@mastra/voice-google-gemini-live 0.11.3 → 0.11.4-alpha.0

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 CHANGED
@@ -1,5 +1,14 @@
1
1
  # @mastra/voice-google-gemini-live
2
2
 
3
+ ## 0.11.4-alpha.0
4
+
5
+ ### Patch Changes
6
+
7
+ - dependencies updates: ([#15538](https://github.com/mastra-ai/mastra/pull/15538))
8
+ - Updated dependency [`ws@^8.20.0` ↗︎](https://www.npmjs.com/package/ws/v/8.20.0) (from `^8.19.0`, in `dependencies`)
9
+ - Updated dependencies [[`20f59b8`](https://github.com/mastra-ai/mastra/commit/20f59b876cf91199efbc49a0e36b391240708f08), [`e2687a7`](https://github.com/mastra-ai/mastra/commit/e2687a7408790c384563816a9a28ed06735684c9), [`8f1b280`](https://github.com/mastra-ai/mastra/commit/8f1b280b7fe6999ec654f160cb69c1a8719e7a57), [`12df98c`](https://github.com/mastra-ai/mastra/commit/12df98c4904643d9481f5c78f3bed443725b4c96)]:
10
+ - @mastra/core@1.26.0-alpha.11
11
+
3
12
  ## 0.11.3
4
13
 
5
14
  ### Patch Changes
@@ -3,7 +3,7 @@ name: mastra-voice-google-gemini-live
3
3
  description: Documentation for @mastra/voice-google-gemini-live. Use when working with @mastra/voice-google-gemini-live APIs, configuration, or implementation.
4
4
  metadata:
5
5
  package: "@mastra/voice-google-gemini-live"
6
- version: "0.11.3"
6
+ version: "0.11.4-alpha.0"
7
7
  ---
8
8
 
9
9
  ## When to use
@@ -1,5 +1,5 @@
1
1
  {
2
- "version": "0.11.3",
2
+ "version": "0.11.4-alpha.0",
3
3
  "package": "@mastra/voice-google-gemini-live",
4
4
  "exports": {},
5
5
  "modules": {}
@@ -16,7 +16,7 @@ const voiceAgent = new Agent({
16
16
  id: 'voice-agent',
17
17
  name: 'Voice Agent',
18
18
  instructions: 'You are a voice assistant that can help users with their tasks.',
19
- model: 'openai/gpt-5.1',
19
+ model: 'openai/gpt-5.4',
20
20
  voice: new OpenAIVoice(),
21
21
  })
22
22
  ```
@@ -40,7 +40,7 @@ const voiceAgent = new Agent({
40
40
  id: 'voice-agent',
41
41
  name: 'Voice Agent',
42
42
  instructions: 'You are a voice assistant that can help users with their tasks.',
43
- model: 'openai/gpt-5.1',
43
+ model: 'openai/gpt-5.4',
44
44
  voice: new OpenAIVoice(),
45
45
  })
46
46
 
@@ -68,7 +68,7 @@ const voiceAgent = new Agent({
68
68
  id: 'voice-agent',
69
69
  name: 'Voice Agent',
70
70
  instructions: 'You are a voice assistant that can help users with their tasks.',
71
- model: 'openai/gpt-5.1',
71
+ model: 'openai/gpt-5.4',
72
72
  voice: new AzureVoice(),
73
73
  })
74
74
 
@@ -95,7 +95,7 @@ const voiceAgent = new Agent({
95
95
  id: 'voice-agent',
96
96
  name: 'Voice Agent',
97
97
  instructions: 'You are a voice assistant that can help users with their tasks.',
98
- model: 'openai/gpt-5.1',
98
+ model: 'openai/gpt-5.4',
99
99
  voice: new ElevenLabsVoice(),
100
100
  })
101
101
 
@@ -122,7 +122,7 @@ const voiceAgent = new Agent({
122
122
  id: 'voice-agent',
123
123
  name: 'Voice Agent',
124
124
  instructions: 'You are a voice assistant that can help users with their tasks.',
125
