@ai-sdk/hume 3.0.0-beta.3 → 3.0.0-beta.31
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 +238 -0
- package/README.md +2 -0
- package/dist/index.d.ts +44 -34
- package/dist/index.js +86 -68
- package/dist/index.js.map +1 -1
- package/package.json +11 -11
- package/src/hume-config.ts +2 -2
- package/src/hume-provider.ts +34 -5
- package/src/hume-speech-model-options.ts +77 -0
- package/src/hume-speech-model.ts +28 -87
- package/src/index.ts +1 -1
- package/dist/index.d.mts +0 -92
- package/dist/index.mjs +0 -255
- package/dist/index.mjs.map +0 -1
package/dist/index.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/index.ts","../src/hume-provider.ts","../src/hume-speech-model.ts","../src/hume-error.ts","../src/version.ts"],"sourcesContent":["export { createHume, hume } from './hume-provider';\nexport type { HumeProvider, HumeProviderSettings } from './hume-provider';\nexport type { HumeSpeechModelOptions } from './hume-speech-model';\nexport { VERSION } from './version';\n","import { SpeechModelV3, ProviderV3 } from '@ai-sdk/provider';\nimport {\n FetchFunction,\n loadApiKey,\n withUserAgentSuffix,\n} from '@ai-sdk/provider-utils';\nimport { HumeSpeechModel } from './hume-speech-model';\nimport { VERSION } from './version';\n\nexport interface HumeProvider extends Pick<ProviderV3, 'speechModel'> {\n (settings?: {}): {\n speech: HumeSpeechModel;\n };\n\n /**\n * Creates a model for speech synthesis.\n */\n speech(): SpeechModelV3;\n}\n\nexport interface HumeProviderSettings {\n /**\n * API key for authenticating requests.\n */\n apiKey?: string;\n\n /**\n * Custom headers to include in the requests.\n */\n headers?: Record<string, string>;\n\n /**\n * Custom fetch implementation. You can use it as a middleware to intercept requests,\n * or to provide a custom fetch implementation for e.g. testing.\n */\n fetch?: FetchFunction;\n}\n\n/**\n * Create an Hume provider instance.\n */\nexport function createHume(options: HumeProviderSettings = {}): HumeProvider {\n const getHeaders = () =>\n withUserAgentSuffix(\n {\n 'X-Hume-Api-Key': loadApiKey({\n apiKey: options.apiKey,\n environmentVariableName: 'HUME_API_KEY',\n description: 'Hume',\n }),\n ...options.headers,\n },\n `ai-sdk/hume/${VERSION}`,\n );\n\n const createSpeechModel = () =>\n new HumeSpeechModel('', {\n provider: `hume.speech`,\n url: ({ path }) => `https://api.hume.ai${path}`,\n headers: getHeaders,\n fetch: options.fetch,\n });\n\n const provider = function () {\n return {\n speech: createSpeechModel(),\n };\n };\n\n provider.speech = createSpeechModel;\n provider.speechModel = createSpeechModel;\n\n return provider satisfies HumeProvider;\n}\n\n/**\n * Default Hume provider instance.\n */\nexport const hume = createHume();\n","import { SpeechModelV3, SharedV3Warning } from '@ai-sdk/provider';\nimport {\n combineHeaders,\n createBinaryResponseHandler,\n parseProviderOptions,\n postJsonToApi,\n} from '@ai-sdk/provider-utils';\nimport { z } from 'zod/v4';\nimport { HumeConfig } from './hume-config';\nimport { humeFailedResponseHandler } from './hume-error';\nimport { HumeSpeechAPITypes } from './hume-api-types';\n\n// https://dev.hume.ai/reference/text-to-speech-tts/synthesize-file\nconst humeSpeechModelOptionsSchema = z.object({\n /**\n * Context for the speech synthesis request.\n * Can be either a generationId for retrieving a previous generation,\n * or a list of utterances to synthesize.\n */\n context: z\n .object({\n /**\n * ID of a previously generated speech synthesis to retrieve.\n */\n generationId: z.string(),\n })\n .or(\n z.object({\n /**\n * List of utterances to synthesize into speech.\n */\n utterances: z.array(\n z.object({\n /**\n * The text content to convert to speech.\n */\n text: z.string(),\n /**\n * Optional description or instructions for how the text should be spoken.\n */\n description: z.string().optional(),\n /**\n * Optional speech rate multiplier.\n */\n speed: z.number().optional(),\n /**\n * Optional duration of silence to add after the utterance in seconds.\n */\n trailingSilence: z.number().optional(),\n /**\n * Voice configuration for the utterance.\n * Can be specified by ID or name.\n */\n voice: z\n .object({\n /**\n * ID of the voice to use.\n */\n id: z.string(),\n /**\n * Provider of the voice, either Hume's built-in voices or a custom voice.\n */\n provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),\n })\n .or(\n z.object({\n /**\n * Name of the voice to use.\n */\n name: z.string(),\n /**\n * Provider of the voice, either Hume's built-in voices or a custom voice.\n */\n provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),\n }),\n )\n .optional(),\n }),\n ),\n }),\n )\n .nullish(),\n});\n\nexport type HumeSpeechModelOptions = z.infer<\n typeof humeSpeechModelOptionsSchema\n>;\n\ninterface HumeSpeechModelConfig extends HumeConfig {\n _internal?: {\n currentDate?: () => Date;\n };\n}\n\nexport class HumeSpeechModel implements SpeechModelV3 {\n readonly specificationVersion = 'v3';\n\n get provider(): string {\n return this.config.provider;\n }\n\n constructor(\n readonly modelId: '',\n private readonly config: HumeSpeechModelConfig,\n ) {}\n\n private async getArgs({\n text,\n voice = 'd8ab67c6-953d-4bd8-9370-8fa53a0f1453',\n outputFormat = 'mp3',\n speed,\n instructions,\n language,\n providerOptions,\n }: Parameters<SpeechModelV3['doGenerate']>[0]) {\n const warnings: SharedV3Warning[] = [];\n\n // Parse provider options\n const humeOptions = await parseProviderOptions({\n provider: 'hume',\n providerOptions,\n schema: humeSpeechModelOptionsSchema,\n });\n\n // Create request body\n const requestBody: HumeSpeechAPITypes = {\n utterances: [\n {\n text,\n speed,\n description: instructions,\n voice: {\n id: voice,\n provider: 'HUME_AI',\n },\n },\n ],\n format: { type: 'mp3' },\n };\n\n if (outputFormat) {\n if (['mp3', 'pcm', 'wav'].includes(outputFormat)) {\n requestBody.format = { type: outputFormat as 'mp3' | 'pcm' | 'wav' };\n } else {\n warnings.push({\n type: 'unsupported',\n feature: 'outputFormat',\n details: `Unsupported output format: ${outputFormat}. Using mp3 instead.