;AAAA,EACA,UAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,UAAA;AAAA,EACA,YAAA;AAAA,EACA,WAAA;AAAA,EACA;AACF,CAAA;;;ACKO,IAAM,aAAA,GAAN,cAA4B,WAAA,CAAY;AAAA,EACrC,YAAA;AAAA,EACA,eAAA;AAAA,EACA,iBAAA;AAAA,EACA,oBAAA;AAAA,EACA,aAAA;AAAA,EAER,WAAA,CAAY;AAAA,IACV,WAAA;AAAA,IACA,cAAA;AAAA,IACA;AAAA,GACF,GAA4G,EAAC,EAAG;AAC9G,IAAA,MAAM,aAAA,GAAgB,QAAQ,GAAA,CAAI,gBAAA;AAElC,IAAA,MAAM,kBAAA,GAA+D;AAAA,MACnE,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,MAAM,qBAAA,GAAkE;AAAA,MACtE,IAAA,EAAM,MAAA;AAAA,MACN,MAAA,EAAQ;AAAA,KACV;AAEA,IAAA,KAAA,CAAM;AAAA,MACJ,WAAA,EAAa;AAAA,QACX,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,QAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,OACpD;AAAA,MACA,cAAA,EAAgB;AAAA,QACd,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,QACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,OAC1D;AAAA,MACA;AAAA,KACD,CAAA;AAED,IAAA,IAAA,CAAK,iBAAA,GAAoB;AAAA,MACvB,IAAA,EAAM,WAAA,EAAa,IAAA,IAAQ,kBAAA,CAAmB,IAAA;AAAA,MAC9C,MAAA,EAAQ,WAAA,EAAa,MAAA,IAAU,kBAAA,CAAmB;AAAA,KACpD;AACA,IAAA,IAAA,CAAK,oBAAA,GAAuB;AAAA,MAC1B,IAAA,EAAM,cAAA,EAAgB,IAAA,IAAQ,qBAAA,CAAsB,IAAA;AAAA,MACpD,MAAA,EAAQ,cAAA,EAAgB,MAAA,IAAU,qBAAA,CAAsB;AAAA,KAC1D;AAEA,IAAA,MAAM,YAAA,GAAe,aAAa,MAAA,IAAU,aAAA;AAC5C,IAAA,MAAM,eAAA,GAAkB,gBAAgB,MAAA,IAAU,aAAA;AAElD,IAAA,IAAI,CAAC,YAAA,IAAgB,CAAC,eAAA,EAAiB;AACrC,MAAA,MAAM,IAAI,MAAM,4FAA4F,CAAA;AAAA,IAC9G;AAEA,IAAA,IAAI,YAAA,EAAc;AAChB,MAAA,IAAA,CAAK,YAAA,GAAe,aAAa,YAAY,CAAA;AAAA,IAC/C;AACA,IAAA,IAAI,eAAA,EAAiB;AACnB,MAAA,IAAA,CAAK,eAAA,GAAkB,aAAa,eAAe,CAAA;AAAA,IACrD;AAEA,IAAA,IAAA,CAAK,gBAAgB,OAAA,IAAW,YAAA;AAAA,EAClC;AAAA,EAEA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,eAAA,CAAgB,IAAI,CAAA,KAAA,MAAU;AAAA,MACnC,OAAA,EAAS;AAAA,KACX,CAAE,CAAA;AAAA,EACJ;AAAA,EAEA,MAAM,KAAA,CACJ,KAAA,EACA,OAAA,EAIgC;AAChC,IAAA,IAAI,CAAC,KAAK,YAAA,EAAc;AACtB,MAAA,MAAM,IAAI,MAAM,uCAAuC,CAAA;AAAA,IACzD;AAEA,IAAA,IAAI,IAAA;AACJ,IAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,MAAA,MAAM,SAAmB,EAAC;AAC1B,MAAA,WAAA,MAAiB,SAAS,KAAA,EAAO;AAC/B,QAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,UAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,QAChC,CAAA,MAAO;AACL,UAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,QACnB;AAAA,MACF;AACA,MAAA,IAAA,GAAO,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA,CAAE,SAAS,OAAO,CAAA;AAAA,IAC/C,CAAA,MAAO;AACL,MAAA,IAAA,GAAO,KAAA;AAAA,IACT;AAEA,IAAA,IAAI,IAAA,CAAK,IAAA,EAAK,CAAE,MAAA,KAAW,CAAA,EAAG;AAC5B,MAAA,MAAM,IAAI,MAAM,qBAAqB,CAAA;AAAA,IACvC;AAEA,IAAA,MAAM,SAAA,GAAY,KAAK,iBAAA,EAAmB,IAAA;AAC1C,IAAA,MAAM,SAAA,GAAY,OAAA,EAAS,OAAA,IAAW,IAAA,CAAK,aAAA;AAE3C,IAAA,MAAM,SAAA,GACJ,SAAA,IAAa,SAAA,GACT,SAAA,CAAU,WAAW,CAAA,EAAG,SAAS,CAAA,CAAA,CAAG,CAAA,GAClC,YACA,CAAA,EAAG,SAAS,CAAA,CAAA,EAAI,SAAS,KAC3B,SAAA,IAAa,SAAA;AAEnB,IAAA,MAAM,WAAA,GAAc,KAAK,YAAA,CAAa,KAAA;AACtC,IAAA,MAAM,QAAA,GAAW,MAAM,WAAA,CAAY,OAAA;AAAA,MACjC,EAAE,IAAA,EAAK;AAAA,MACP;AAAA,QACE,KAAA,EAAO,SAAA;AAAA,QACP,GAAG,MAAA,CAAO,WAAA,CAAY,MAAA,CAAO,OAAA,CAAQ,WAAW,EAAE,CAAA,CAAE,MAAA,CAAO,CAAC,CAAC,CAAC,CAAA,KAAM,CAAA,KAAM,SAAS,CAAC;AAAA;AACtF,KACF;AAEA,IAAA,MAAM,SAAA,GAAY,MAAM,QAAA,CAAS,SAAA,EAAU;AAC3C,IAAA,IAAI,CAAC,SAAA,EAAW;AACd,MAAA,MAAM,IAAI,MAAM,kCAAkC,CAAA;AAAA,IACpD;AAEA,IAAA,MAAM,MAAA,GAAS,UAAU,SAAA,EAAU;AACnC,IAAA,MAAM,UAAA,GAAa,IAAI,WAAA,EAAY;AAGnC,IAAA,CAAC,YAAY;AACX,MAAA,IAAI;AACF,QAAA,OAAO,IAAA,EAAM;AACX,UAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAM,GAAI,MAAM,OAAO,IAAA,EAAK;AAC1C,UAAA,IAAI,IAAA,EAAM;AACR,YAAA,UAAA,CAAW,GAAA,EAAI;AACf,YAAA;AAAA,UACF;AACA,UAAA,UAAA,CAAW,MAAM,KAAK,CAAA;AAAA,QACxB;AAAA,MACF,SAAS,KAAA,EAAO;AACd,QAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,MACnC;AAAA,IACF,CAAA,GAAG,CAAE,KAAA,CAAM,CAAA,KAAA,KAAS;AAClB,MAAA,UAAA,CAAW,QAAQ,KAAc,CAAA;AAAA,IACnC,CAAC,CAAA;AAED,IAAA,OAAO,UAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,WAAA,GAAc;AAClB,IAAA,OAAO,EAAE,SAAS,IAAA,EAAK;AAAA,EACzB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAaA,MAAM,MAAA,CACJ,WAAA,EACA,OAAA,EAIc;AACd,IAAA,IAAI,CAAC,KAAK,eAAA,EAAiB;AACzB,MAAA,MAAM,IAAI,MAAM,0CAA0C,CAAA;AAAA,IAC5D;AAEA,IAAA,MAAM,SAAmB,EAAC;AAC1B,IAAA,WAAA,MAAiB,SAAS,WAAA,EAAa;AACrC,MAAA,IAAI,OAAO,UAAU,QAAA,EAAU;AAC7B,QAAA,MAAA,CAAO,IAAA,CAAK,MAAA,CAAO,IAAA,CAAK,KAAK,CAAC,CAAA;AAAA,MAChC,CAAA,MAAO;AACL,QAAA,MAAA,CAAO,KAAK,KAAK,CAAA;AAAA,MACnB;AAAA,IACF;AACA,IAAA,MAAM,MAAA,GAAS,MAAA,CAAO,MAAA,CAAO,MAAM,CAAA;AACnC,IAAA,MAAM,EAAE,SAAS,qBAAA,EAAuB,CAAA,EAAG,GAAG,WAAA,EAAY,GAAI,WAAW,EAAC;AAC1E,IAAA,MAAM,EAAE,MAAA,EAAQ,KAAA,EAAM,GAAI,MAAM,KAAK,eAAA,CAAgB,MAAA,CAAO,WAAA,CAAY,cAAA,CAAe,MAAA,EAAQ;AAAA,MAC7F,GAAG,WAAA;AAAA,MACH,KAAA,EAAO,KAAK,oBAAA,EAAsB,IAAA;AAAA,MAClC;AAAA,KACD,CAAA;AACD,IAAA,IAAI,KAAA,EAAO;AACT,MAAA,MAAM,KAAA;AAAA,IACR;AAEA,IAAA,MAAM,OAAA,GAAU,MAAA,CAAO,OAAA,EAAS,QAAA,GAAW,CAAC,CAAA;AAC5C,IAAA,MAAM,GAAA,GAKU,OAAA,EAAS,YAAA,GAAe,CAAC,CAAA;AAEzC,IAAA,IAAI,CAAC,GAAA,EAAK;AACR,MAAA,OAAO;AAAA,QACL,UAAA,EAAY,EAAA;AAAA,QACZ,OAAO,EAAC;AAAA,QACR,GAAA,EAAK;AAAA,OACP;AAAA,IACF;AAEA,IAAA,MAAM,QAAA,GAAgB;AAAA,MACpB,YAAY,GAAA,CAAI,UAAA;AAAA,MAChB,OAAO,GAAA,CAAI,KAAA;AAAA,MACX,GAAA,EAAK;AAAA,KACP;AAEA,IAAA,IAAI,OAAA,IAAW,IAAI,KAAA,EAAO;AACxB,MAAA,QAAA,CAAS,eAAA,GAAkB,GAAA,CAAI,KAAA,CAAM,GAAA,CAAI,CAAC,CAAA,MAAqB;AAAA,QAC7D,MAAM,CAAA,CAAE,IAAA;AAAA,QACR,SAAS,CAAA,CAAE,OAAA;AAAA,QACX,OAAO,CAAA,CAAE,KAAA;AAAA,QACT,KAAK,CAAA,CAAE;AAAA,OACT,CAAE,CAAA;AAAA,IACJ;AAEA,IAAA,OAAO,QAAA;AAAA,EACT;AACF","file":"index.js","sourcesContent":["/**\n * List of available Deepgram voice models for text-to-speech\n * Each voice is designed for specific use cases and languages\n * Format: {name}-{language} (e.g. asteria-en)\n */\nexport const DEEPGRAM_VOICES = [\n 'asteria-en',\n 'luna-en',\n 'stella-en',\n 'athena-en',\n 'hera-en',\n 'orion-en',\n 'arcas-en',\n 'perseus-en',\n 'angus-en',\n 'orpheus-en',\n 'helios-en',\n 'zeus-en',\n] as const;\n\nexport type DeepgramVoiceId = (typeof DEEPGRAM_VOICES)[number];\n\n/**\n * List of available Deepgram models for text-to-speech and speech-to-text\n */\nexport const DEEPGRAM_MODELS = ['aura', 'whisper', 'base', 'enhanced', 'nova', 'nova-2', 'nova-3'] as const;\n\nexport type DeepgramModel = (typeof DEEPGRAM_MODELS)[number];\n","import { PassThrough } from 'stream';\n\nimport { createClient } from '@deepgram/sdk';\nimport { MastraVoice } from '@mastra/core/voice';\n\nimport { DEEPGRAM_VOICES } from './voices';\nimport type { DeepgramVoiceId, DeepgramModel } from './voices';\n\ninterface DeepgramVoiceConfig {\n name?: DeepgramModel;\n apiKey?: string;\n properties?: Record<string, any>;\n language?: string;\n}\n\ninterface DeepgramWord {\n word: string;\n start?: number;\n end?: number;\n confidence?: number;\n speaker?: number;\n}\n\nexport class DeepgramVoice extends MastraVoice {\n private speechClient?: ReturnType<typeof createClient>;\n private listeningClient?: ReturnType<typeof createClient>;\n private storedSpeechModel?: { name: DeepgramModel; apiKey?: string };\n private storedListeningModel?: { name: DeepgramModel; apiKey?: string };\n private storedSpeaker?: DeepgramVoiceId;\n\n constructor({\n speechModel,\n listeningModel,\n speaker,\n }: { speechModel?: DeepgramVoiceConfig; listeningModel?: DeepgramVoiceConfig; speaker?: DeepgramVoiceId } = {}) {\n const defaultApiKey = process.env.DEEPGRAM_API_KEY;\n\n const defaultSpeechModel: { name: DeepgramModel; apiKey?: string } = {\n name: 'aura',\n apiKey: defaultApiKey,\n };\n\n const defaultListeningModel: { name: DeepgramModel; apiKey?: string } = {\n name: 'nova',\n apiKey: defaultApiKey,\n };\n\n super({\n speechModel: {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n },\n listeningModel: {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n },\n speaker,\n });\n\n this.storedSpeechModel = {\n name: speechModel?.name ?? defaultSpeechModel.name,\n apiKey: speechModel?.apiKey ?? defaultSpeechModel.apiKey,\n };\n this.storedListeningModel = {\n name: listeningModel?.name ?? defaultListeningModel.name,\n apiKey: listeningModel?.apiKey ?? defaultListeningModel.apiKey,\n };\n\n const speechApiKey = speechModel?.apiKey || defaultApiKey;\n const listeningApiKey = listeningModel?.apiKey || defaultApiKey;\n\n if (!speechApiKey && !listeningApiKey) {\n throw new Error('At least one of DEEPGRAM_API_KEY, speechModel.apiKey, or listeningModel.apiKey must be set');\n }\n\n if (speechApiKey) {\n this.speechClient = createClient(speechApiKey);\n }\n if (listeningApiKey) {\n this.listeningClient = createClient(listeningApiKey);\n }\n\n this.storedSpeaker = speaker || 'asteria-en';\n }\n\n async getSpeakers() {\n return DEEPGRAM_VOICES.map(voice => ({\n voiceId: voice,\n }));\n }\n\n async speak(\n input: string | NodeJS.ReadableStream,\n options?: {\n speaker?: string;\n [key: string]: any;\n },\n ): Promise<NodeJS.ReadableStream> {\n if (!this.speechClient) {\n throw new Error('Deepgram speech client not configured');\n }\n\n let text: string;\n if (typeof input !== 'string') {\n const chunks: Buffer[] = [];\n for await (const chunk of input) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n text = Buffer.concat(chunks).toString('utf-8');\n } else {\n text = input;\n }\n\n if (text.trim().length === 0) {\n throw new Error('Input text is empty');\n }\n\n const baseModel = this.storedSpeechModel?.name;\n const speakerId = options?.speaker || this.storedSpeaker;\n\n const modelName =\n baseModel && speakerId\n ? speakerId.startsWith(`${baseModel}-`)\n ? speakerId\n : `${baseModel}-${speakerId}`\n : baseModel || speakerId;\n\n const speakClient = this.speechClient.speak;\n const response = await speakClient.request(\n { text },\n {\n model: modelName,\n ...Object.fromEntries(Object.entries(options ?? {}).filter(([k]) => k !== 'speaker')),\n },\n );\n\n const webStream = await response.getStream();\n if (!webStream) {\n throw new Error('No stream returned from Deepgram');\n }\n\n const reader = webStream.getReader();\n const nodeStream = new PassThrough();\n\n // Add error handling for the stream processing\n (async () => {\n try {\n while (true) {\n const { done, value } = await reader.read();\n if (done) {\n nodeStream.end();\n break;\n }\n nodeStream.write(value);\n }\n } catch (error) {\n nodeStream.destroy(error as Error);\n }\n })().catch(error => {\n nodeStream.destroy(error as Error);\n });\n\n return nodeStream;\n }\n\n /**\n * Checks if listening capabilities are enabled.\n *\n * @returns {Promise<{ enabled: boolean }>}\n */\n async getListener() {\n return { enabled: true };\n }\n\n /**\n * Transcribes audio with optional speaker diarization.\n *\n * @param audioStream - Audio input stream\n * @param options - Transcription options (diarize, language, etc.)