- model: 'openai/gpt-5.1',
125
+ model: 'openai/gpt-5.4',
126
126
  voice: new PlayAIVoice(),
127
127
  })
128
128
 
@@ -149,7 +149,7 @@ const voiceAgent = new Agent({
149
149
  id: 'voice-agent',
150
150
  name: 'Voice Agent',
151
151
  instructions: 'You are a voice assistant that can help users with their tasks.',
152
- model: 'openai/gpt-5.1',
152
+ model: 'openai/gpt-5.4',
153
153
  voice: new GoogleVoice(),
154
154
  })
155
155
 
@@ -176,7 +176,7 @@ const voiceAgent = new Agent({
176
176
  id: 'voice-agent',
177
177
  name: 'Voice Agent',
178
178
  instructions: 'You are a voice assistant that can help users with their tasks.',
179
- model: 'openai/gpt-5.1',
179
+ model: 'openai/gpt-5.4',
180
180
  voice: new CloudflareVoice(),
181
181
  })
182
182
 
@@ -203,7 +203,7 @@ const voiceAgent = new Agent({
203
203
  id: 'voice-agent',
204
204
  name: 'Voice Agent',
205
205
  instructions: 'You are a voice assistant that can help users with their tasks.',
206
- model: 'openai/gpt-5.1',
206
+ model: 'openai/gpt-5.4',
207
207
  voice: new DeepgramVoice(),
208
208
  })
209
209
 
@@ -230,7 +230,7 @@ const voiceAgent = new Agent({
230
230
  id: 'voice-agent',
231
231
  name: 'Voice Agent',
232
232
  instructions: 'You are a voice assistant that can help users with their tasks.',
233
- model: 'openai/gpt-5.1',
233
+ model: 'openai/gpt-5.4',
234
234
  voice: new SpeechifyVoice(),
235
235
  })
236
236
 
@@ -257,7 +257,7 @@ const voiceAgent = new Agent({
257
257
  id: 'voice-agent',
258
258
  name: 'Voice Agent',
259
259
  instructions: 'You are a voice assistant that can help users with their tasks.',
260
- model: 'openai/gpt-5.1',
260
+ model: 'openai/gpt-5.4',
261
261
  voice: new SarvamVoice(),
262
262
  })
263
263
 
@@ -265,7 +265,7 @@ const { text } = await voiceAgent.generate('What color is the sky?')
265
265
 
266
266
  // Convert text to speech to an Audio Stream
267
267
  const audioStream = await voiceAgent.voice.speak(text, {
268
- speaker: 'default', // Optional: specify a speaker
268
+ speaker: 'shubh', // Optional: specify a bulbul:v3 speaker
269
269
  })
270
270
 
271
271
  playAudio(audioStream)
@@ -284,7 +284,7 @@ const voiceAgent = new Agent({
284
284
  id: 'voice-agent',
285
285
  name: 'Voice Agent',
286
286
  instructions: 'You are a voice assistant that can help users with their tasks.',
287
- model: 'openai/gpt-5.1',
287
+ model: 'openai/gpt-5.4',
288
288
  voice: new MurfVoice(),
289
289
  })
290
290
 
@@ -319,7 +319,7 @@ const voiceAgent = new Agent({
319
319
  id: 'voice-agent',
320
320
  name: 'Voice Agent',
321
321
  instructions: 'You are a voice assistant that can help users with their tasks.',
322
- model: 'openai/gpt-5.1',
322
+ model: 'openai/gpt-5.4',
323
323
  voice: new OpenAIVoice(),
324
324
  })
325
325
 
@@ -348,7 +348,7 @@ const voiceAgent = new Agent({
348
348
  id: 'voice-agent',
349
349
  name: 'Voice Agent',
350
350
  instructions: 'You are a voice assistant that can help users with their tasks.',
