together-ai 0.28.0 → 0.29.0

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.
Files changed (24) hide show
  1. package/CHANGELOG.md +8 -0
  2. package/package.json +1 -1
  3. package/resources/audio/audio.d.mts +20 -8
  4. package/resources/audio/audio.d.mts.map +1 -1
  5. package/resources/audio/audio.d.ts +20 -8
  6. package/resources/audio/audio.d.ts.map +1 -1
  7. package/resources/audio/audio.js.map +1 -1
  8. package/resources/audio/audio.mjs.map +1 -1
  9. package/resources/audio/transcriptions.d.mts +67 -0
  10. package/resources/audio/transcriptions.d.mts.map +1 -1
  11. package/resources/audio/transcriptions.d.ts +67 -0
  12. package/resources/audio/transcriptions.d.ts.map +1 -1
  13. package/resources/audio/translations.d.mts +4 -0
  14. package/resources/audio/translations.d.mts.map +1 -1
  15. package/resources/audio/translations.d.ts +4 -0
  16. package/resources/audio/translations.d.ts.map +1 -1
  17. package/src/resources/audio/audio.ts +20 -8
  18. package/src/resources/audio/transcriptions.ts +80 -0
  19. package/src/resources/audio/translations.ts +5 -0
  20. package/src/version.ts +1 -1
  21. package/version.d.mts +1 -1
  22. package/version.d.ts +1 -1
  23. package/version.js +1 -1
  24. package/version.mjs +1 -1
package/CHANGELOG.md CHANGED
@@ -1,5 +1,13 @@
1
1
  # Changelog
2
2
 
3
+ ## 0.29.0 (2025-10-30)
4
+
5
+ Full Changelog: [v0.28.0...v0.29.0](https://github.com/togethercomputer/together-typescript/compare/v0.28.0...v0.29.0)
6
+
7
+ ### Features
8
+
9
+ * **api:** api update ([aeac27b](https://github.com/togethercomputer/together-typescript/commit/aeac27be216e075e69f1c031e732c178f94925b5))
10
+
3
11
  ## 0.28.0 (2025-10-23)
4
12
 
5
13
  Full Changelog: [v0.27.0...v0.28.0](https://github.com/togethercomputer/together-typescript/compare/v0.27.0...v0.28.0)
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "together-ai",
3
- "version": "0.28.0",
3
+ "version": "0.29.0",
4
4
  "description": "The official TypeScript library for the Together API",
5
5
  "author": "Together <dev-feedback@TogetherAI.com>",
6
6
  "types": "./index.d.ts",
package/resources/audio/audio.d.mts CHANGED
@@ -17,8 +17,8 @@ export declare class Audio extends APIResource {
17
17
  * ```ts
18
18
  * const audio = await client.audio.create({
19
19
  * input: 'input',
20
- * model: 'cartesia/sonic',
21
- * voice: 'laidback woman',
20
+ * model: 'canopylabs/orpheus-3b-0.1-ft',
21
+ * voice: 'voice',
22
22
  * });
23
23
  *
24
24
  * const content = await audio.blob();
@@ -56,15 +56,24 @@ export interface AudioCreateParamsBase {
56
56
  * The name of the model to query.
57
57
  *
58
58
  * [See all of Together AI's chat models](https://docs.together.ai/docs/serverless-models#audio-models)
59
+ * The current supported tts models are: - cartesia/sonic - hexgrad/Kokoro-82M -
60
+ * canopylabs/orpheus-3b-0.1-ft
59
61
  */
60
- model: 'cartesia/sonic' | (string & {});
62
+ model: 'cartesia/sonic' | 'hexgrad/Kokoro-82M' | 'canopylabs/orpheus-3b-0.1-ft' | (string & {});
61
63
  /**
62
- * The voice to use for generating the audio.
64
+ * The voice to use for generating the audio. The voices supported are different
65
+ * for each model. For eg - for canopylabs/orpheus-3b-0.1-ft, one of the voices
66
+ * supported is tara, for hexgrad/Kokoro-82M, one of the voices supported is
67
+ * af_alloy and for cartesia/sonic, one of the voices supported is "friendly
68
+ * sidekick".
69
+ *
70
+ * You can view the voices supported for each model using the /v1/voices endpoint
71
+ * sending the model name as the query parameter.
63
72
  * [View all supported voices here](https://docs.together.ai/docs/text-to-speech#voices-available).
64
73
  */
65
- voice: 'laidback woman' | 'polite man' | 'storyteller lady' | 'friendly sidekick' | (string & {});
74
+ voice: string;
66
75
  /**
67
- * Language of input text
76
+ * Language of input text.
68
77
  */
69
78
  language?: 'en' | 'de' | 'fr' | 'es' | 'hi' | 'it' | 'ja' | 'ko' | 'nl' | 'pl' | 'pt' | 'ru' | 'sv' | 'tr' | 'zh';
70
79
  /**
@@ -72,11 +81,14 @@ export interface AudioCreateParamsBase {
72
81
  */
73
82
  response_encoding?: 'pcm_f32le' | 'pcm_s16le' | 'pcm_mulaw' | 'pcm_alaw';
74
83
  /**
75
- * The format of audio output
84
+ * The format of audio output. Supported formats are mp3, wav, raw if streaming is
85
+ * false. If streaming is true, the only supported format is raw.
76
86
  */
77
87
  response_format?: 'mp3' | 'wav' | 'raw';
78
88
  /**
79
- * Sampling rate to use for the output audio
89
+ * Sampling rate to use for the output audio. The default sampling rate for
90
+ * canopylabs/orpheus-3b-0.1-ft and hexgrad/Kokoro-82M is 24000 and for
91
+ * cartesia/sonic is 44100.