\n * @returns Promise resolving to:\n * - transcript: Full transcript string\n * - words: Array of word objects with timing and confidence\n * - raw: Complete Deepgram API response\n * - speakerSegments: (when diarize=true) Array of {word, speaker, start, end}\n */\n async listen(\n audioStream: NodeJS.ReadableStream,\n options?: {\n diarize?: boolean;\n [key: string]: any;\n },\n ): Promise<any> {\n if (!this.listeningClient) {\n throw new Error('Deepgram listening client not configured');\n }\n\n const chunks: Buffer[] = [];\n for await (const chunk of audioStream) {\n if (typeof chunk === 'string') {\n chunks.push(Buffer.from(chunk));\n } else {\n chunks.push(chunk);\n }\n }\n const buffer = Buffer.concat(chunks);\n const { diarize, diarize_speaker_count: _, ...restOptions } = options ?? {};\n const { result, error } = await this.listeningClient.listen.prerecorded.transcribeFile(buffer, {\n ...restOptions,\n model: this.storedListeningModel?.name,\n diarize,\n });\n if (error) {\n throw error;\n }\n\n const channel = result.results?.channels?.[0];\n const alt:\n | {\n transcript?: string;\n words?: DeepgramWord[];\n }\n | undefined = channel?.alternatives?.[0];\n\n if (!alt) {\n return {\n transcript: '',\n words: [],\n raw: result,\n };\n }\n\n const response: any = {\n transcript: alt.transcript,\n words: alt.words,\n raw: result,\n };\n\n if (diarize && alt.words) {\n response.speakerSegments = alt.words.map((w: DeepgramWord) => ({\n word: w.word,\n speaker: w.speaker,\n start: w.start,\n end: w.end,\n }));\n }\n\n return response;\n }\n}\n\nexport type { DeepgramVoiceConfig, DeepgramVoiceId, DeepgramModel };\n"]}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@mastra/voice-deepgram",
|
|
3
|
-
"version": "0.12.0-beta.
|
|
3
|
+
"version": "0.12.0-beta.1",
|
|
4
4
|
"description": "Mastra Deepgram voice integration",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"files": [
|
|
@@ -28,14 +28,16 @@
|
|
|
28
28
|
},
|
|
29
29
|
"devDependencies": {
|
|
30
30
|
"@microsoft/api-extractor": "^7.52.8",
|
|
31
|
-
"@types/node": "
|
|
31
|
+
"@types/node": "22.13.17",
|
|
32
|
+
"@vitest/coverage-v8": "4.0.12",
|
|
33
|
+
"@vitest/ui": "4.0.12",
|
|
32
34
|
"eslint": "^9.37.0",
|
|
33
35
|
"tsup": "^8.5.0",
|
|
34
36
|
"typescript": "^5.8.3",
|
|
35
|
-
"vitest": "
|
|
37
|
+
"vitest": "4.0.12",
|
|
36
38
|
"@internal/lint": "0.0.53",
|
|
37
|
-
"@
|
|
38
|
-
"@
|
|
39
|
+
"@internal/types-builder": "0.0.28",
|
|
40
|
+
"@mastra/core": "1.0.0-beta.5"
|
|
39
41
|
},
|
|
40
42
|
"peerDependencies": {
|
|
41
43
|
"@mastra/core": ">=1.0.0-0 <2.0.0-0",
|