`,\n });\n }\n }\n\n // Add provider-specific options\n if (humeOptions) {\n const speechModelOptions: Omit<\n HumeSpeechAPITypes,\n 'utterances' | 'format'\n > = {};\n\n if (humeOptions.context) {\n if ('generationId' in humeOptions.context) {\n speechModelOptions.context = {\n generation_id: humeOptions.context.generationId,\n };\n } else {\n speechModelOptions.context = {\n utterances: humeOptions.context.utterances.map(utterance => ({\n text: utterance.text,\n description: utterance.description,\n speed: utterance.speed,\n trailing_silence: utterance.trailingSilence,\n voice: utterance.voice,\n })),\n };\n }\n }\n\n for (const key in speechModelOptions) {\n const value =\n speechModelOptions[\n key as keyof Omit<HumeSpeechAPITypes, 'utterances' | 'format'>\n ];\n if (value !== undefined) {\n (requestBody as Record<string, unknown>)[key] = value;\n }\n }\n }\n\n if (language) {\n warnings.push({\n type: 'unsupported',\n feature: 'language',\n details: `Hume speech models do not support language selection. Language parameter \"${language}\" was ignored.`,\n });\n }\n\n return {\n requestBody,\n warnings,\n };\n }\n\n async doGenerate(\n options: Parameters<SpeechModelV3['doGenerate']>[0],\n ): Promise<Awaited<ReturnType<SpeechModelV3['doGenerate']>>> {\n const currentDate = this.config._internal?.currentDate?.() ?? new Date();\n const { requestBody, warnings } = await this.getArgs(options);\n\n const {\n value: audio,\n responseHeaders,\n rawValue: rawResponse,\n } = await postJsonToApi({\n url: this.config.url({\n path: '/v0/tts/file',\n modelId: this.modelId,\n }),\n headers: combineHeaders(this.config.headers(), options.headers),\n body: requestBody,\n failedResponseHandler: humeFailedResponseHandler,\n successfulResponseHandler: createBinaryResponseHandler(),\n abortSignal: options.abortSignal,\n fetch: this.config.fetch,\n });\n\n return {\n audio,\n warnings,\n request: {\n body: JSON.stringify(requestBody),\n },\n response: {\n timestamp: currentDate,\n modelId: this.modelId,\n headers: responseHeaders,\n body: rawResponse,\n },\n };\n }\n}\n","import { z } from 'zod/v4';\nimport { createJsonErrorResponseHandler } from '@ai-sdk/provider-utils';\n\nexport const humeErrorDataSchema = z.object({\n error: z.object({\n message: z.string(),\n code: z.number(),\n }),\n});\n\nexport type HumeErrorData = z.infer<typeof humeErrorDataSchema>;\n\nexport const humeFailedResponseHandler = createJsonErrorResponseHandler({\n errorSchema: humeErrorDataSchema,\n errorToMessage: data => data.error.message,\n});\n","// Version string of this package injected at build time.\ndeclare const __PACKAGE_VERSION__: string | undefined;\nexport const VERSION: string =\n typeof __PACKAGE_VERSION__ !== 'undefined'\n ? __PACKAGE_VERSION__\n : '0.0.0-test';\n"],"mappings":";;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;;;ACCA,IAAAA,yBAIO;;;ACJP,IAAAC,yBAKO;AACP,IAAAC,aAAkB;;;ACPlB,gBAAkB;AAClB,4BAA+C;AAExC,IAAM,sBAAsB,YAAE,OAAO;AAAA,EAC1C,OAAO,YAAE,OAAO;AAAA,IACd,SAAS,YAAE,OAAO;AAAA,IAClB,MAAM,YAAE,OAAO;AAAA,EACjB,CAAC;AACH,CAAC;AAIM,IAAM,gCAA4B,sDAA+B;AAAA,EACtE,aAAa;AAAA,EACb,gBAAgB,UAAQ,KAAK,MAAM;AACrC,CAAC;;;ADFD,IAAM,+BAA+B,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAM5C,SAAS,aACN,OAAO;AAAA;AAAA;AAAA;AAAA,IAIN,cAAc,aAAE,OAAO;AAAA,EACzB,CAAC,EACA;AAAA,IACC,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,MAIP,YAAY,aAAE;AAAA,QACZ,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,UAIP,MAAM,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,UAIf,aAAa,aAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA,UAIjC,OAAO,aAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA,UAI3B,iBAAiB,aAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA;AAAA,UAKrC,OAAO,aACJ,OAAO;AAAA;AAAA;AAAA;AAAA,YAIN,IAAI,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,YAIb,UAAU,aAAE,KAAK,CAAC,WAAW,cAAc,CAAC,EAAE,SAAS;AAAA,UACzD,CAAC,EACA;AAAA,YACC,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,cAIP,MAAM,aAAE,OAAO;AAAA;AAAA;AAAA;AAAA,cAIf,UAAU,aAAE,KAAK,CAAC,WAAW,cAAc,CAAC,EAAE,SAAS;AAAA,YACzD,CAAC;AAAA,UACH,EACC,SAAS;AAAA,QACd,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,EACC,QAAQ;AACb,CAAC;AAYM,IAAM,kBAAN,MAA+C;AAAA,EAOpD,YACW,SACQ,QACjB;AAFS;AACQ;AARnB,SAAS,uBAAuB;AAAA,EAS7B;AAAA,EAPH,IAAI,WAAmB;AACrB,WAAO,KAAK,OAAO;AAAA,EACrB;AAAA,EAOA,MAAc,QAAQ;AAAA,IACpB;AAAA,IACA,QAAQ;AAAA,IACR,eAAe;AAAA,IACf;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,GAA+C;AAC7C,UAAM,WAA8B,CAAC;AAGrC,UAAM,cAAc,UAAM,6CAAqB;AAAA,MAC7C,UAAU;AAAA,MACV;AAAA,MACA,QAAQ;AAAA,IACV,CAAC;AAGD,UAAM,cAAkC;AAAA,MACtC,YAAY;AAAA,QACV;AAAA,UACE;AAAA,UACA;AAAA,UACA,aAAa;AAAA,UACb,OAAO;AAAA,YACL,IAAI;AAAA,YACJ,UAAU;AAAA,UACZ;AAAA,QACF;AAAA,MACF;AAAA,MACA,QAAQ,EAAE,MAAM,MAAM;AAAA,IACxB;AAEA,QAAI,cAAc;AAChB,UAAI,CAAC,OAAO,OAAO,KAAK,EAAE,SAAS,YAAY,GAAG;AAChD,oBAAY,SAAS,EAAE,MAAM,aAAsC;AAAA,MACrE,OAAO;AACL,iBAAS,KAAK;AAAA,UACZ,MAAM;AAAA,UACN,SAAS;AAAA,UACT,SAAS,8BAA8B,YAAY;AAAA,QACrD,CAAC;AAAA,MACH;AAAA,IACF;AAGA,QAAI,aAAa;AACf,YAAM,qBAGF,CAAC;AAEL,UAAI,YAAY,SAAS;AACvB,YAAI,kBAAkB,YAAY,SAAS;AACzC,6BAAmB,UAAU;AAAA,YAC3B,eAAe,YAAY,QAAQ;AAAA,UACrC;AAAA,QACF,OAAO;AACL,6BAAmB,UAAU;AAAA,YAC3B,YAAY,YAAY,QAAQ,WAAW,IAAI,gBAAc;AAAA,cAC3D,MAAM,UAAU;AAAA,cAChB,aAAa,UAAU;AAAA,cACvB,OAAO,UAAU;AAAA,cACjB,kBAAkB,UAAU;AAAA,cAC5B,OAAO,UAAU;AAAA,YACnB,EAAE;AAAA,UACJ;AAAA,QACF;AAAA,MACF;AAEA,iBAAW,OAAO,oBAAoB;AACpC,cAAM,QACJ,mBACE,GACF;AACF,YAAI,UAAU,QAAW;AACvB,UAAC,YAAwC,GAAG,IAAI;AAAA,QAClD;AAAA,MACF;AAAA,IACF;AAEA,QAAI,UAAU;AACZ,eAAS,KAAK;AAAA,QACZ,MAAM;AAAA,QACN,SAAS;AAAA,QACT,SAAS,6EAA6E,QAAQ;AAAA,MAChG,CAAC;AAAA,IACH;AAEA,WAAO;AAAA,MACL;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAM,WACJ,SAC2D;AA5M/D;AA6MI,UAAM,eAAc,sBAAK,OAAO,cAAZ,mBAAuB,gBAAvB,4CAA0C,oBAAI,KAAK;AACvE,UAAM,EAAE,aAAa,SAAS,IAAI,MAAM,KAAK,QAAQ,OAAO;AAE5D,UAAM;AAAA,MACJ,OAAO;AAAA,MACP;AAAA,MACA,UAAU;AAAA,IACZ,IAAI,UAAM,sCAAc;AAAA,MACtB,KAAK,KAAK,OAAO,IAAI;AAAA,QACnB,MAAM;AAAA,QACN,SAAS,KAAK;AAAA,MAChB,CAAC;AAAA,MACD,aAAS,uCAAe,KAAK,OAAO,QAAQ,GAAG,QAAQ,OAAO;AAAA,MAC9D,MAAM;AAAA,MACN,uBAAuB;AAAA,MACvB,+BAA2B,oDAA4B;AAAA,MACvD,aAAa,QAAQ;AAAA,MACrB,OAAO,KAAK,OAAO;AAAA,IACrB,CAAC;AAED,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA,SAAS;AAAA,QACP,MAAM,KAAK,UAAU,WAAW;AAAA,MAClC;AAAA,MACA,UAAU;AAAA,QACR,WAAW;AAAA,QACX,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,QACT,MAAM;AAAA,MACR;AAAA,IACF;AAAA,EACF;AACF;;;AE7OO,IAAM,UACX,OACI,iBACA;;;AHoCC,SAAS,WAAW,UAAgC,CAAC,GAAiB;AAC3E,QAAM,aAAa,UACjB;AAAA,IACE;AAAA,MACE,sBAAkB,mCAAW;AAAA,QAC3B,QAAQ,QAAQ;AAAA,QAChB,yBAAyB;AAAA,QACzB,aAAa;AAAA,MACf,CAAC;AAAA,MACD,GAAG,QAAQ;AAAA,IACb;AAAA,IACA,eAAe,OAAO;AAAA,EACxB;AAEF,QAAM,oBAAoB,MACxB,IAAI,gBAAgB,IAAI;AAAA,IACtB,UAAU;AAAA,IACV,KAAK,CAAC,EAAE,KAAK,MAAM,sBAAsB,IAAI;AAAA,IAC7C,SAAS;AAAA,IACT,OAAO,QAAQ;AAAA,EACjB,CAAC;AAEH,QAAM,WAAW,WAAY;AAC3B,WAAO;AAAA,MACL,QAAQ,kBAAkB;AAAA,IAC5B;AAAA,EACF;AAEA,WAAS,SAAS;AAClB,WAAS,cAAc;AAEvB,SAAO;AACT;AAKO,IAAM,OAAO,WAAW;","names":["import_provider_utils","import_provider_utils","import_v4"]}
|
|
1
|
+
{"version":3,"sources":["../src/hume-provider.ts","../src/hume-speech-model.ts","../src/hume-error.ts","../src/hume-speech-model-options.ts","../src/version.ts"],"sourcesContent":["import {\n NoSuchModelError,\n type SpeechModelV4,\n type ProviderV4,\n} from '@ai-sdk/provider';\nimport {\n loadApiKey,\n withUserAgentSuffix,\n type FetchFunction,\n} from '@ai-sdk/provider-utils';\nimport { HumeSpeechModel } from './hume-speech-model';\nimport { VERSION } from './version';\n\nexport interface HumeProvider extends ProviderV4 {\n (settings?: {}): {\n speech: HumeSpeechModel;\n };\n\n /**\n * Creates a model for speech synthesis.\n */\n speech(): SpeechModelV4;\n}\n\nexport interface HumeProviderSettings {\n /**\n * API key for authenticating requests.\n */\n apiKey?: string;\n\n /**\n * Custom headers to include in the requests.\n */\n headers?: Record<string, string>;\n\n /**\n * Custom fetch implementation. You can use it as a middleware to intercept requests,\n * or to provide a custom fetch implementation for e.g. testing.\n */\n fetch?: FetchFunction;\n}\n\n/**\n * Create an Hume provider instance.\n */\nexport function createHume(options: HumeProviderSettings = {}): HumeProvider {\n const getHeaders = () =>\n withUserAgentSuffix(\n {\n 'X-Hume-Api-Key': loadApiKey({\n apiKey: options.apiKey,\n environmentVariableName: 'HUME_API_KEY',\n description: 'Hume',\n }),\n ...options.headers,\n },\n `ai-sdk/hume/${VERSION}`,\n );\n\n const createSpeechModel = () =>\n new HumeSpeechModel('', {\n provider: `hume.speech`,\n url: ({ path }) => `https://api.hume.ai${path}`,\n headers: getHeaders,\n fetch: options.fetch,\n });\n\n const provider = function () {\n return {\n speech: createSpeechModel(),\n };\n };\n\n provider.specificationVersion = 'v4' as const;\n provider.speech = createSpeechModel;\n provider.speechModel = createSpeechModel;\n\n provider.languageModel = (modelId: string) => {\n throw new NoSuchModelError({\n modelId,\n modelType: 'languageModel',\n message: 'Hume does not provide language models',\n });\n };\n\n provider.embeddingModel = (modelId: string) => {\n throw new NoSuchModelError({\n modelId,\n modelType: 'embeddingModel',\n message: 'Hume does not provide embedding models',\n });\n };\n\n provider.imageModel = (modelId: string) => {\n throw new NoSuchModelError({\n modelId,\n modelType: 'imageModel',\n message: 'Hume does not provide image models',\n });\n };\n\n return provider as HumeProvider;\n}\n\n/**\n * Default Hume provider instance.\n */\nexport const hume = createHume();\n","import type { SpeechModelV4, SharedV4Warning } from '@ai-sdk/provider';\nimport {\n combineHeaders,\n createBinaryResponseHandler,\n parseProviderOptions,\n postJsonToApi,\n serializeModelOptions,\n WORKFLOW_SERIALIZE,\n WORKFLOW_DESERIALIZE,\n} from '@ai-sdk/provider-utils';\nimport type { HumeConfig } from './hume-config';\nimport { humeFailedResponseHandler } from './hume-error';\nimport { humeSpeechModelOptionsSchema } from './hume-speech-model-options';\nimport type { HumeSpeechAPITypes } from './hume-api-types';\n\ninterface HumeSpeechModelConfig extends HumeConfig {\n _internal?: {\n currentDate?: () => Date;\n };\n}\n\nexport class HumeSpeechModel implements SpeechModelV4 {\n readonly specificationVersion = 'v4';\n\n get provider(): string {\n return this.config.provider;\n }\n\n static [WORKFLOW_SERIALIZE](model: HumeSpeechModel) {\n return serializeModelOptions({\n modelId: model.modelId,\n config: model.config,\n });\n }\n\n static [WORKFLOW_DESERIALIZE](options: {\n modelId: '';\n config: HumeSpeechModelConfig;\n }) {\n return new HumeSpeechModel(options.modelId as '', options.config);\n }\n\n constructor(\n readonly modelId: '',\n private readonly config: HumeSpeechModelConfig,\n ) {}\n\n private async getArgs({\n text,\n voice = 'd8ab67c6-953d-4bd8-9370-8fa53a0f1453',\n