80
92
  */
81
93
  sample_rate?: number;
82
94
  /**
package/resources/audio/audio.d.mts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"audio.d.mts","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,KAAK,QAAQ;OACb,KAAK,iBAAiB;OACtB,EAAE,yBAAyB,EAAE,2BAA2B,EAAE,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAE,uBAAuB,EAAE,yBAAyB,EAAE,YAAY,EAAE;OACpE,EAAE,UAAU,EAAE;OACd,EAAE,MAAM,EAAE;OAEV,EAAE,cAAc,EAAE;AAEzB,qBAAa,KAAM,SAAQ,WAAW;IACpC,cAAc,EAAE,iBAAiB,CAAC,cAAc,CAAsD;IACtG,YAAY,EAAE,eAAe,CAAC,YAAY,CAAkD;IAE5F;;;;;;;;;;;;;;OAcG;IACH,MAAM,CAAC,IAAI,EAAE,6BAA6B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,QAAQ,CAAC;IAC3F,MAAM,CACJ,IAAI,EAAE,0BAA0B,EAChC,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,CAAC;IAC7C,MAAM,CACJ,IAAI,EAAE,qBAAqB,EAC3B,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,GAAG,QAAQ,CAAC;CAazD;AAED,MAAM,MAAM,SAAS,GAAG,SAAS,CAAC,sBAAsB,GAAG,SAAS,CAAC,cAAc,CAAC;AAEpF,yBAAiB,SAAS,CAAC;IACzB,UAAiB,sBAAsB;QACrC,IAAI,EAAE,QAAQ,CAAC,sBAAsB,CAAC;KACvC;IAED,UAAiB,cAAc;QAC7B,IAAI,EAAE,QAAQ,CAAC;KAChB;CACF;AAED,MAAM,WAAW,sBAAsB;IACrC;;OAEG;IACH,GAAG,EAAE,MAAM,CAAC;IAEZ,KAAK,EAAE,MAAM,CAAC;IAEd,MAAM,EAAE,iBAAiB,CAAC;CAC3B;AAED,MAAM,MAAM,iBAAiB,GAAG,6BAA6B,GAAG,0BAA0B,CAAC;AAE3F,MAAM,WAAW,qBAAqB;IACpC;;OAEG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;;;OAIG;IACH,KAAK,EAAE,gBAAgB,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAExC;;;OAGG;IACH,KAAK,EAAE,gBAAgB,GAAG,YAAY,GAAG,kBAAkB,GAAG,mBAAmB,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAElG;;OAEG;IACH,QAAQ,CAAC,EACL,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,CAAC;IAET;;OAEG;IACH,iBAAiB,CAAC,EAAE,WAAW,GAAG,WAAW,GAAG,WAAW,GAAG,UAAU,CAAC;IAEzE;;OAEG;IACH,eAAe,CAAC,EAAE,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;IAExC;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,yBAAiB,iBAAiB,CAAC;IACjC,KAAY,6BAA6B,GAAG,QAAQ,CAAC,6BAA6B,CAAC;IACnF,KAAY,0BAA0B,GAAG,QAAQ,CAAC,0BAA0B,CAAC;CAC9E;AAED,MAAM,WAAW,6BAA8B,SAAQ,qBAAqB;IAC1E;;;;OAIG;IACH,MAAM,CAAC,EAAE,KAAK,CAAC;CAChB;AAED,MAAM,WAAW,0BAA2B,SAAQ,qBAAqB;IACvE;;;;OAIG;IACH,MAAM,EAAE,IAAI,CAAC;CACd;AAKD,MAAM,CAAC,OAAO,WAAW,KAAK,CAAC;IAC7B,OAAO,EACL,KAAK,SAAS,IAAI,SAAS,EAC3B,KAAK,sBAAsB,IAAI,sBAAsB,EACrD,KAAK,iBAAiB,IAAI,iBAAiB,EAC3C,KAAK,6BAA6B,IAAI,6BAA6B,EACnE,KAAK,0BAA0B,IAAI,0BAA0B,GAC9D,CAAC;IAEF,OAAO,EACL,cAAc,IAAI,cAAc,EAChC,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;IAEF,OAAO,EACL,YAAY,IAAI,YAAY,EAC5B,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
1
+ {"version":3,"file":"audio.d.mts","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,KAAK,QAAQ;OACb,KAAK,iBAAiB;OACtB,EAAE,yBAAyB,EAAE,2BAA2B,EAAE,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAE,uBAAuB,EAAE,yBAAyB,EAAE,YAAY,EAAE;OACpE,EAAE,UAAU,EAAE;OACd,EAAE,MAAM,EAAE;OAEV,EAAE,cAAc,EAAE;AAEzB,qBAAa,KAAM,SAAQ,WAAW;IACpC,cAAc,EAAE,iBAAiB,CAAC,cAAc,CAAsD;IACtG,YAAY,EAAE,eAAe,CAAC,YAAY,CAAkD;IAE5F;;;;;;;;;;;;;;OAcG;IACH,MAAM,CAAC,IAAI,EAAE,6BAA6B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,QAAQ,CAAC;IAC3F,MAAM,CACJ,IAAI,EAAE,0BAA0B,EAChC,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,CAAC;IAC7C,MAAM,CACJ,IAAI,EAAE,qBAAqB,EAC3B,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,GAAG,QAAQ,CAAC;CAazD;AAED,MAAM,MAAM,SAAS,GAAG,SAAS,CAAC,sBAAsB,GAAG,SAAS,CAAC,cAAc,CAAC;AAEpF,yBAAiB,SAAS,CAAC;IACzB,UAAiB,sBAAsB;QACrC,IAAI,EAAE,QAAQ,CAAC,sBAAsB,CAAC;KACvC;IAED,UAAiB,cAAc;QAC7B,IAAI,EAAE,QAAQ,CAAC;KAChB;CACF;AAED,MAAM,WAAW,sBAAsB;IACrC;;OAEG;IACH,GAAG,EAAE,MAAM,CAAC;IAEZ,KAAK,EAAE,MAAM,CAAC;IAEd,MAAM,EAAE,iBAAiB,CAAC;CAC3B;AAED,MAAM,MAAM,iBAAiB,GAAG,6BAA6B,GAAG,0BAA0B,CAAC;AAE3F,MAAM,WAAW,qBAAqB;IACpC;;OAEG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;;;;;OAMG;IACH,KAAK,EAAE,gBAAgB,GAAG,oBAAoB,GAAG,8BAA8B,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAEhG;;;;;;;;;;OAUG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;OAEG;IACH,QAAQ,CAAC,EACL,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,CAAC;IAET;;OAEG;IACH,iBAAiB,CAAC,EAAE,WAAW,GAAG,WAAW,GAAG,WAAW,GAAG,UAAU,CAAC;IAEzE;;;OAGG;IACH,eAAe,CAAC,EAAE,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;IAExC;;;;OAIG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,yBAAiB,iBAAiB,CAAC;IACjC,KAAY,6BAA6B,GAAG,QAAQ,CAAC,6BAA6B,CAAC;IACnF,KAAY,0BAA0B,GAAG,QAAQ,CAAC,0BAA0B,CAAC;CAC9E;AAED,MAAM,WAAW,6BAA8B,SAAQ,qBAAqB;IAC1E;;;;OAIG;IACH,MAAM,CAAC,EAAE,KAAK,CAAC;CAChB;AAED,MAAM,WAAW,0BAA2B,SAAQ,qBAAqB;IACvE;;;;OAIG;IACH,MAAM,EAAE,IAAI,CAAC;CACd;AAKD,MAAM,CAAC,OAAO,WAAW,KAAK,CAAC;IAC7B,OAAO,EACL,KAAK,SAAS,IAAI,SAAS,EAC3B,KAAK,sBAAsB,IAAI,sBAAsB,EACrD,KAAK,iBAAiB,IAAI,iBAAiB,EAC3C,KAAK,6BAA6B,IAAI,6BAA6B,EACnE,KAAK,0BAA0B,IAAI,0BAA0B,GAC9D,CAAC;IAEF,OAAO,EACL,cAAc,IAAI,cAAc,EAChC,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;IAEF,OAAO,EACL,YAAY,IAAI,YAAY,EAC5B,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
package/resources/audio/audio.d.ts CHANGED
@@ -17,8 +17,8 @@ export declare class Audio extends APIResource {
17
17
  * ```ts
18
18
  * const audio = await client.audio.create({
19
19
  * input: 'input',
20
- * model: 'cartesia/sonic',
21
- * voice: 'laidback woman',
20
+ * model: 'canopylabs/orpheus-3b-0.1-ft',
21
+ * voice: 'voice',
22
22
  * });
23
23
  *
24
24
  * const content = await audio.blob();
@@ -56,15 +56,24 @@ export interface AudioCreateParamsBase {
56
56
  * The name of the model to query.
