openapi: 3.1.0
info:
  title: ElevenLabs Audio Isolation Agents Speech to Text API
  description: The ElevenLabs Audio Isolation API removes background noise from audio recordings, isolating vocal tracks from ambient sounds and interference. This is useful for cleaning up recordings, improving audio quality for podcasts and interviews, and preparing audio files for further processing such as voice cloning or transcription.
  version: '1.0'
  contact:
    name: ElevenLabs Support
    url: https://help.elevenlabs.io
  termsOfService: https://elevenlabs.io/terms-of-service
servers:
- url: https://api.elevenlabs.io
  description: Production Server
security:
- apiKeyAuth: []
tags:
- name: Speech to Text
  description: Endpoints for converting audio into text transcriptions with support for multiple languages and audio formats.
paths:
  /v1/speech-to-text:
    post:
      operationId: convertSpeechToText
      summary: Convert speech to text
      description: Transcribes an audio file into text. Supports multiple audio formats and languages. Returns the transcribed text along with optional word-level timing information.
      tags:
      - Speech to Text
      requestBody:
        required: true
        content:
          multipart/form-data:
            schema:
              $ref: '#/components/schemas/SpeechToTextRequest'
      responses:
        '200':
          description: Transcription completed successfully
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/TranscriptionResponse'
        '400':
          description: Bad request - invalid audio file or parameters
        '401':
          description: Unauthorized - invalid or missing API key
        '422':
          description: Unprocessable entity - validation error
  /v1/speech-to-text/async:
    post:
      operationId: convertSpeechToTextAsync
      summary: Convert speech to text asynchronously
      description: Submits an audio file for asynchronous transcription. Returns a task identifier that can be used to poll for the transcription result or receive it via webhook callback.
      tags:
      - Speech to Text
      requestBody:
        required: true
        content:
          multipart/form-data:
            schema:
              $ref: '#/components/schemas/SpeechToTextAsyncRequest'
      responses:
        '200':
          description: Transcription task submitted successfully
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/AsyncTranscriptionResponse'
        '400':
          description: Bad request - invalid audio file or parameters
        '401':
          description: Unauthorized - invalid or missing API key
        '422':
          description: Unprocessable entity - validation error
components:
  schemas:
    TranscriptionResponse:
      type: object
      properties:
        text:
          type: string
          description: The full transcribed text.
        language_code:
          type: string
          description: The detected language of the audio content.
        language_probability:
          type: number
          description: Confidence score for the detected language.
        words:
          type: array
          description: Word-level timing information when timestamps are requested.
          items:
            $ref: '#/components/schemas/TranscriptionWord'
        audio_events:
          type: array
          description: Non-speech audio events detected in the recording.
          items:
            $ref: '#/components/schemas/AudioEvent'
    SpeechToTextRequest:
      type: object
      required:
      - file
      properties:
        file:
          type: string
          format: binary
          description: The audio file to transcribe. Supports formats including MP3, WAV, FLAC, OGG, and M4A.
        model_id:
          type: string
          description: The identifier of the speech-to-text model to use for transcription.
        language_code:
          type: string
          description: Language code in ISO 639-1 format to hint the expected language of the audio content.
        tag_audio_events:
          type: boolean
          description: Whether to tag non-speech audio events such as music, laughter, or applause in the transcription output.
          default: false
        timestamps_granularity:
          type: string
          description: The level of timestamp granularity to include in the response.
          enum:
          - none
          - word
          - character
    SpeechToTextAsyncRequest:
      type: object
      required:
      - file
      properties:
        file:
          type: string
          format: binary
          description: The audio file to transcribe asynchronously.
        model_id:
          type: string
          description: The identifier of the speech-to-text model to use.
        language_code:
          type: string
          description: Language code in ISO 639-1 format.
        webhook_url:
          type: string
          format: uri
          description: A URL to receive a webhook notification when the transcription is complete.
    AudioEvent:
      type: object
      properties:
        type:
          type: string
          description: The type of audio event detected.
        start:
          type: number
          description: Start time of the event in seconds.
        end:
          type: number
          description: End time of the event in seconds.
    AsyncTranscriptionResponse:
      type: object
      properties:
        task_id:
          type: string
          description: The identifier of the asynchronous transcription task.
        status:
          type: string
          description: The current status of the transcription task.
          enum:
          - pending
          - processing
          - completed
          - failed
    TranscriptionWord:
      type: object
      properties:
        text:
          type: string
          description: The transcribed word.
        start:
          type: number
          description: Start time of the word in seconds.
        end:
          type: number
          description: End time of the word in seconds.
        confidence:
          type: number
          description: Confidence score for the word transcription.
          minimum: 0
          maximum: 1
  securitySchemes:
    apiKeyAuth:
      type: apiKey
      in: header
      name: xi-api-key
      description: ElevenLabs API key passed in the xi-api-key header for authentication.
externalDocs:
  description: ElevenLabs Audio Isolation API Documentation
  url: https://elevenlabs.io/docs/api-reference/audio-isolation/audio-isolation