outputFormat = 'mp3',\n speed,\n instructions,\n language,\n providerOptions,\n }: Parameters<SpeechModelV4['doGenerate']>[0]) {\n const warnings: SharedV4Warning[] = [];\n\n // Parse provider options\n const humeOptions = await parseProviderOptions({\n provider: 'hume',\n providerOptions,\n schema: humeSpeechModelOptionsSchema,\n });\n\n // Create request body\n const requestBody: HumeSpeechAPITypes = {\n utterances: [\n {\n text,\n speed,\n description: instructions,\n voice: {\n id: voice,\n provider: 'HUME_AI',\n },\n },\n ],\n format: { type: 'mp3' },\n };\n\n if (outputFormat) {\n if (['mp3', 'pcm', 'wav'].includes(outputFormat)) {\n requestBody.format = { type: outputFormat as 'mp3' | 'pcm' | 'wav' };\n } else {\n warnings.push({\n type: 'unsupported',\n feature: 'outputFormat',\n details: `Unsupported output format: ${outputFormat}. Using mp3 instead.`,\n });\n }\n }\n\n // Add provider-specific options\n if (humeOptions) {\n const speechModelOptions: Omit<\n HumeSpeechAPITypes,\n 'utterances' | 'format'\n > = {};\n\n if (humeOptions.context) {\n if ('generationId' in humeOptions.context) {\n speechModelOptions.context = {\n generation_id: humeOptions.context.generationId,\n };\n } else {\n speechModelOptions.context = {\n utterances: humeOptions.context.utterances.map(utterance => ({\n text: utterance.text,\n description: utterance.description,\n speed: utterance.speed,\n trailing_silence: utterance.trailingSilence,\n voice: utterance.voice,\n })),\n };\n }\n }\n\n for (const key in speechModelOptions) {\n const value =\n speechModelOptions[\n key as keyof Omit<HumeSpeechAPITypes, 'utterances' | 'format'>\n ];\n if (value !== undefined) {\n (requestBody as Record<string, unknown>)[key] = value;\n }\n }\n }\n\n if (language) {\n warnings.push({\n type: 'unsupported',\n feature: 'language',\n details: `Hume speech models do not support language selection. Language parameter \"${language}\" was ignored.`,\n });\n }\n\n return {\n requestBody,\n warnings,\n };\n }\n\n async doGenerate(\n options: Parameters<SpeechModelV4['doGenerate']>[0],\n ): Promise<Awaited<ReturnType<SpeechModelV4['doGenerate']>>> {\n const currentDate = this.config._internal?.currentDate?.() ?? new Date();\n const { requestBody, warnings } = await this.getArgs(options);\n\n const {\n value: audio,\n responseHeaders,\n rawValue: rawResponse,\n } = await postJsonToApi({\n url: this.config.url({\n path: '/v0/tts/file',\n modelId: this.modelId,\n }),\n headers: combineHeaders(this.config.headers?.(), options.headers),\n body: requestBody,\n failedResponseHandler: humeFailedResponseHandler,\n successfulResponseHandler: createBinaryResponseHandler(),\n abortSignal: options.abortSignal,\n fetch: this.config.fetch,\n });\n\n return {\n audio,\n warnings,\n request: {\n body: JSON.stringify(requestBody),\n },\n response: {\n timestamp: currentDate,\n modelId: this.modelId,\n headers: responseHeaders,\n body: rawResponse,\n },\n };\n }\n}\n","import { z } from 'zod/v4';\nimport { createJsonErrorResponseHandler } from '@ai-sdk/provider-utils';\n\nexport const humeErrorDataSchema = z.object({\n error: z.object({\n message: z.string(),\n code: z.number(),\n }),\n});\n\nexport type HumeErrorData = z.infer<typeof humeErrorDataSchema>;\n\nexport const humeFailedResponseHandler = createJsonErrorResponseHandler({\n errorSchema: humeErrorDataSchema,\n errorToMessage: data => data.error.message,\n});\n","import { z } from 'zod/v4';\n\n// https://dev.hume.ai/reference/text-to-speech-tts/synthesize-file\nexport const humeSpeechModelOptionsSchema = z.object({\n /**\n * Context for the speech synthesis request.\n * Can be either a generationId for retrieving a previous generation,\n * or a list of utterances to synthesize.\n */\n context: z\n .object({\n /**\n * ID of a previously generated speech synthesis to retrieve.\n */\n generationId: z.string(),\n })\n .or(\n z.object({\n /**\n * List of utterances to synthesize into speech.\n */\n utterances: z.array(\n z.object({\n /**\n * The text content to convert to speech.\n */\n text: z.string(),\n /**\n * Optional description or instructions for how the text should be spoken.\n */\n description: z.string().optional(),\n /**\n * Optional speech rate multiplier.\n */\n speed: z.number().optional(),\n /**\n * Optional duration of silence to add after the utterance in seconds.\n */\n trailingSilence: z.number().optional(),\n /**\n * Voice configuration for the utterance.\n * Can be specified by ID or name.\n */\n voice: z\n .object({\n /**\n * ID of the voice to use.\n */\n id: z.string(),\n /**\n * Provider of the voice, either Hume's built-in voices or a custom voice.\n */\n provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),\n })\n .or(\n z.object({\n /**\n * Name of the voice to use.\n */\n name: z.string(),\n /**\n * Provider of the voice, either Hume's built-in voices or a custom voice.\n */\n provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),\n }),\n )\n .optional(),\n }),\n ),\n }),\n )\n .nullish(),\n});\n\nexport type HumeSpeechModelOptions = z.infer<\n typeof humeSpeechModelOptionsSchema\n>;\n","// Version string of this package injected at build time.\ndeclare const __PACKAGE_VERSION__: string | undefined;\nexport const VERSION: string =\n typeof __PACKAGE_VERSION__ !