57
57
  *
58
58
  * [See all of Together AI's chat models](https://docs.together.ai/docs/serverless-models#audio-models)
59
+ * The current supported tts models are: - cartesia/sonic - hexgrad/Kokoro-82M -
60
+ * canopylabs/orpheus-3b-0.1-ft
59
61
  */
60
- model: 'cartesia/sonic' | (string & {});
62
+ model: 'cartesia/sonic' | 'hexgrad/Kokoro-82M' | 'canopylabs/orpheus-3b-0.1-ft' | (string & {});
61
63
  /**
62
- * The voice to use for generating the audio.
64
+ * The voice to use for generating the audio. The voices supported are different
65
+ * for each model. For eg - for canopylabs/orpheus-3b-0.1-ft, one of the voices
66
+ * supported is tara, for hexgrad/Kokoro-82M, one of the voices supported is
67
+ * af_alloy and for cartesia/sonic, one of the voices supported is "friendly
68
+ * sidekick".
69
+ *
70
+ * You can view the voices supported for each model using the /v1/voices endpoint
71
+ * sending the model name as the query parameter.
63
72
  * [View all supported voices here](https://docs.together.ai/docs/text-to-speech#voices-available).
64
73
  */
65
- voice: 'laidback woman' | 'polite man' | 'storyteller lady' | 'friendly sidekick' | (string & {});
74
+ voice: string;
66
75
  /**
67
- * Language of input text
76
+ * Language of input text.
68
77
  */
69
78
  language?: 'en' | 'de' | 'fr' | 'es' | 'hi' | 'it' | 'ja' | 'ko' | 'nl' | 'pl' | 'pt' | 'ru' | 'sv' | 'tr' | 'zh';
70
79
  /**
@@ -72,11 +81,14 @@ export interface AudioCreateParamsBase {
72
81
  */
73
82
  response_encoding?: 'pcm_f32le' | 'pcm_s16le' | 'pcm_mulaw' | 'pcm_alaw';
74
83
  /**
75
- * The format of audio output
84
+ * The format of audio output. Supported formats are mp3, wav, raw if streaming is
85
+ * false. If streaming is true, the only supported format is raw.
76
86
  */
77
87
  response_format?: 'mp3' | 'wav' | 'raw';
78
88
  /**
79
- * Sampling rate to use for the output audio
89
+ * Sampling rate to use for the output audio. The default sampling rate for
90
+ * canopylabs/orpheus-3b-0.1-ft and hexgrad/Kokoro-82M is 24000 and for
91
+ * cartesia/sonic is 44100.
80
92
  */
81
93
  sample_rate?: number;
82
94
  /**
package/resources/audio/audio.d.ts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"audio.d.ts","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,KAAK,QAAQ;OACb,KAAK,iBAAiB;OACtB,EAAE,yBAAyB,EAAE,2BAA2B,EAAE,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAE,uBAAuB,EAAE,yBAAyB,EAAE,YAAY,EAAE;OACpE,EAAE,UAAU,EAAE;OACd,EAAE,MAAM,EAAE;OAEV,EAAE,cAAc,EAAE;AAEzB,qBAAa,KAAM,SAAQ,WAAW;IACpC,cAAc,EAAE,iBAAiB,CAAC,cAAc,CAAsD;IACtG,YAAY,EAAE,eAAe,CAAC,YAAY,CAAkD;IAE5F;;;;;;;;;;;;;;OAcG;IACH,MAAM,CAAC,IAAI,EAAE,6BAA6B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,QAAQ,CAAC;IAC3F,MAAM,CACJ,IAAI,EAAE,0BAA0B,EAChC,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,CAAC;IAC7C,MAAM,CACJ,IAAI,EAAE,qBAAqB,EAC3B,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,GAAG,QAAQ,CAAC;CAazD;AAED,MAAM,MAAM,SAAS,GAAG,SAAS,CAAC,sBAAsB,GAAG,SAAS,CAAC,cAAc,CAAC;AAEpF,yBAAiB,SAAS,CAAC;IACzB,UAAiB,sBAAsB;QACrC,IAAI,EAAE,QAAQ,CAAC,sBAAsB,CAAC;KACvC;IAED,UAAiB,cAAc;QAC7B,IAAI,EAAE,QAAQ,CAAC;KAChB;CACF;AAED,MAAM,WAAW,sBAAsB;IACrC;;OAEG;IACH,GAAG,EAAE,MAAM,CAAC;IAEZ,KAAK,EAAE,MAAM,CAAC;IAEd,MAAM,EAAE,iBAAiB,CAAC;CAC3B;AAED,MAAM,MAAM,iBAAiB,GAAG,6BAA6B,GAAG,0BAA0B,CAAC;AAE3F,MAAM,WAAW,qBAAqB;IACpC;;OAEG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;;;OAIG;IACH,KAAK,EAAE,gBAAgB,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAExC;;;OAGG;IACH,KAAK,EAAE,gBAAgB,GAAG,YAAY,GAAG,kBAAkB,GAAG,mBAAmB,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAElG;;OAEG;IACH,QAAQ,CAAC,EACL,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,CAAC;IAET;;OAEG;IACH,iBAAiB,CAAC,EAAE,WAAW,GAAG,WAAW,GAAG,WAAW,GAAG,UAAU,CAAC;IAEzE;;OAEG;IACH,eAAe,CAAC,EAAE,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;IAExC;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,yBAAiB,iBAAiB,CAAC;IACjC,KAAY,6BAA6B,GAAG,QAAQ,CAAC,6BAA6B,CAAC;IACnF,KAAY,0BAA0B,GAAG,QAAQ,CAAC,0BAA0B,CAAC;CAC9E;AAED,MAAM,WAAW,6BAA8B,SAAQ,qBAAqB;IAC1E;;;;OAIG;IACH,MAAM,CAAC,EAAE,KAAK,CAAC;CAChB;AAED,MAAM,WAAW,0BAA2B,SAAQ,qBAAqB;IACvE;;;;OAIG;IACH,MAAM,EAAE,IAAI,CAAC;CACd;AAKD,MAAM,CAAC,OAAO,WAAW,KAAK,CAAC;IAC7B,OAAO,EACL,KAAK,SAAS,IAAI,SAAS,EAC3B,KAAK,sBAAsB,IAAI,sBAAsB,EACrD,KAAK,iBAAiB,IAAI,iBAAiB,EAC3C,KAAK,6BAA6B,IAAI,6BAA6B,EACnE,KAAK,0BAA0B,IAAI,0BAA0B,GAC9D,CAAC;IAEF,OAAO,EACL,cAAc,IAAI,cAAc,EAChC,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;IAEF,OAAO,EACL,YAAY,IAAI,YAAY,EAC5B,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
1
+ {"version":3,"file":"audio.d.ts","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,KAAK,QAAQ;OACb,KAAK,iBAAiB;OACtB,EAAE,yBAAyB,EAAE,2BAA2B,EAAE,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAE,uBAAuB,EAAE,yBAAyB,EAAE,YAAY,EAAE;OACpE,EAAE,UAAU,EAAE;OACd,EAAE,MAAM,EAAE;OAEV,EAAE,cAAc,EAAE;AAEzB,qBAAa,KAAM,SAAQ,WAAW;IACpC,cAAc,EAAE,iBAAiB,CAAC,cAAc,CAAsD;IACtG,YAAY,EAAE,eAAe,CAAC,YAAY,CAAkD;IAE5F;;;;;;;;;;;;;;OAcG;IACH,MAAM,CAAC,IAAI,EAAE,6BAA6B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,QAAQ,CAAC;IAC3F,MAAM,CACJ,IAAI,EAAE,0BAA0B,EAChC,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,CAAC;IAC7C,MAAM,CACJ,IAAI,EAAE,qBAAqB,EAC3B,OAAO,CAAC,EAAE,cAAc,GACvB,UAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,GAAG,QAAQ,CAAC;CAazD;AAED,MAAM,MAAM,SAAS,GAAG,SAAS,CAAC,sBAAsB,GAAG,SAAS,CAAC,cAAc,CAAC;AAEpF,yBAAiB,SAAS,CAAC;IACzB,UAAiB,sBAAsB;QACrC,IAAI,EAAE,QAAQ,CAAC,sBAAsB,CAAC;KACvC;IAED,UAAiB