> ## Documentation Index > Fetch the complete documentation index at: https://docs.voice.wixzel.com/llms.txt > Use this file to discover all available pages before exploring further. # Clone a voice > Makes a cloned voice from recordings of one person speaking, and charges the one-off fee in the price book (`voice_clone`). `consent` must be `true`. Send the recordings as `files` in a multipart request, or as `audio_urls` in JSON. The recordings are passed to the provider and not stored. Put the returned `provider_voice_id` in an agent’s `tts.voice`; only engines listed in `engines` can speak in it. Requires `voices:write`. ## OpenAPI ````yaml /openapi.json post /v1/voices openapi: 3.1.0 info: title: Wixzel Voice API version: '2026-09-01' description: >- APIs for agentic telephony. One API key, one balance, every voice engine. ## Authentication Send your key as `Authorization: Bearer wv_live_...`. Keys are scoped: grant only what an integration needs. There is no admin scope. ## Versioning The `/v1` prefix covers additive changes. Behavioural changes ship behind a dated `Wixzel-Version` header, and existing keys keep the behaviour they were created with. ## Money Amounts are integer **micro-USD** (1,000,000 = $1.00). Voice usage is billed per second, per token and per character, so a float dollar figure cannot represent it without disagreeing with the ledger. ## Errors Every error carries a stable `code` and a `request_id`. Match on `code`; the `message` is for humans and may change. contact: name: Wixzel Voice url: https://voice.wixzel.com servers: - url: https://api.voice.wixzel.com description: Production security: [] tags: - name: Calls description: Place calls and read what happened on them. - name: Realtime description: Talk to an agent from a browser or app, with no phone line. - name: Agents description: The prompt, voice and behaviour of a caller. - name: Leads description: People to call, and data to merge into prompts. - name: Campaigns description: Call a list of leads with one agent. - name: Knowledge bases description: Facts an agent can draw on mid-call. - name: Phone numbers description: Numbers on your SIP trunks. - name: SIP trunks description: Your carrier connections. - name: Appointments description: Bookings, including ones agents make on calls. - name: Usage description: Itemised billing lines. - name: Billing description: Balance, ledger and top-ups. - name: API keys description: Create, scope and rotate keys. - name: Engines description: What the platform can serve, and what it costs. - name: Voices description: Cloned voices, made from your own recordings, for your agents to speak in. - name: Webhooks description: Where events are sent, and what happened when they got there. - name: MCP servers description: Other apps’ tools, for agents to use while they are on a call. paths: /v1/voices: post: tags: - Voices summary: Clone a voice description: >- Makes a cloned voice from recordings of one person speaking, and charges the one-off fee in the price book (`voice_clone`). `consent` must be `true`. Send the recordings as `files` in a multipart request, or as `audio_urls` in JSON. The recordings are passed to the provider and not stored. Put the returned `provider_voice_id` in an agent’s `tts.voice`; only engines listed in `engines` can speak in it. Requires `voices:write`. requestBody: content: multipart/form-data: schema: $ref: '#/components/schemas/CreateVoiceMultipart' application/json: schema: $ref: '#/components/schemas/CreateVoice' responses: '201': description: The new voice content: application/json: schema: $ref: '#/components/schemas/Voice' '400': description: Invalid request content: application/json: schema: $ref: '#/components/schemas/Error' '401': description: Missing or invalid API key content: application/json: schema: $ref: '#/components/schemas/Error' '402': description: Not enough credit for the cloning fee content: application/json: schema: $ref: '#/components/schemas/Error' '403': description: Key lacks the required scope content: application/json: schema: $ref: '#/components/schemas/Error' '409': description: The account already has its maximum number of voices content: application/json: schema: $ref: '#/components/schemas/Error' '413': description: A recording is larger than 10 MB content: application/json: schema: $ref: '#/components/schemas/Error' '429': description: Rate limited. Retry after the interval in `Retry-After`. content: application/json: schema: $ref: '#/components/schemas/Error' '503': description: Voice cloning is temporarily unavailable. Nothing was charged. content: application/json: schema: $ref: '#/components/schemas/Error' security: - bearerAuth: [] components: schemas: CreateVoiceMultipart: type: object properties: name: type: string minLength: 1 maxLength: 100 example: Dr. Meera, front desk description: type: string maxLength: 500 provider: type: string enum: - elevenlabs - sarvam default: elevenlabs description: >- Where the clone is made, which decides the engines it speaks on. `elevenlabs` for the `classic` and `deepgram_agent` engines; `sarvam` for the `sarvam` engine, which is the one to use for Indian languages. language: type: string pattern: ^[a-z]{2}-[A-Z]{2}$ description: The language spoken in the recording. Required for `sarvam`. example: ml-IN remove_background_noise: type: boolean description: >- ElevenLabs only. Cleans up a noisy recording; it can make a clean one worse. consent: type: boolean description: >- Must be `true`. You confirm that the voice is yours, or that the speaker has given you permission to clone it and use it on calls. Recorded with the time and your account. files: type: array items: type: string format: binary minItems: 1 maxItems: 5 description: >- The recordings. 30 seconds to a few minutes of one person speaking clearly, with no music or other voices. mp3, wav, m4a, ogg, webm or flac, up to 10 MB each. ElevenLabs takes up to 5 files; Sarvam takes 1, and 10 to 30 seconds is enough. required: - name - consent - files CreateVoice: type: object properties: name: type: string minLength: 1 maxLength: 100 example: Dr. Meera, front desk description: type: string maxLength: 500 provider: type: string enum: - elevenlabs - sarvam default: elevenlabs description: >- Where the clone is made, which decides the engines it speaks on. `elevenlabs` for the `classic` and `deepgram_agent` engines; `sarvam` for the `sarvam` engine, which is the one to use for Indian languages. language: type: string pattern: ^[a-z]{2}-[A-Z]{2}$ description: The language spoken in the recording. Required for `sarvam`. example: ml-IN remove_background_noise: type: boolean description: >- ElevenLabs only. Cleans up a noisy recording; it can make a clean one worse. consent: type: boolean description: >- Must be `true`. You confirm that the voice is yours, or that the speaker has given you permission to clone it and use it on calls. Recorded with the time and your account. audio_urls: type: array items: type: string maxLength: 2048 format: uri minItems: 1 maxItems: 5 description: >- JSON requests only: public https links to the recordings, fetched once and not stored. Multipart requests send the files themselves as `files` instead. required: - name - consent Voice: type: object properties: id: type: string pattern: ^[0-9a-f]{24}$ example: 6a96a3ead6e886d42462dd3e object: type: string enum: - voice name: type: string description: type: - string - 'null' provider: type: string enum: - elevenlabs - sarvam description: >- Where the clone is made, which decides the engines it speaks on. `elevenlabs` for the `classic` and `deepgram_agent` engines; `sarvam` for the `sarvam` engine, which is the one to use for Indian languages. provider_voice_id: type: string description: Put this in an agent’s `tts.voice` to make it speak in this voice. example: c38kUX8pkfYO2kHyqfFy engines: type: array items: type: string description: The engines that can speak in this voice. example: - classic - deepgram_agent language: type: - string - 'null' description: >- The language of the recording. Sarvam clones can still speak every language Sarvam supports. created_at: type: string format: date-time description: ISO 8601, always UTC. example: '2026-09-01T12:00:00.000Z' updated_at: type: string format: date-time description: ISO 8601, always UTC. example: '2026-09-01T12:00:00.000Z' required: - id - object - name - description - provider - provider_voice_id - engines - language - created_at - updated_at Error: type: object properties: error: type: object properties: type: type: string enum: - invalid_request_error - authentication_error - permission_error - rate_limit_error - insufficient_credits - not_found_error - conflict_error - api_error code: type: string description: Stable machine-readable code. example: agent_not_found message: type: string description: >- Human-readable explanation. Do not match on this — match on code. param: type: string description: Which field caused the failure, when applicable. doc_url: type: string request_id: type: string description: Quote this when asking for help. example: req_01HXYZ... required: - type - code - message - request_id required: - error securitySchemes: bearerAuth: type: http scheme: bearer description: >- Your Wixzel Voice API key: `Authorization: Bearer wv_live_...`. Keys are scoped; grant only what the integration needs. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.