== 'undefined'\n ? __PACKAGE_VERSION__\n : '0.0.0-test';\n"],"mappings":";AAAA;AAAA,EACE;AAAA,OAGK;AACP;AAAA,EACE;AAAA,EACA;AAAA,OAEK;;;ACRP;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;;;ACTP,SAAS,SAAS;AAClB,SAAS,sCAAsC;AAExC,IAAM,sBAAsB,EAAE,OAAO;AAAA,EAC1C,OAAO,EAAE,OAAO;AAAA,IACd,SAAS,EAAE,OAAO;AAAA,IAClB,MAAM,EAAE,OAAO;AAAA,EACjB,CAAC;AACH,CAAC;AAIM,IAAM,4BAA4B,+BAA+B;AAAA,EACtE,aAAa;AAAA,EACb,gBAAgB,UAAQ,KAAK,MAAM;AACrC,CAAC;;;ACfD,SAAS,KAAAA,UAAS;AAGX,IAAM,+BAA+BA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAMnD,SAASA,GACN,OAAO;AAAA;AAAA;AAAA;AAAA,IAIN,cAAcA,GAAE,OAAO;AAAA,EACzB,CAAC,EACA;AAAA,IACCA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,MAIP,YAAYA,GAAE;AAAA,QACZA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,UAIP,MAAMA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,UAIf,aAAaA,GAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA,UAIjC,OAAOA,GAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA,UAI3B,iBAAiBA,GAAE,OAAO,EAAE,SAAS;AAAA;AAAA;AAAA;AAAA;AAAA,UAKrC,OAAOA,GACJ,OAAO;AAAA;AAAA;AAAA;AAAA,YAIN,IAAIA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,YAIb,UAAUA,GAAE,KAAK,CAAC,WAAW,cAAc,CAAC,EAAE,SAAS;AAAA,UACzD,CAAC,EACA;AAAA,YACCA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,cAIP,MAAMA,GAAE,OAAO;AAAA;AAAA;AAAA;AAAA,cAIf,UAAUA,GAAE,KAAK,CAAC,WAAW,cAAc,CAAC,EAAE,SAAS;AAAA,YACzD,CAAC;AAAA,UACH,EACC,SAAS;AAAA,QACd,CAAC;AAAA,MACH;AAAA,IACF,CAAC;AAAA,EACH,EACC,QAAQ;AACb,CAAC;;;AFnDM,IAAM,kBAAN,MAAM,iBAAyC;AAAA,EAqBpD,YACW,SACQ,QACjB;AAFS;AACQ;AAtBnB,SAAS,uBAAuB;AAAA,EAuB7B;AAAA,EArBH,IAAI,WAAmB;AACrB,WAAO,KAAK,OAAO;AAAA,EACrB;AAAA,EAEA,QAAQ,kBAAkB,EAAE,OAAwB;AAClD,WAAO,sBAAsB;AAAA,MAC3B,SAAS,MAAM;AAAA,MACf,QAAQ,MAAM;AAAA,IAChB,CAAC;AAAA,EACH;AAAA,EAEA,QAAQ,oBAAoB,EAAE,SAG3B;AACD,WAAO,IAAI,iBAAgB,QAAQ,SAAe,QAAQ,MAAM;AAAA,EAClE;AAAA,EAOA,MAAc,QAAQ;AAAA,IACpB;AAAA,IACA,QAAQ;AAAA,IACR,eAAe;AAAA,IACf;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,GAA+C;AAC7C,UAAM,WAA8B,CAAC;AAGrC,UAAM,cAAc,MAAM,qBAAqB;AAAA,MAC7C,UAAU;AAAA,MACV;AAAA,MACA,QAAQ;AAAA,IACV,CAAC;AAGD,UAAM,cAAkC;AAAA,MACtC,YAAY;AAAA,QACV;AAAA,UACE;AAAA,UACA;AAAA,UACA,aAAa;AAAA,UACb,OAAO;AAAA,YACL,IAAI;AAAA,YACJ,UAAU;AAAA,UACZ;AAAA,QACF;AAAA,MACF;AAAA,MACA,QAAQ,EAAE,MAAM,MAAM;AAAA,IACxB;AAEA,QAAI,cAAc;AAChB,UAAI,CAAC,OAAO,OAAO,KAAK,EAAE,SAAS,YAAY,GAAG;AAChD,oBAAY,SAAS,EAAE,MAAM,aAAsC;AAAA,MACrE,OAAO;AACL,iBAAS,KAAK;AAAA,UACZ,MAAM;AAAA,UACN,SAAS;AAAA,UACT,SAAS,8BAA8B,YAAY;AAAA,QACrD,CAAC;AAAA,MACH;AAAA,IACF;AAGA,QAAI,aAAa;AACf,YAAM,qBAGF,CAAC;AAEL,UAAI,YAAY,SAAS;AACvB,YAAI,kBAAkB,YAAY,SAAS;AACzC,6BAAmB,UAAU;AAAA,YAC3B,eAAe,YAAY,QAAQ;AAAA,UACrC;AAAA,QACF,OAAO;AACL,6BAAmB,UAAU;AAAA,YAC3B,YAAY,YAAY,QAAQ,WAAW,IAAI,gBAAc;AAAA,cAC3D,MAAM,UAAU;AAAA,cAChB,aAAa,UAAU;AAAA,cACvB,OAAO,UAAU;AAAA,cACjB,kBAAkB,UAAU;AAAA,cAC5B,OAAO,UAAU;AAAA,YACnB,EAAE;AAAA,UACJ;AAAA,QACF;AAAA,MACF;AAEA,iBAAW,OAAO,oBAAoB;AACpC,cAAM,QACJ,mBACE,GACF;AACF,YAAI,UAAU,QAAW;AACvB,UAAC,YAAwC,GAAG,IAAI;AAAA,QAClD;AAAA,MACF;AAAA,IACF;AAEA,QAAI,UAAU;AACZ,eAAS,KAAK;AAAA,QACZ,MAAM;AAAA,QACN,SAAS;AAAA,QACT,SAAS,6EAA6E,QAAQ;AAAA,MAChG,CAAC;AAAA,IACH;AAEA,WAAO;AAAA,MACL;AAAA,MACA;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAM,WACJ,SAC2D;AAjJ/D;AAkJI,UAAM,eAAc,sBAAK,OAAO,cAAZ,mBAAuB,gBAAvB,4CAA0C,oBAAI,KAAK;AACvE,UAAM,EAAE,aAAa,SAAS,IAAI,MAAM,KAAK,QAAQ,OAAO;AAE5D,UAAM;AAAA,MACJ,OAAO;AAAA,MACP;AAAA,MACA,UAAU;AAAA,IACZ,IAAI,MAAM,cAAc;AAAA,MACtB,KAAK,KAAK,OAAO,IAAI;AAAA,QACnB,MAAM;AAAA,QACN,SAAS,KAAK;AAAA,MAChB,CAAC;AAAA,MACD,SAAS,gBAAe,gBAAK,QAAO,YAAZ,6BAAyB,QAAQ,OAAO;AAAA,MAChE,MAAM;AAAA,MACN,uBAAuB;AAAA,MACvB,2BAA2B,4BAA4B;AAAA,MACvD,aAAa,QAAQ;AAAA,MACrB,OAAO,KAAK,OAAO;AAAA,IACrB,CAAC;AAED,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA,SAAS;AAAA,QACP,MAAM,KAAK,UAAU,WAAW;AAAA,MAClC;AAAA,MACA,UAAU;AAAA,QACR,WAAW;AAAA,QACX,SAAS,KAAK;AAAA,QACd,SAAS;AAAA,QACT,MAAM;AAAA,MACR;AAAA,IACF;AAAA,EACF;AACF;;;AGlLO,IAAM,UACX,OACI,kBACA;;;AJwCC,SAAS,WAAW,UAAgC,CAAC,GAAiB;AAC3E,QAAM,aAAa,MACjB;AAAA,IACE;AAAA,MACE,kBAAkB,WAAW;AAAA,QAC3B,QAAQ,QAAQ;AAAA,QAChB,yBAAyB;AAAA,QACzB,aAAa;AAAA,MACf,CAAC;AAAA,MACD,GAAG,QAAQ;AAAA,IACb;AAAA,IACA,eAAe,OAAO;AAAA,EACxB;AAEF,QAAM,oBAAoB,MACxB,IAAI,gBAAgB,IAAI;AAAA,IACtB,UAAU;AAAA,IACV,KAAK,CAAC,EAAE,KAAK,MAAM,sBAAsB,IAAI;AAAA,IAC7C,SAAS;AAAA,IACT,OAAO,QAAQ;AAAA,EACjB,CAAC;AAEH,QAAM,WAAW,WAAY;AAC3B,WAAO;AAAA,MACL,QAAQ,kBAAkB;AAAA,IAC5B;AAAA,EACF;AAEA,WAAS,uBAAuB;AAChC,WAAS,SAAS;AAClB,WAAS,cAAc;AAEvB,WAAS,gBAAgB,CAAC,YAAoB;AAC5C,UAAM,IAAI,iBAAiB;AAAA,MACzB;AAAA,MACA,WAAW;AAAA,MACX,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AAEA,WAAS,iBAAiB,CAAC,YAAoB;AAC7C,UAAM,IAAI,iBAAiB;AAAA,MACzB;AAAA,MACA,WAAW;AAAA,MACX,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AAEA,WAAS,aAAa,CAAC,YAAoB;AACzC,UAAM,IAAI,iBAAiB;AAAA,MACzB;AAAA,MACA,WAAW;AAAA,MACX,SAAS;AAAA,IACX,CAAC;AAAA,EACH;AAEA,SAAO;AACT;AAKO,IAAM,OAAO,WAAW;","names":["z"]}
|
package/package.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@ai-sdk/hume",
|
|
3
|
-
"version": "3.0.0-beta.
|
|
3
|
+
"version": "3.0.0-beta.31",
|
|
4
|
+
"type": "module",
|
|
4
5
|
"license": "Apache-2.0",
|
|
5
6
|
"sideEffects": false,
|
|
6
7
|
"main": "./dist/index.js",
|
|
7
|
-
"module": "./dist/index.mjs",
|
|
8
8
|
"types": "./dist/index.d.ts",
|
|
9
9
|
"files": [
|
|
10
10
|
"dist/**/*",
|
|
@@ -24,20 +24,20 @@
|
|
|
24
24
|
"./package.json": "./package.json",
|
|
25
25
|
".": {
|
|
26
26
|
"types": "./dist/index.d.ts",
|
|
27
|
-
"import": "./dist/index.
|
|
28
|
-
"
|
|
27
|
+
"import": "./dist/index.js",
|
|
28
|
+
"default": "./dist/index.js"
|
|
29
29
|
}
|
|
30
30
|
},
|
|
31
31
|
"dependencies": {
|
|
32
|
-
"@ai-sdk/provider": "4.0.0-beta.
|
|
33
|
-
"@ai-sdk/provider-utils": "5.0.0-beta.
|
|
32
|
+
"@ai-sdk/provider": "4.0.0-beta.14",
|
|
33
|
+
"@ai-sdk/provider-utils": "5.0.0-beta.30"
|
|
34
34
|
},
|
|
35
35
|
"devDependencies": {
|
|
36
36
|
"@types/node": "20.17.24",
|
|
37
37
|
"tsup": "^8",
|
|
38
38
|
"typescript": "5.6.3",
|
|
39
39
|
"zod": "3.25.76",
|
|
40
|
-
"@ai-sdk/test-server": "2.0.0-beta.