,cAAc;QAC7B,IAAI,EAAE,QAAQ,CAAC;KAChB;CACF;AAED,MAAM,WAAW,sBAAsB;IACrC;;OAEG;IACH,GAAG,EAAE,MAAM,CAAC;IAEZ,KAAK,EAAE,MAAM,CAAC;IAEd,MAAM,EAAE,iBAAiB,CAAC;CAC3B;AAED,MAAM,MAAM,iBAAiB,GAAG,6BAA6B,GAAG,0BAA0B,CAAC;AAE3F,MAAM,WAAW,qBAAqB;IACpC;;OAEG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;;;;;OAMG;IACH,KAAK,EAAE,gBAAgB,GAAG,oBAAoB,GAAG,8BAA8B,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAC;IAEhG;;;;;;;;;;OAUG;IACH,KAAK,EAAE,MAAM,CAAC;IAEd;;OAEG;IACH,QAAQ,CAAC,EACL,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,GACJ,IAAI,CAAC;IAET;;OAEG;IACH,iBAAiB,CAAC,EAAE,WAAW,GAAG,WAAW,GAAG,WAAW,GAAG,UAAU,CAAC;IAEzE;;;OAGG;IACH,eAAe,CAAC,EAAE,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;IAExC;;;;OAIG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB;AAED,yBAAiB,iBAAiB,CAAC;IACjC,KAAY,6BAA6B,GAAG,QAAQ,CAAC,6BAA6B,CAAC;IACnF,KAAY,0BAA0B,GAAG,QAAQ,CAAC,0BAA0B,CAAC;CAC9E;AAED,MAAM,WAAW,6BAA8B,SAAQ,qBAAqB;IAC1E;;;;OAIG;IACH,MAAM,CAAC,EAAE,KAAK,CAAC;CAChB;AAED,MAAM,WAAW,0BAA2B,SAAQ,qBAAqB;IACvE;;;;OAIG;IACH,MAAM,EAAE,IAAI,CAAC;CACd;AAKD,MAAM,CAAC,OAAO,WAAW,KAAK,CAAC;IAC7B,OAAO,EACL,KAAK,SAAS,IAAI,SAAS,EAC3B,KAAK,sBAAsB,IAAI,sBAAsB,EACrD,KAAK,iBAAiB,IAAI,iBAAiB,EAC3C,KAAK,6BAA6B,IAAI,6BAA6B,EACnE,KAAK,0BAA0B,IAAI,0BAA0B,GAC9D,CAAC;IAEF,OAAO,EACL,cAAc,IAAI,cAAc,EAChC,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;IAEF,OAAO,EACL,YAAY,IAAI,YAAY,EAC5B,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
package/resources/audio/audio.js.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"audio.js","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":";AAAA,sFAAsF;;;;AAEtF,qDAAkD;AAElD,+EAAsD;AACtD,wDAA0G;AAC1G,2EAAkD;AAClD,oDAAkG;AAGlG,uDAAsD;AAGtD,MAAa,KAAM,SAAQ,sBAAW;IAAtC;;QACE,mBAAc,GAAqC,IAAI,iBAAiB,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACtG,iBAAY,GAAiC,IAAI,eAAe,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAsC9F,CAAC;IAZC,MAAM,CACJ,IAAuB,EACvB,OAAwB;QAExB,OAAO,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,eAAe,EAAE;YACxC,IAAI;YACJ,GAAG,OAAO;YACV,OAAO,EAAE,IAAA,sBAAY,EAAC,CAAC,EAAE,MAAM,EAAE,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC;YACjF,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,KAAK;YAC5B,gBAAgB,EAAE,IAAI;SACvB,CAAsE,CAAC;IAC1E,CAAC;CACF;AAxCD,sBAwCC;AAgHD,KAAK,CAAC,cAAc,GAAG,+BAAc,CAAC;AACtC,KAAK,CAAC,YAAY,GAAG,2BAAY,CAAC"}
1
+ {"version":3,"file":"audio.js","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":";AAAA,sFAAsF;;;;AAEtF,qDAAkD;AAElD,+EAAsD;AACtD,wDAA0G;AAC1G,2EAAkD;AAClD,oDAAkG;AAGlG,uDAAsD;AAGtD,MAAa,KAAM,SAAQ,sBAAW;IAAtC;;QACE,mBAAc,GAAqC,IAAI,iBAAiB,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACtG,iBAAY,GAAiC,IAAI,eAAe,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAsC9F,CAAC;IAZC,MAAM,CACJ,IAAuB,EACvB,OAAwB;QAExB,OAAO,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,eAAe,EAAE;YACxC,IAAI;YACJ,GAAG,OAAO;YACV,OAAO,EAAE,IAAA,sBAAY,EAAC,CAAC,EAAE,MAAM,EAAE,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC;YACjF,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,KAAK;YAC5B,gBAAgB,EAAE,IAAI;SACvB,CAAsE,CAAC;IAC1E,CAAC;CACF;AAxCD,sBAwCC;AA4HD,KAAK,CAAC,cAAc,GAAG,+BAAc,CAAC;AACtC,KAAK,CAAC,YAAY,GAAG,2BAAY,CAAC"}
package/resources/audio/audio.mjs.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"audio.mjs","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"AAAA,sFAAsF;OAE/E,EAAE,WAAW,EAAE;OAEf,KAAK,iBAAiB;OACtB,EAA0D,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAsD,YAAY,EAAE;OAGpE,EAAE,YAAY,EAAE;AAGvB,MAAM,OAAO,KAAM,SAAQ,WAAW;IAAtC;;QACE,mBAAc,GAAqC,IAAI,iBAAiB,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACtG,iBAAY,GAAiC,IAAI,eAAe,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAsC9F,CAAC;IAZC,MAAM,CACJ,IAAuB,EACvB,OAAwB;QAExB,OAAO,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,eAAe,EAAE;YACxC,IAAI;YACJ,GAAG,OAAO;YACV,OAAO,EAAE,YAAY,CAAC,CAAC,EAAE,MAAM,EAAE,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC;YACjF,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,KAAK;YAC5B,gBAAgB,EAAE,IAAI;SACvB,CAAsE,CAAC;IAC1E,CAAC;CACF;AAgHD,KAAK,CAAC,cAAc,GAAG,cAAc,CAAC;AACtC,KAAK,CAAC,YAAY,GAAG,YAAY,CAAC"}
1
+ {"version":3,"file":"audio.mjs","sourceRoot":"","sources":["../../src/resources/audio/audio.ts"],"names":[],"mappings":"AAAA,sFAAsF;OAE/E,EAAE,WAAW,EAAE;OAEf,KAAK,iBAAiB;OACtB,EAA0D,cAAc,EAAE;OAC1E,KAAK,eAAe;OACpB,EAAsD,YAAY,EAAE;OAGpE,EAAE,YAAY,EAAE;AAGvB,MAAM,OAAO,KAAM,SAAQ,WAAW;IAAtC;;QACE,mBAAc,GAAqC,IAAI,iBAAiB,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACtG,iBAAY,GAAiC,IAAI,eAAe,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAsC9F,CAAC;IAZC,MAAM,CACJ,IAAuB,EACvB,OAAwB;QAExB,OAAO,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,eAAe,EAAE;YACxC,IAAI;YACJ,GAAG,OAAO;YACV,OAAO,EAAE,YAAY,CAAC,CAAC,EAAE,MAAM,EAAE,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC;YACjF,MAAM,EAAE,IAAI,CAAC,MAAM,IAAI,KAAK;YAC5B,gBAAgB,EAAE,IAAI;SACvB,CAAsE,CAAC;IAC1E,CAAC;CACF;AA4HD,KAAK,CAAC,cAAc,GAAG,cAAc,CAAC;AACtC,KAAK,CAAC,YAAY,GAAG,YAAY,CAAC"}
package/resources/audio/transcriptions.d.mts CHANGED
@@ -45,6 +45,10 @@ export declare namespace TranscriptionCreateResponse {
45
45
  * The transcribed text
46
46
  */
47
47
  text: string;
48
+ /**
49
+ * Array of transcription speaker segments (only when diarize is enabled)
50
+ */
51
+ speaker_segments?: Array<AudioTranscriptionVerboseJsonResponse.SpeakerSegment>;
48
52
  /**
49
53
  * Array of transcription words (only when timestamp_granularities includes 'word')
50
54
  */