351
- model: 'openai/gpt-5.1',
351
+ model: 'openai/gpt-5.4',
352
352
  voice: new AzureVoice(),
353
353
  })
354
354
 
@@ -376,7 +376,7 @@ const voiceAgent = new Agent({
376
376
  id: 'voice-agent',
377
377
  name: 'Voice Agent',
378
378
  instructions: 'You are a voice assistant that can help users with their tasks.',
379
- model: 'openai/gpt-5.1',
379
+ model: 'openai/gpt-5.4',
380
380
  voice: new ElevenLabsVoice(),
381
381
  })
382
382
 
@@ -404,7 +404,7 @@ const voiceAgent = new Agent({
404
404
  id: 'voice-agent',
405
405
  name: 'Voice Agent',
406
406
  instructions: 'You are a voice assistant that can help users with their tasks.',
407
- model: 'openai/gpt-5.1',
407
+ model: 'openai/gpt-5.4',
408
408
  voice: new GoogleVoice(),
409
409
  })
410
410
 
@@ -432,7 +432,7 @@ const voiceAgent = new Agent({
432
432
  id: 'voice-agent',
433
433
  name: 'Voice Agent',
434
434
  instructions: 'You are a voice assistant that can help users with their tasks.',
435
- model: 'openai/gpt-5.1',
435
+ model: 'openai/gpt-5.4',
436
436
  voice: new CloudflareVoice(),
437
437
  })
438
438
 
@@ -460,7 +460,7 @@ const voiceAgent = new Agent({
460
460
  id: 'voice-agent',
461
461
  name: 'Voice Agent',
462
462
  instructions: 'You are a voice assistant that can help users with their tasks.',
463
- model: 'openai/gpt-5.1',
463
+ model: 'openai/gpt-5.4',
464
464
  voice: new DeepgramVoice(),
465
465
  })
466
466
 
@@ -488,7 +488,7 @@ const voiceAgent = new Agent({
488
488
  id: 'voice-agent',
489
489
  name: 'Voice Agent',
490
490
  instructions: 'You are a voice assistant that can help users with their tasks.',
491
- model: 'openai/gpt-5.1',
491
+ model: 'openai/gpt-5.4',
492
492
  voice: new SarvamVoice(),
493
493
  })
494
494
 
@@ -520,7 +520,7 @@ const voiceAgent = new Agent({
520
520
  id: 'voice-agent',
521
521
  name: 'Voice Agent',
522
522
  instructions: 'You are a voice assistant that can help users with their tasks.',
523
- model: 'openai/gpt-5.1',
523
+ model: 'openai/gpt-5.4',
524
524
  voice: new OpenAIRealtimeVoice(),
525
525
  })
526
526
 
@@ -550,7 +550,7 @@ const voiceAgent = new Agent({
550
550
  id: 'voice-agent',
551
551
  name: 'Voice Agent',
552
552
  instructions: 'You are a voice assistant that can help users with their tasks.',
553
- model: 'openai/gpt-5.1',
553
+ model: 'openai/gpt-5.4',
554
554
  voice: new GeminiLiveVoice({
555
555
  // Live API mode
556
556
  apiKey: process.env.GOOGLE_API_KEY,
@@ -760,12 +760,15 @@ Visit the [Speechify Voice Reference](https://mastra.ai/reference/voice/speechif
760
760
  // Sarvam Voice Configuration
761
761
  const voice = new SarvamVoice({
762
762
  speechModel: {
763
- name: 'sarvam-voice', // Example model name
763
+ model: 'bulbul:v3', // TTS model (bulbul:v2 or bulbul:v3)
764
+ apiKey: process.env.SARVAM_API_KEY,
765
+ language: 'en-IN', // BCP-47 language code
766
+ },
767
+ listeningModel: {
768