|
|
40
|
+
"@ai-sdk/test-server": "2.0.0-beta.3",
|
|
41
41
|
"@vercel/ai-tsconfig": "0.0.0"
|
|
42
42
|
},
|
|
43
43
|
"peerDependencies": {
|
|
@@ -47,12 +47,14 @@
|
|
|
47
47
|
"node": ">=18"
|
|
48
48
|
},
|
|
49
49
|
"publishConfig": {
|
|
50
|
-
"access": "public"
|
|
50
|
+
"access": "public",
|
|
51
|
+
"provenance": true
|
|
51
52
|
},
|
|
52
53
|
"homepage": "https://ai-sdk.dev/docs",
|
|
53
54
|
"repository": {
|
|
54
55
|
"type": "git",
|
|
55
|
-
"url": "
|
|
56
|
+
"url": "https://github.com/vercel/ai",
|
|
57
|
+
"directory": "packages/hume"
|
|
56
58
|
},
|
|
57
59
|
"bugs": {
|
|
58
60
|
"url": "https://github.com/vercel/ai/issues"
|
|
@@ -64,9 +66,7 @@
|
|
|
64
66
|
"build": "tsup --tsconfig tsconfig.build.json",
|
|
65
67
|
"build:watch": "tsup --tsconfig tsconfig.build.json --watch",
|
|
66
68
|
"clean": "del-cli dist docs",
|
|
67
|
-
"lint": "eslint \"./**/*.ts*\"",
|
|
68
69
|
"type-check": "tsc --noEmit",
|
|
69
|
-
"prettier-check": "prettier --check \"./**/*.ts*\"",
|
|
70
70
|
"test": "pnpm test:node && pnpm test:edge",
|
|
71
71
|
"test:watch": "vitest --config vitest.node.config.js --watch",
|
|
72
72
|
"test:edge": "vitest --config vitest.edge.config.js --run",
|
package/src/hume-config.ts
CHANGED
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
import { FetchFunction } from '@ai-sdk/provider-utils';
|
|
1
|
+
import type { FetchFunction } from '@ai-sdk/provider-utils';
|
|
2
2
|
|
|
3
3
|
export type HumeConfig = {
|
|
4
4
|
provider: string;
|
|
5
5
|
url: (options: { modelId: string; path: string }) => string;
|
|
6
|
-
headers
|
|
6
|
+
headers?: () => Record<string, string | undefined>;
|
|
7
7
|
fetch?: FetchFunction;
|
|
8
8
|
generateId?: () => string;
|
|
9
9
|
};
|
package/src/hume-provider.ts
CHANGED
|
@@ -1,13 +1,17 @@
|
|
|
1
|
-
import { SpeechModelV3, ProviderV3 } from '@ai-sdk/provider';
|
|
2
1
|
import {
|
|
3
|
-
|
|
2
|
+
NoSuchModelError,
|
|
3
|
+
type SpeechModelV4,
|
|
4
|
+
type ProviderV4,
|
|
5
|
+
} from '@ai-sdk/provider';
|
|
6
|
+
import {
|
|
4
7
|
loadApiKey,
|
|
5
8
|
withUserAgentSuffix,
|
|
9
|
+
type FetchFunction,
|
|
6
10
|
} from '@ai-sdk/provider-utils';
|
|
7
11
|
import { HumeSpeechModel } from './hume-speech-model';
|
|
8
12
|
import { VERSION } from './version';
|
|
9
13
|
|
|
10
|
-
export interface HumeProvider extends
|
|
14
|
+
export interface HumeProvider extends ProviderV4 {
|
|
11
15
|
(settings?: {}): {
|
|
12
16
|
speech: HumeSpeechModel;
|
|
13
17
|
};
|
|
@@ -15,7 +19,7 @@ export interface HumeProvider extends Pick<ProviderV3, 'speechModel'> {
|
|
|
15
19
|
/**
|
|
16
20
|
* Creates a model for speech synthesis.
|
|
17
21
|
*/
|
|
18
|
-
speech():
|
|
22
|
+
speech(): SpeechModelV4;
|
|
19
23
|
}
|
|
20
24
|
|
|
21
25
|
export interface HumeProviderSettings {
|
|
@@ -67,10 +71,35 @@ export function createHume(options: HumeProviderSettings = {}): HumeProvider {
|
|
|
67
71
|
};
|
|
68
72
|
};
|
|
69
73
|
|
|
74
|
+
provider.specificationVersion = 'v4' as const;
|
|
70
75
|
provider.speech = createSpeechModel;
|
|
71
76
|
provider.speechModel = createSpeechModel;
|
|
72
77
|
|
|
73
|
-
|
|
78
|
+
provider.languageModel = (modelId: string) => {
|
|
79
|
+
throw new NoSuchModelError({
|
|
80
|
+
modelId,
|
|
81
|
+
modelType: 'languageModel',
|
|
82
|
+
message: 'Hume does not provide language models',
|
|
83
|
+
});
|
|
84
|
+
};
|
|
85
|
+
|
|
86
|
+
provider.embeddingModel = (modelId: string) => {
|
|
87
|
+
throw new NoSuchModelError({
|
|
88
|
+
modelId,
|
|
89
|
+
modelType: 'embeddingModel',
|
|
90
|
+
message: 'Hume does not provide embedding models',
|
|
91
|
+
});
|
|
92
|
+
};
|
|
93
|
+
|
|
94
|
+
provider.imageModel = (modelId: string) => {
|
|
95
|
+
throw new NoSuchModelError({
|
|
96
|
+
modelId,
|
|
97
|
+
modelType: 'imageModel',
|
|
98
|
+
message: 'Hume does not provide image models',
|
|
99
|
+
});
|
|
100
|
+
};
|
|
101
|
+
|
|
102
|
+
return provider as HumeProvider;
|
|
74
103
|
}
|
|
75
104
|
|
|
76
105
|
/**
|
|
@@ -0,0 +1,77 @@
|
|
|
1
|
+
import { z } from 'zod/v4';
|
|
2
|
+
|
|
3
|
+
// https://dev.hume.ai/reference/text-to-speech-tts/synthesize-file
|
|
4
|
+
export const humeSpeechModelOptionsSchema = z.object({
|
|
5
|
+
/**
|
|
6
|
+
* Context for the speech synthesis request.
|
|
7
|
+
* Can be either a generationId for retrieving a previous generation,
|
|
8
|
+
* or a list of utterances to synthesize.
|
|
9
|
+
*/
|
|
10
|
+
context: z
|
|
11
|
+
.object({
|
|
12
|
+
/**
|
|
13
|
+
* ID of a previously generated speech synthesis to retrieve.
|
|
14
|
+
*/
|
|
15
|
+
generationId: z.string(),
|
|
16
|
+
})
|
|
17
|
+
.or(
|
|
18
|
+
z.object({
|
|
19
|
+
/**
|
|
20
|
+
* List of utterances to synthesize into speech.
|
|
21
|
+
*/
|
|
22
|
+
utterances: z.array(
|
|
23
|
+
z.object({
|
|
24
|
+
/**
|
|
25
|
+
* The text content to convert to speech.
|
|
26
|
+
*/
|
|
27
|
+
text: z.string(),
|
|
28
|
+
/**
|
|
29
|
+
* Optional description or instructions for how the text should be spoken.
|
|
30
|
+
*/
|
|
31
|
+
description: z.string().optional(),
|
|
32
|
+
/**
|
|
33
|
+
* Optional speech rate multiplier.
|
|
34
|
+
*/
|
|
35
|
+
speed: z.number().optional(),
|
|
36
|
+
/**
|
|
37
|
+
* Optional duration of silence to add after the utterance in seconds.