@@ -69,6 +73,52 @@ export declare namespace TranscriptionCreateResponse {
69
73
  */
70
74
  text: string;
71
75
  }
76
+ interface SpeakerSegment {
77
+ /**
78
+ * Unique identifier for the speaker segment
79
+ */
80
+ id: number;
81
+ /**
82
+ * End time of the speaker segment in seconds
83
+ */
84
+ end: number;
85
+ /**
86
+ * The speaker identifier
87
+ */
88
+ speaker_id: string;
89
+ /**
90
+ * Start time of the speaker segment in seconds
91
+ */
92
+ start: number;
93
+ /**
94
+ * The full text spoken by this speaker in this segment
95
+ */
96
+ text: string;
97
+ /**
98
+ * Array of words spoken by this speaker in this segment
99
+ */
100
+ words: Array<SpeakerSegment.Word>;
101
+ }
102
+ namespace SpeakerSegment {
103
+ interface Word {
104
+ /**
105
+ * End time of the word in seconds
106
+ */
107
+ end: number;
108
+ /**
109
+ * Start time of the word in seconds
110
+ */
111
+ start: number;
112
+ /**
113
+ * The word
114
+ */
115
+ word: string;
116
+ /**
117
+ * The speaker id for the word (only when diarize is enabled)
118
+ */
119
+ speaker_id?: string;
120
+ }
121
+ }
72
122
  interface Word {
73
123
  /**
74
124
  * End time of the word in seconds
@@ -82,6 +132,10 @@ export declare namespace TranscriptionCreateResponse {
82
132
  * The word
83
133
  */
84
134
  word: string;
135
+ /**
136
+ * The speaker id for the word (only when diarize is enabled)
137
+ */
138
+ speaker_id?: string;
85
139
  }
86
140
  }
87
141
  }
@@ -90,6 +144,19 @@ export interface TranscriptionCreateParams {
90
144
  * Audio file to transcribe
91
145
  */
92
146
  file: Uploadable;
147
+ /**
148
+ * Whether to enable speaker diarization. When enabled, you will get the speaker id
149
+ * for each word in the transcription. In the response, in the words array, you
150
+ * will get the speaker id for each word. In addition, we also return the
151
+ * speaker_segments array which contains the speaker id for each speaker segment
152
+ * along with the start and end time of the segment along with all the words in the
153
+ * segment.
154
+ *
155
+ * For eg - ... "speaker_segments": [ "speaker_id": "SPEAKER_00", "start": 0,
156
+ * "end": 30.02, "words": [ { "id": 0, "word": "Tijana", "start": 0, "end": 11.475,
157
+ * "speaker_id": "SPEAKER_00" }, ...
158
+ */
159
+ diarize?: boolean;
93
160
  /**
94
161
  * Optional ISO 639-1 language code. If `auto` is provided, language is
95
162
  * auto-detected.
package/resources/audio/transcriptions.d.mts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"transcriptions.d.mts","sourceRoot":"","sources":["../../src/resources/audio/transcriptions.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,cAAe,SAAQ,WAAW;IAC7C;;;;;;;;;;OAUG;IACH,MAAM,CAAC,IAAI,EAAE,yBAAyB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,2BAA2B,CAAC;CAM3G;AAED,MAAM,MAAM,2BAA2B,GACnC,2BAA2B,CAAC,8BAA8B,GAC1D,2BAA2B,CAAC,qCAAqC,CAAC;AAEtE,yBAAiB,2BAA2B,CAAC;IAC3C,UAAiB,8BAA8B;QAC7C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,qCAAqC;QACpD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,qCAAqC,CAAC,OAAO,CAAC,CAAC;QAE/D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,IAAI,CAAC,CAAC;KAC3D;IAED,UAAiB,qCAAqC,CAAC;QACrD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;KACF;CACF;AAED,MAAM,WAAW,yBAAyB;IACxC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,cAAc,CAAC;IACtC,OAAO,EACL,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;CACH"}
1
+ {"version":3,"file":"transcriptions.d.mts","sourceRoot":"","sources":["../../src/resources/audio/transcriptions.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,cAAe,SAAQ,WAAW;IAC7C;;;;;;;;;;OAUG;IACH,MAAM,CAAC,IAAI,EAAE,yBAAyB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,2BAA2B,CAAC;CAM3G;AAED,MAAM,MAAM,2BAA2B,GACnC,2BAA2B,CAAC,8BAA8B,GAC1D,2BAA2B,CAAC,qCAAqC,CAAC;AAEtE,yBAAiB,2BAA2B,CAAC;IAC3C,UAAiB,8BAA8B;QAC7C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,qCAAqC;QACpD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,qCAAqC,CAAC,OAAO,CAAC,CAAC;QAE/D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,gBAAgB,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,cAAc,CAAC,CAAC;QAE/E;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,IAAI,CAAC,CAAC;KAC3D;IAED,UAAiB,qCAAqC,CAAC;QACrD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,cAAc;YAC7B;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,UAAU,EAAE,MAAM,CAAC;YAEnB;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,KAAK,EAAE,KAAK,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC;SACnC;QAED,UAAiB,cAAc,CAAC;YAC9B,UAAiB,IAAI;gBACnB;;mBAEG;gBACH,GAAG,EAAE,MAAM,CAAC;gBAEZ;;mBAEG;gBACH,KAAK,EAAE,MAAM,CAAC;gBAEd;;mBAEG;gBACH,IAAI,EAAE,MAAM,CAAC;gBAEb;;mBAEG;gBACH,UAAU,CAAC,EAAE,MAAM,CAAC;aACrB;SACF;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,UAAU,CAAC,EAAE,MAAM,CAAC;SACrB;KACF;CACF;AAED,MAAM,WAAW,yBAAyB;IACxC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAElB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,cAAc,CAAC;IACtC,OAAO,EACL,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;CACH"}
package/resources/audio/transcriptions.d.ts CHANGED
@@ -45,6 +45,10 @@ export declare namespace TranscriptionCreateResponse {
45
45
  * The transcribed text
46
46
  */
47
47
  text: string;
48
+ /**
49
+ * Array of transcription speaker segments (only when diarize is enabled)
50
+ */
51
+ speaker_segments?: Array<AudioTranscriptionVerboseJsonResponse.SpeakerSegment>;
48
52
  /**
49
53
  * Array of transcription words (only when timestamp_granularities includes 'word')