+ model: 'saarika:v2.5', // STT model (saarika:v2.5 or saaras:v3)
764
769
  apiKey: process.env.SARVAM_API_KEY,
765
- language: 'en-IN', // Language code
766
- style: 'conversational', // Style setting
767
770
  },
768
- // Sarvam may not have a separate listening model
771
+ speaker: 'shubh', // Default bulbul:v3 speaker
769
772
  })
770
773
  ```
771
774
 
@@ -844,7 +847,7 @@ const voiceAgent = new Agent({
844
847
  id: 'aisdk-voice-agent',
845
848
  name: 'AI SDK Voice Agent',
846
849
  instructions: 'You are a helpful assistant with voice capabilities.',
847
- model: 'openai/gpt-5.1',
850
+ model: 'openai/gpt-5.4',
848
851
  voice,
849
852
  })
850
853
  ```
@@ -32,7 +32,7 @@ const agent = new Agent({
32
32
  id: 'agent',
33
33
  name: 'OpenAI Realtime Agent',
34
34
  instructions: `You are a helpful assistant with real-time voice capabilities.`,
35
- model: 'openai/gpt-5.1',
35
+ model: 'openai/gpt-5.4',
36
36
  voice: new OpenAIRealtimeVoice(),
37
37
  })
38
38
 
@@ -66,7 +66,7 @@ const agent = new Agent({
66
66
  name: 'Gemini Live Agent',
67
67
  instructions: 'You are a helpful assistant with real-time voice capabilities.',
68
68
  // Model used for text generation; voice provider handles realtime audio
69
- model: 'openai/gpt-5.1',
69
+ model: 'openai/gpt-5.4',
70
70
  voice: new GeminiLiveVoice({
71
71
  apiKey: process.env.GOOGLE_API_KEY,
72
72
  model: 'gemini-2.0-flash-exp',
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@mastra/voice-google-gemini-live",
3
- "version": "0.11.3",
3
+ "version": "0.11.4-alpha.0",
4
4
  "description": "Mastra Google Gemini Live API integration",
5
5
  "type": "module",
6
6
  "files": [
@@ -24,29 +24,28 @@
24
24
  },
25
25
  "license": "Apache-2.0",
26
26
  "dependencies": {
27
- "@google/genai": "latest",
28
- "google-auth-library": "^10.5.0",
29
- "ws": "^8.19.0",
30
- "zod": "^4.3.6",
31
- "zod-to-json-schema": "^3.24.6"
27
+ "@google/genai": "^1.45.0",
28
+ "google-auth-library": "^10.6.1",
29
+ "ws": "^8.20.0"
32
30
  },
33
31
  "devDependencies": {
34
- "@types/node": "22.19.13",
32
+ "@types/node": "22.19.15",
35
33
  "@types/ws": "^8.18.1",
36
- "@vitest/coverage-v8": "4.0.18",
37
- "@vitest/ui": "4.0.18",
38
- "eslint": "^9.39.4",
34
+ "@vitest/coverage-v8": "4.1.4",
35
+ "@vitest/ui": "4.1.4",
36
+ "eslint": "^10.2.1",
39
37
  "tsup": "^8.5.1",
40
- "tsx": "latest",
38
+ "tsx": "^4.21.0",
41
39
  "typescript": "^5.9.3",
42
- "vitest": "4.0.18",
43
- "@internal/lint": "0.0.67",
44
- "@mastra/core": "1.11.0",
45
- "@internal/types-builder": "0.0.42"
40
+ "vitest": "4.1.4",
41
+ "zod": "^4.3.6",
42
+ "zod-to-json-schema": "^3.25.1",
43
+ "@internal/types-builder": "0.0.58",
44
+ "@internal/lint": "0.0.83",
45
+ "@mastra/core": "1.26.0-alpha.11"
46
46
  },
47
47
  "peerDependencies": {
48
- "@mastra/core": ">=1.0.0-0 <2.0.0-0",
49
- "zod": "^3.0.0"
48
+ "@mastra/core": ">=1.0.0-0 <2.0.0-0"
50
49
  },
51
50
  "homepage": "https://mastra.ai",
52
51
  "repository": {