|
|
38
|
+
*/
|
|
39
|
+
trailingSilence: z.number().optional(),
|
|
40
|
+
/**
|
|
41
|
+
* Voice configuration for the utterance.
|
|
42
|
+
* Can be specified by ID or name.
|
|
43
|
+
*/
|
|
44
|
+
voice: z
|
|
45
|
+
.object({
|
|
46
|
+
/**
|
|
47
|
+
* ID of the voice to use.
|
|
48
|
+
*/
|
|
49
|
+
id: z.string(),
|
|
50
|
+
/**
|
|
51
|
+
* Provider of the voice, either Hume's built-in voices or a custom voice.
|
|
52
|
+
*/
|
|
53
|
+
provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),
|
|
54
|
+
})
|
|
55
|
+
.or(
|
|
56
|
+
z.object({
|
|
57
|
+
/**
|
|
58
|
+
* Name of the voice to use.
|
|
59
|
+
*/
|
|
60
|
+
name: z.string(),
|
|
61
|
+
/**
|
|
62
|
+
* Provider of the voice, either Hume's built-in voices or a custom voice.
|
|
63
|
+
*/
|
|
64
|
+
provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),
|
|
65
|
+
}),
|
|
66
|
+
)
|
|
67
|
+
.optional(),
|
|
68
|
+
}),
|
|
69
|
+
),
|
|
70
|
+
}),
|
|
71
|
+
)
|
|
72
|
+
.nullish(),
|
|
73
|
+
});
|
|
74
|
+
|
|
75
|
+
export type HumeSpeechModelOptions = z.infer<
|
|
76
|
+
typeof humeSpeechModelOptionsSchema
|
|
77
|
+
>;
|
package/src/hume-speech-model.ts
CHANGED
|
@@ -1,90 +1,17 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import type { SpeechModelV4, SharedV4Warning } from '@ai-sdk/provider';
|
|
2
2
|
import {
|
|
3
3
|
combineHeaders,
|
|
4
4
|
createBinaryResponseHandler,
|
|
5
5
|
parseProviderOptions,
|
|
6
6
|
postJsonToApi,
|
|
7
|
+
serializeModelOptions,
|
|
8
|
+
WORKFLOW_SERIALIZE,
|
|
9
|
+
WORKFLOW_DESERIALIZE,
|
|
7
10
|
} from '@ai-sdk/provider-utils';
|
|
8
|
-
import {
|
|
9
|
-
import { HumeConfig } from './hume-config';
|
|
11
|
+
import type { HumeConfig } from './hume-config';
|
|
10
12
|
import { humeFailedResponseHandler } from './hume-error';
|
|
11
|
-
import {
|
|
12
|
-
|
|
13
|
-
// https://dev.hume.ai/reference/text-to-speech-tts/synthesize-file
|
|
14
|
-
const humeSpeechModelOptionsSchema = z.object({
|
|
15
|
-
/**
|
|
16
|
-
* Context for the speech synthesis request.
|
|
17
|
-
* Can be either a generationId for retrieving a previous generation,
|
|
18
|
-
* or a list of utterances to synthesize.
|
|
19
|
-
*/
|
|
20
|
-
context: z
|
|
21
|
-
.object({
|
|
22
|
-
/**
|
|
23
|
-
* ID of a previously generated speech synthesis to retrieve.
|
|
24
|
-
*/
|
|
25
|
-
generationId: z.string(),
|
|
26
|
-
})
|
|
27
|
-
.or(
|
|
28
|
-
z.object({
|
|
29
|
-
/**
|
|
30
|
-
* List of utterances to synthesize into speech.
|
|
31
|
-
*/
|
|
32
|
-
utterances: z.array(
|
|
33
|
-
z.object({
|
|
34
|
-
/**
|
|
35
|
-
* The text content to convert to speech.
|
|
36
|
-
*/
|
|
37
|
-
text: z.string(),
|
|
38
|
-
/**
|
|
39
|
-
* Optional description or instructions for how the text should be spoken.
|
|
40
|
-
*/
|
|
41
|
-
description: z.string().optional(),
|
|
42
|
-
/**
|
|
43
|
-
* Optional speech rate multiplier.
|
|
44
|
-
*/
|
|
45
|
-
speed: z.number().optional(),
|
|
46
|
-
/**
|
|
47
|
-
* Optional duration of silence to add after the utterance in seconds.
|
|
48
|
-
*/
|
|
49
|
-
trailingSilence: z.number().optional(),
|
|
50
|
-
/**
|
|
51
|
-
* Voice configuration for the utterance.
|
|
52
|
-
* Can be specified by ID or name.
|
|
53
|
-
*/
|
|
54
|
-
voice: z
|
|
55
|
-
.object({
|
|
56
|
-
/**
|
|
57
|
-
* ID of the voice to use.
|
|
58
|
-
*/
|
|
59
|
-
id: z.string(),
|
|
60
|
-
/**
|
|
61
|
-
* Provider of the voice, either Hume's built-in voices or a custom voice.
|
|
62
|
-
*/
|
|
63
|
-
provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),
|
|
64
|
-
})
|
|
65
|
-
.or(
|
|
66
|
-
z.object({
|
|
67
|
-
/**
|
|
68
|
-
* Name of the voice to use.
|
|
69
|
-
*/
|
|
70
|
-
name: z.string(),
|
|
71
|
-
/**
|
|
72
|
-
* Provider of the voice, either Hume's built-in voices or a custom voice.
|
|
73
|
-
*/
|
|
74
|
-
provider: z.enum(['HUME_AI', 'CUSTOM_VOICE']).optional(),
|
|
75
|
-
}),
|
|
76
|
-
)
|
|
77
|
-
.optional(),
|
|
78
|
-
}),
|
|
79
|
-
),
|
|
80
|
-
}),
|
|
81
|
-
)
|
|
82
|
-
.nullish(),
|
|
83
|
-
});
|
|
84
|
-
|
|
85
|
-
export type HumeSpeechModelOptions = z.infer<
|
|
86
|
-
typeof humeSpeechModelOptionsSchema
|
|
87
|
-
>;
|
|
13
|
+
import { humeSpeechModelOptionsSchema } from './hume-speech-model-options';
|
|
14
|
+
import type { HumeSpeechAPITypes } from './hume-api-types';
|
|
88
15
|
|
|
89
16
|
interface HumeSpeechModelConfig extends HumeConfig {
|
|
90
17
|
_internal?: {
|
|
@@ -92,13 +19,27 @@ interface HumeSpeechModelConfig extends HumeConfig {
|
|
|
92
19
|
};
|
|
93
20
|
}
|
|
94
21
|
|
|
95
|
-
export class HumeSpeechModel implements
|
|
96
|
-
readonly specificationVersion = '
|
|
22
|
+
export class HumeSpeechModel implements SpeechModelV4 {
|
|
23
|
+
readonly specificationVersion = 'v4';
|
|
97
24
|
|
|
98
25
|
get provider(): string {
|
|
99
26
|
return this.config.provider;
|
|
100
27
|
}
|
|
101
28
|
|
|
29
|
+
static [WORKFLOW_SERIALIZE](model: HumeSpeechModel) {
|
|
30
|
+
return serializeModelOptions({
|
|
31
|
+
modelId: model.modelId,
|
|
32
|
+
config: model.config,
|
|
33
|
+
});
|
|
34
|
+
}
|
|
35
|
+
|
|
36
|
+
static [WORKFLOW_DESERIALIZE](options: {
|
|
37
|
+
modelId: '';
|
|
38
|
+
config: HumeSpeechModelConfig;
|
|
39
|
+
}) {
|
|
40
|
+
return new HumeSpeechModel(options.modelId as '', options.config);
|
|
41
|
+
}
|
|
42
|
+
|
|
102
43
|
constructor(
|
|
103
44
|
readonly modelId: '',
|
|
104
45
|
private readonly config: HumeSpeechModelConfig,
|
|