50
54
  */
@@ -69,6 +73,52 @@ export declare namespace TranscriptionCreateResponse {
69
73
  */
70
74
  text: string;
71
75
  }
76
+ interface SpeakerSegment {
77
+ /**
78
+ * Unique identifier for the speaker segment
79
+ */
80
+ id: number;
81
+ /**
82
+ * End time of the speaker segment in seconds
83
+ */
84
+ end: number;
85
+ /**
86
+ * The speaker identifier
87
+ */
88
+ speaker_id: string;
89
+ /**
90
+ * Start time of the speaker segment in seconds
91
+ */
92
+ start: number;
93
+ /**
94
+ * The full text spoken by this speaker in this segment
95
+ */
96
+ text: string;
97
+ /**
98
+ * Array of words spoken by this speaker in this segment
99
+ */
100
+ words: Array<SpeakerSegment.Word>;
101
+ }
102
+ namespace SpeakerSegment {
103
+ interface Word {
104
+ /**
105
+ * End time of the word in seconds
106
+ */
107
+ end: number;
108
+ /**
109
+ * Start time of the word in seconds
110
+ */
111
+ start: number;
112
+ /**
113
+ * The word
114
+ */
115
+ word: string;
116
+ /**
117
+ * The speaker id for the word (only when diarize is enabled)
118
+ */
119
+ speaker_id?: string;
120
+ }
121
+ }
72
122
  interface Word {
73
123
  /**
74
124
  * End time of the word in seconds
@@ -82,6 +132,10 @@ export declare namespace TranscriptionCreateResponse {
82
132
  * The word
83
133
  */
84
134
  word: string;
135
+ /**
136
+ * The speaker id for the word (only when diarize is enabled)
137
+ */
138
+ speaker_id?: string;
85
139
  }
86
140
  }
87
141
  }
@@ -90,6 +144,19 @@ export interface TranscriptionCreateParams {
90
144
  * Audio file to transcribe
91
145
  */
92
146
  file: Uploadable;
147
+ /**
148
+ * Whether to enable speaker diarization. When enabled, you will get the speaker id
149
+ * for each word in the transcription. In the response, in the words array, you
150
+ * will get the speaker id for each word. In addition, we also return the
151
+ * speaker_segments array which contains the speaker id for each speaker segment
152
+ * along with the start and end time of the segment along with all the words in the
153
+ * segment.
154
+ *
155
+ * For eg - ... "speaker_segments": [ "speaker_id": "SPEAKER_00", "start": 0,
156
+ * "end": 30.02, "words": [ { "id": 0, "word": "Tijana", "start": 0, "end": 11.475,
157
+ * "speaker_id": "SPEAKER_00" }, ...
158
+ */
159
+ diarize?: boolean;
93
160
  /**
94
161
  * Optional ISO 639-1 language code. If `auto` is provided, language is
95
162
  * auto-detected.
package/resources/audio/transcriptions.d.ts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"transcriptions.d.ts","sourceRoot":"","sources":["../../src/resources/audio/transcriptions.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,cAAe,SAAQ,WAAW;IAC7C;;;;;;;;;;OAUG;IACH,MAAM,CAAC,IAAI,EAAE,yBAAyB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,2BAA2B,CAAC;CAM3G;AAED,MAAM,MAAM,2BAA2B,GACnC,2BAA2B,CAAC,8BAA8B,GAC1D,2BAA2B,CAAC,qCAAqC,CAAC;AAEtE,yBAAiB,2BAA2B,CAAC;IAC3C,UAAiB,8BAA8B;QAC7C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,qCAAqC;QACpD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,qCAAqC,CAAC,OAAO,CAAC,CAAC;QAE/D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,IAAI,CAAC,CAAC;KAC3D;IAED,UAAiB,qCAAqC,CAAC;QACrD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;KACF;CACF;AAED,MAAM,WAAW,yBAAyB;IACxC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,cAAc,CAAC;IACtC,OAAO,EACL,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;CACH"}
1
+ {"version":3,"file":"transcriptions.d.ts","sourceRoot":"","sources":["../../src/resources/audio/transcriptions.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,cAAe,SAAQ,WAAW;IAC7C;;;;;;;;;;OAUG;IACH,MAAM,CAAC,IAAI,EAAE,yBAAyB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,2BAA2B,CAAC;CAM3G;AAED,MAAM,MAAM,2BAA2B,GACnC,2BAA2B,CAAC,8BAA8B,GAC1D,2BAA2B,CAAC,qCAAqC,CAAC;AAEtE,yBAAiB,2BAA2B,CAAC;IAC3C,UAAiB,8BAA8B;QAC7C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,qCAAqC;QACpD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,qCAAqC,CAAC,OAAO,CAAC,CAAC;QAE/D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,gBAAgB,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,cAAc,CAAC,CAAC;QAE/E;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,qCAAqC,CAAC,IAAI,CAAC,CAAC;KAC3D;IAED,UAAiB,qCAAqC,CAAC;QACrD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,cAAc;YAC7B;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,UAAU,EAAE,MAAM,CAAC;YAEnB;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,KAAK,EAAE,KAAK,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC;SACnC;QAED,UAAiB,cAAc,CAAC;YAC9B,UAAiB,IAAI;gBACnB;;mBAEG;gBACH,GAAG,EAAE,MAAM,CAAC;gBAEZ;;mBAEG;gBACH,KAAK,EAAE,MAAM,CAAC;gBAEd;;mBAEG;gBACH,IAAI,EAAE,MAAM,CAAC;gBAEb;;mBAEG;gBACH,UAAU,CAAC,EAAE,MAAM,CAAC;aACrB;SACF;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,UAAU,CAAC,EAAE,MAAM,CAAC;SACrB;KACF;CACF;AAED,MAAM,WAAW,yBAAyB;IACxC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;;;;;;;;;OAWG;IACH,OAAO,CAAC,EAAE,OAAO,CAAC;IAElB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,cAAc,CAAC;IACtC,OAAO,EACL,KAAK,2BAA2B,IAAI,2BAA2B,EAC/D,KAAK,yBAAyB,IAAI,yBAAyB,GAC5D,CAAC;CACH"}
package/resources/audio/translations.d.mts CHANGED
@@ -81,6 +81,10 @@ export declare namespace TranslationCreateResponse {
81
81
  * The word
82
82
  */
83
83
  word: string;
84
+ /**
85
+ * The speaker id for the word (only when diarize is enabled)
86
+ */
87
+ speaker_id?: string;
84
88
  }
85
89
  }
86
90
  }
package/resources/audio/translations.d.mts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"translations.d.mts","sourceRoot":"","sources":["../../src/resources/audio/translations.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,YAAa,SAAQ,WAAW;IAC3C;;;;;;;;;OASG;IACH,MAAM,CAAC,IAAI,EAAE,uBAAuB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,yBAAyB,CAAC;CAMvG;AAED,MAAM,MAAM,yBAAyB,GACjC,yBAAyB,CAAC,4BAA4B,GACtD,yBAAyB,CAAC,mCAAmC,CAAC;AAElE,yBAAiB,yBAAyB,CAAC;IACzC,UAAiB,4BAA4B;QAC3C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,mCAAmC;QAClD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,mCAAmC,CAAC,OAAO,CAAC,CAAC;QAE7D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,mCAAmC,CAAC,IAAI,CAAC,CAAC;KACzD;IAED,UAAiB,mCAAmC,CAAC;QACnD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;KACF;CACF;AAED,MAAM,WAAW,uBAAuB;IACtC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,YAAY,CAAC;IACpC,OAAO,EACL,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
1
+ {"version":3,"file":"translations.d.mts","sourceRoot":"","sources":["../../src/resources/audio/translations.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,YAAa,SAAQ,WAAW;IAC3C;;;;;;;;;OASG;IACH,MAAM,CAAC,IAAI,EAAE,uBAAuB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,yBAAyB,CAAC;CAMvG;AAED,MAAM,MAAM,yBAAyB,GACjC,yBAAyB,CAAC,4BAA4B,GACtD,yBAAyB,CAAC,mCAAmC,CAAC;AAElE,yBAAiB,yBAAyB,CAAC;IACzC,UAAiB,4BAA4B;QAC3C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,mCAAmC;QAClD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,mCAAmC,CAAC,OAAO,CAAC,CAAC;QAE7D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,mCAAmC,CAAC,IAAI,CAAC,CAAC;KACzD;IAED,UAAiB,mCAAmC,CAAC;QACnD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,UAAU,CAAC,EAAE,MAAM,CAAC;SACrB;KACF;CACF;AAED,MAAM,WAAW,uBAAuB;IACtC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,YAAY,CAAC;IACpC,OAAO,EACL,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
package/resources/audio/translations.d.ts CHANGED
@@ -81,6 +81,10 @@ export declare namespace TranslationCreateResponse {
81
81
  * The word
82
82
  */
83
83
  word: string;
84
+ /**
85
+ * The speaker id for the word (only when diarize is enabled)
86
+ */
87
+ speaker_id?: string;
84
88
  }
85
89
  }
86
90
  }
package/resources/audio/translations.d.ts.map CHANGED
@@ -1 +1 @@
1
- {"version":3,"file":"translations.d.ts","sourceRoot":"","sources":["../../src/resources/audio/translations.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,YAAa,SAAQ,WAAW;IAC3C;;;;;;;;;OASG;IACH,MAAM,CAAC,IAAI,EAAE,uBAAuB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,yBAAyB,CAAC;CAMvG;AAED,MAAM,MAAM,yBAAyB,GACjC,yBAAyB,CAAC,4BAA4B,GACtD,yBAAyB,CAAC,mCAAmC,CAAC;AAElE,yBAAiB,yBAAyB,CAAC;IACzC,UAAiB,4BAA4B;QAC3C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,mCAAmC;QAClD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,mCAAmC,CAAC,OAAO,CAAC,CAAC;QAE7D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,mCAAmC,CAAC,IAAI,CAAC,CAAC;KACzD;IAED,UAAiB,mCAAmC,CAAC;QACnD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;KACF;CACF;AAED,MAAM,WAAW,uBAAuB;IACtC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,YAAY,CAAC;IACpC,OAAO,EACL,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
1
+ {"version":3,"file":"translations.d.ts","sourceRoot":"","sources":["../../src/resources/audio/translations.ts"],"names":[],"mappings":"OAEO,EAAE,WAAW,EAAE;OACf,EAAE,UAAU,EAAE;OACd,EAAE,KAAK,UAAU,EAAE;OACnB,EAAE,cAAc,EAAE;AAGzB,qBAAa,YAAa,SAAQ,WAAW;IAC3C;;;;;;;;;OASG;IACH,MAAM,CAAC,IAAI,EAAE,uBAAuB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,UAAU,CAAC,yBAAyB,CAAC;CAMvG;AAED,MAAM,MAAM,yBAAyB,GACjC,yBAAyB,CAAC,4BAA4B,GACtD,yBAAyB,CAAC,mCAAmC,CAAC;AAElE,yBAAiB,yBAAyB,CAAC;IACzC,UAAiB,4BAA4B;QAC3C;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;KACd;IAED,UAAiB,mCAAmC;QAClD;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,MAAM,CAAC;QAEjB;;WAEG;QACH,QAAQ,EAAE,KAAK,CAAC,mCAAmC,CAAC,OAAO,CAAC,CAAC;QAE7D;;WAEG;QACH,IAAI,EAAE,YAAY,GAAG,WAAW,CAAC;QAEjC;;WAEG;QACH,IAAI,EAAE,MAAM,CAAC;QAEb;;WAEG;QACH,KAAK,CAAC,EAAE,KAAK,CAAC,mCAAmC,CAAC,IAAI,CAAC,CAAC;KACzD;IAED,UAAiB,mCAAmC,CAAC;QACnD,UAAiB,OAAO;YACtB;;eAEG;YACH,EAAE,EAAE,MAAM,CAAC;YAEX;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;SACd;QAED,UAAiB,IAAI;YACnB;;eAEG;YACH,GAAG,EAAE,MAAM,CAAC;YAEZ;;eAEG;YACH,KAAK,EAAE,MAAM,CAAC;YAEd;;eAEG;YACH,IAAI,EAAE,MAAM,CAAC;YAEb;;eAEG;YACH,UAAU,CAAC,EAAE,MAAM,CAAC;SACrB;KACF;CACF;AAED,MAAM,WAAW,uBAAuB;IACtC;;OAEG;IACH,IAAI,EAAE,UAAU,CAAC;IAEjB;;;OAGG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAElB;;OAEG;IACH,KAAK,CAAC,EAAE,yBAAyB,CAAC;IAElC;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAC;IAEhB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,GAAG,cAAc,CAAC;IAE1C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;;;OAIG;IACH,uBAAuB,CAAC,EAAE,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,SAAS,GAAG,MAAM,CAAC,CAAC;CAC1E;AAED,MAAM,CAAC,OAAO,WAAW,YAAY,CAAC;IACpC,OAAO,EACL,KAAK,yBAAyB,IAAI,yBAAyB,EAC3D,KAAK,uBAAuB,IAAI,uBAAuB,GACxD,CAAC;CACH"}
package/src/resources/audio/audio.ts CHANGED
@@ -22,8 +22,8 @@ export class Audio extends APIResource {
22
22
  * ```ts
23
23
  * const audio = await client.audio.create({
24
24
  * input: 'input',
25
- * model: 'cartesia/sonic',
26
- * voice: 'laidback woman',
25
+ * model: 'canopylabs/orpheus-3b-0.1-ft',
26
+ * voice: 'voice',
27
27
  * });
28
28
  *
29
29
  * const content = await audio.blob();
@@ -88,17 +88,26 @@ export interface AudioCreateParamsBase {
88
88
  * The name of the model to query.