@@ -112,8 +53,8 @@ export class HumeSpeechModel implements SpeechModelV3 {
|
|
|
112
53
|
instructions,
|
|
113
54
|
language,
|
|
114
55
|
providerOptions,
|
|
115
|
-
}: Parameters<
|
|
116
|
-
const warnings:
|
|
56
|
+
}: Parameters<SpeechModelV4['doGenerate']>[0]) {
|
|
57
|
+
const warnings: SharedV4Warning[] = [];
|
|
117
58
|
|
|
118
59
|
// Parse provider options
|
|
119
60
|
const humeOptions = await parseProviderOptions({
|
|
@@ -201,8 +142,8 @@ export class HumeSpeechModel implements SpeechModelV3 {
|
|
|
201
142
|
}
|
|
202
143
|
|
|
203
144
|
async doGenerate(
|
|
204
|
-
options: Parameters<
|
|
205
|
-
): Promise<Awaited<ReturnType<
|
|
145
|
+
options: Parameters<SpeechModelV4['doGenerate']>[0],
|
|
146
|
+
): Promise<Awaited<ReturnType<SpeechModelV4['doGenerate']>>> {
|
|
206
147
|
const currentDate = this.config._internal?.currentDate?.() ?? new Date();
|
|
207
148
|
const { requestBody, warnings } = await this.getArgs(options);
|
|
208
149
|
|
|
@@ -215,7 +156,7 @@ export class HumeSpeechModel implements SpeechModelV3 {
|
|
|
215
156
|
path: '/v0/tts/file',
|
|
216
157
|
modelId: this.modelId,
|
|
217
158
|
}),
|
|
218
|
-
headers: combineHeaders(this.config.headers(), options.headers),
|
|
159
|
+
headers: combineHeaders(this.config.headers?.(), options.headers),
|
|
219
160
|
body: requestBody,
|
|
220
161
|
failedResponseHandler: humeFailedResponseHandler,
|
|
221
162
|
successfulResponseHandler: createBinaryResponseHandler(),
|
package/src/index.ts
CHANGED
|
@@ -1,4 +1,4 @@
|
|
|
1
1
|
export { createHume, hume } from './hume-provider';
|
|
2
2
|
export type { HumeProvider, HumeProviderSettings } from './hume-provider';
|
|
3
|
-
export type { HumeSpeechModelOptions } from './hume-speech-model';
|
|
3
|
+
export type { HumeSpeechModelOptions } from './hume-speech-model-options';
|
|
4
4
|
export { VERSION } from './version';
|
package/dist/index.d.mts
DELETED
|
@@ -1,92 +0,0 @@
|
|
|
1
|
-
import { SpeechModelV3, ProviderV3 } from '@ai-sdk/provider';
|
|
2
|
-
import { FetchFunction } from '@ai-sdk/provider-utils';
|
|
3
|
-
import { z } from 'zod/v4';
|
|
4
|
-
|
|
5
|
-
type HumeConfig = {
|
|
6
|
-
provider: string;
|
|
7
|
-
url: (options: {
|
|
8
|
-
modelId: string;
|
|
9
|
-
path: string;
|
|
10
|
-
}) => string;
|
|
11
|
-
headers: () => Record<string, string | undefined>;
|
|
12
|
-
fetch?: FetchFunction;
|
|
13
|
-
generateId?: () => string;
|
|
14
|
-
};
|
|
15
|
-
|
|
16
|
-
declare const humeSpeechModelOptionsSchema: z.ZodObject<{
|
|
17
|
-
context: z.ZodOptional<z.ZodNullable<z.ZodUnion<[z.ZodObject<{
|
|
18
|
-
generationId: z.ZodString;
|
|
19
|
-
}, z.core.$strip>, z.ZodObject<{
|
|
20
|
-
utterances: z.ZodArray<z.ZodObject<{
|
|
21
|
-
text: z.ZodString;
|
|
22
|
-
description: z.ZodOptional<z.ZodString>;
|
|
23
|
-
speed: z.ZodOptional<z.ZodNumber>;
|
|
24
|
-
trailingSilence: z.ZodOptional<z.ZodNumber>;
|
|
25
|
-
voice: z.ZodOptional<z.ZodUnion<[z.ZodObject<{
|
|
26
|
-
id: z.ZodString;
|
|
27
|
-
provider: z.ZodOptional<z.ZodEnum<{
|
|
28
|
-
HUME_AI: "HUME_AI";
|
|
29
|
-
CUSTOM_VOICE: "CUSTOM_VOICE";
|
|
30
|
-
}>>;
|
|
31
|
-
}, z.core.$strip>, z.ZodObject<{
|
|
32
|
-
name: z.ZodString;
|
|
33
|
-
provider: z.ZodOptional<z.ZodEnum<{
|
|
34
|
-
HUME_AI: "HUME_AI";
|
|
35
|
-
CUSTOM_VOICE: "CUSTOM_VOICE";
|
|
36
|
-
}>>;
|
|
37
|
-
}, z.core.$strip>]>>;
|
|
38
|
-
}, z.core.$strip>>;
|
|
39
|
-
}, z.core.$strip>]>>>;
|
|
40
|
-
}, z.core.$strip>;
|
|
41
|
-
type HumeSpeechModelOptions = z.infer<typeof humeSpeechModelOptionsSchema>;
|
|
42
|
-
interface HumeSpeechModelConfig extends HumeConfig {
|
|
43
|
-
_internal?: {
|
|
44
|
-
currentDate?: () => Date;
|
|
45
|
-
};
|
|
46
|
-
}
|
|
47
|
-
declare class HumeSpeechModel implements SpeechModelV3 {
|
|
48
|
-
readonly modelId: '';
|
|
49
|
-
private readonly config;
|
|
50
|
-
readonly specificationVersion = "v3";
|
|
51
|
-
get provider(): string;
|
|
52
|
-
constructor(modelId: '', config: HumeSpeechModelConfig);
|
|
53
|
-
private getArgs;
|
|
54
|
-
doGenerate(options: Parameters<SpeechModelV3['doGenerate']>[0]): Promise<Awaited<ReturnType<SpeechModelV3['doGenerate']>>>;
|
|
55
|
-
}
|
|
56
|
-
|
|
57
|
-
interface HumeProvider extends Pick<ProviderV3, 'speechModel'> {
|
|
58
|
-
(settings?: {}): {
|
|
59
|
-
speech: HumeSpeechModel;
|
|
60
|
-
};
|
|
61
|
-
/**
|
|
62
|
-
* Creates a model for speech synthesis.
|
|
63
|
-
*/
|
|
64
|
-
speech(): SpeechModelV3;
|
|
65
|
-
}
|
|
66
|
-
interface HumeProviderSettings {
|
|
67
|
-
/**
|
|
68
|
-
* API key for authenticating requests.
|
|
69
|
-
*/
|
|
70
|
-
apiKey?: string;
|
|
71
|
-
/**
|
|
72
|
-
* Custom headers to include in the requests.
|
|
73
|
-
*/
|
|
74
|
-
headers?: Record<string, string>;
|
|
75
|
-
/**
|
|
76
|
-
* Custom fetch implementation. You can use it as a middleware to intercept requests,
|
|
77
|
-
* or to provide a custom fetch implementation for e.g. testing.
|
|
78
|
-
*/
|
|
79
|
-
fetch?: FetchFunction;
|
|
80
|
-
}
|
|
81
|
-
/**
|
|
82
|
-
* Create an Hume provider instance.
|
|
83
|
-
*/
|
|
84
|
-
declare function createHume(options?: HumeProviderSettings): HumeProvider;
|
|
85
|
-
/**
|
|
86
|
-
* Default Hume provider instance.
|
|
87
|
-
*/
|
|
88
|
-
declare const hume: HumeProvider;
|
|
89
|
-
|
|
90
|
-
declare const VERSION: string;
|
|
91
|
-
|
|
92
|
-
export { type HumeProvider, type HumeProviderSettings, type HumeSpeechModelOptions, VERSION, createHume, hume };
|