89
89
  *
90
90
  * [See all of Together AI's chat models](https://docs.together.ai/docs/serverless-models#audio-models)
91
+ * The current supported tts models are: - cartesia/sonic - hexgrad/Kokoro-82M -
92
+ * canopylabs/orpheus-3b-0.1-ft
91
93
  */
92
- model: 'cartesia/sonic' | (string & {});
94
+ model: 'cartesia/sonic' | 'hexgrad/Kokoro-82M' | 'canopylabs/orpheus-3b-0.1-ft' | (string & {});
93
95
 
94
96
  /**
95
- * The voice to use for generating the audio.
97
+ * The voice to use for generating the audio. The voices supported are different
98
+ * for each model. For eg - for canopylabs/orpheus-3b-0.1-ft, one of the voices
99
+ * supported is tara, for hexgrad/Kokoro-82M, one of the voices supported is
100
+ * af_alloy and for cartesia/sonic, one of the voices supported is "friendly
101
+ * sidekick".
102
+ *
103
+ * You can view the voices supported for each model using the /v1/voices endpoint
104
+ * sending the model name as the query parameter.
96
105
  * [View all supported voices here](https://docs.together.ai/docs/text-to-speech#voices-available).
97
106
  */
98
- voice: 'laidback woman' | 'polite man' | 'storyteller lady' | 'friendly sidekick' | (string & {});
107
+ voice: string;
99
108
 
100
109
  /**
101
- * Language of input text
110
+ * Language of input text.
102
111
  */
103
112
  language?:
104
113
  | 'en'
@@ -123,12 +132,15 @@ export interface AudioCreateParamsBase {
123
132
  response_encoding?: 'pcm_f32le' | 'pcm_s16le' | 'pcm_mulaw' | 'pcm_alaw';
124
133
 
125
134
  /**
126
- * The format of audio output
135
+ * The format of audio output. Supported formats are mp3, wav, raw if streaming is
136
+ * false. If streaming is true, the only supported format is raw.
127
137
  */
128
138
  response_format?: 'mp3' | 'wav' | 'raw';
129
139
 
130
140
  /**
131
- * Sampling rate to use for the output audio
141
+ * Sampling rate to use for the output audio. The default sampling rate for
142
+ * canopylabs/orpheus-3b-0.1-ft and hexgrad/Kokoro-82M is 24000 and for
143
+ * cartesia/sonic is 44100.
132
144
  */
133
145
  sample_rate?: number;
134
146
 
package/src/resources/audio/transcriptions.ts CHANGED
@@ -64,6 +64,11 @@ export namespace TranscriptionCreateResponse {
64
64
  */
65
65
  text: string;
66
66
 
67
+ /**
68
+ * Array of transcription speaker segments (only when diarize is enabled)
69
+ */
70
+ speaker_segments?: Array<AudioTranscriptionVerboseJsonResponse.SpeakerSegment>;
71
+
67
72
  /**
68
73
  * Array of transcription words (only when timestamp_granularities includes 'word')
69
74
  */
@@ -93,6 +98,62 @@ export namespace TranscriptionCreateResponse {
93
98
  text: string;
94
99
  }
95
100
 
101
+ export interface SpeakerSegment {
102
+ /**
103
+ * Unique identifier for the speaker segment
104
+ */
105
+ id: number;
106
+
107
+ /**
108
+ * End time of the speaker segment in seconds
109
+ */
110
+ end: number;
111
+
112
+ /**
113
+ * The speaker identifier
114
+ */
115
+ speaker_id: string;
116
+
117
+ /**
118
+ * Start time of the speaker segment in seconds
119
+ */
120
+ start: number;
121
+
122
+ /**
123
+ * The full text spoken by this speaker in this segment
124
+ */
125
+ text: string;
126
+
127
+ /**
128
+ * Array of words spoken by this speaker in this segment
129
+ */
130
+ words: Array<SpeakerSegment.Word>;
131
+ }
132
+
133
+ export namespace SpeakerSegment {
134
+ export interface Word {
135
+ /**
136
+ * End time of the word in seconds
137
+ */
138
+ end: number;
139
+
140
+ /**
141
+ * Start time of the word in seconds
142
+ */
143
+ start: number;
144
+
145
+ /**
146
+ * The word
147
+ */
148
+ word: string;
149
+
150
+ /**
151
+ * The speaker id for the word (only when diarize is enabled)
152
+ */
153
+ speaker_id?: string;
154
+ }
155
+ }
156
+
96
157
  export interface Word {
97
158
  /**
98
159
  * End time of the word in seconds
@@ -108,6 +169,11 @@ export namespace TranscriptionCreateResponse {
108
169
  * The word
109
170
  */
110
171
  word: string;
172
+
173
+ /**
174
+ * The speaker id for the word (only when diarize is enabled)
175
+ */
176
+ speaker_id?: string;
111
177
  }
112
178
  }
113
179
  }
@@ -118,6 +184,20 @@ export interface TranscriptionCreateParams {
118
184
  */
119
185
  file: Uploadable;
120
186
 
187
+ /**
188
+ * Whether to enable speaker diarization. When enabled, you will get the speaker id
189
+ * for each word in the transcription. In the response, in the words array, you
190
+ * will get the speaker id for each word. In addition, we also return the
191
+ * speaker_segments array which contains the speaker id for each speaker segment
192
+ * along with the start and end time of the segment along with all the words in the
193
+ * segment.
194
+ *
195
+ * For eg - ... "speaker_segments": [ "speaker_id": "SPEAKER_00", "start": 0,
196
+ * "end": 30.02, "words": [ { "id": 0, "word": "Tijana", "start": 0, "end": 11.475,
197
+ * "speaker_id": "SPEAKER_00" }, ...
198
+ */
199
+ diarize?: boolean;
200
+
121
201
  /**
122
202
  * Optional ISO 639-1 language code. If `auto` is provided, language is
123
203
  * auto-detected.
package/src/resources/audio/translations.ts CHANGED
@@ -107,6 +107,11 @@ export namespace TranslationCreateResponse {
107
107
  * The word
108
108
  */
109
109
  word: string;
110
+
111
+ /**
112
+ * The speaker id for the word (only when diarize is enabled)
113
+ */
114
+ speaker_id?: string;
110
115
  }
111
116
  }
112
117
  }
package/src/version.ts CHANGED
@@ -1 +1 @@
1
- export const VERSION = '0.28.0'; // x-release-please-version
1
+ export const VERSION = '0.29.0'; // x-release-please-version
package/version.d.mts CHANGED
@@ -1,2 +1,2 @@
1
- export declare const VERSION = "0.28.0";
1
+ export declare const VERSION = "0.29.0";
2
2
  //# sourceMappingURL=version.d.mts.map
package/version.d.ts CHANGED
@@ -1,2 +1,2 @@
1
- export declare const VERSION = "0.28.0";
1
+ export declare const VERSION = "0.29.0";
2
2
  //# sourceMappingURL=version.d.ts.map
package/version.js CHANGED
@@ -1,5 +1,5 @@
1
1
  "use strict";
2
2
  Object.defineProperty(exports, "__esModule", { value: true });
3
3
  exports.VERSION = void 0;
4
- exports.VERSION = '0.28.0'; // x-release-please-version
4
+ exports.VERSION = '0.29.0'; // x-release-please-version
5
5
  //# sourceMappingURL=version.js.map
package/version.mjs CHANGED
@@ -1,2 +1,2 @@
1
- export const VERSION = '0.28.0'; // x-release-please-version
1
+ export const VERSION = '0.29.0'; // x-release-please-version
2
2
  //# sourceMappingURL=version.mjs.map