> ## Documentation Index > Fetch the complete documentation index at: https://developers.liara.ir/llms.txt > Use this file to discover all available pages before exploring further. # Transcribe audio > Transcribes the provided audio using the specified speech-to-text model. The API accepts audio in either JSON or multipart/form-data format. For JSON requests, provide the audio as a Base64-encoded string through the `input_audio.data` field and specify the audio format through `input_audio.format`. For multipart/form-data requests, upload the audio file using the `file` field. The request is billed according to the selected speech-to-text model and the usage associated with the transcription request. ## OpenAPI ````yaml /spec/ai-stt.yaml post /api/{workspaceID}/v1/audio/transcriptions openapi: 3.0.1 info: title: AI - Audio Transcriptions description: |- Audio transcription API for converting audio into text using supported speech-to-text AI models. The API supports both JSON and multipart/form-data requests. For JSON requests, the audio must be provided as a Base64-encoded value in the `input_audio` field. For multipart/form-data requests, the audio file must be provided in the `file` field. Parameters: - `workspaceID`: The ID of the workspace to use for billing and routing. Authentication: API Key (Bearer token) sent via the `Authorization` header. The transcription result is returned as a JSON object. termsOfService: '#' contact: email: info@liara.ir version: 1.0.0 servers: - url: https://ai.liara.ir security: - apiKey: [] tags: - name: Audio Transcriptions description: Transcribe audio into text using AI models externalDocs: description: Find out more about Liara AI url: https://liara.ir paths: /api/{workspaceID}/v1/audio/transcriptions: post: tags: - Audio Transcriptions summary: Transcribe audio description: |- Transcribes the provided audio using the specified speech-to-text model. The API accepts audio in either JSON or multipart/form-data format. For JSON requests, provide the audio as a Base64-encoded string through the `input_audio.data` field and specify the audio format through `input_audio.format`. For multipart/form-data requests, upload the audio file using the `file` field. The request is billed according to the selected speech-to-text model and the usage associated with the transcription request. operationId: createTranscription parameters: - name: workspaceID in: path required: true description: The workspace ID used for billing and routing. schema: type: string pattern: ^[a-f0-9]{24}$ requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/STTRequestJSON' example: model: openai/whisper-1 input_audio: data: UklGRiQAAABXQVZFZm10IBAAAAABAAEA... format: wav language: en response_format: json temperature: 0 timestamp_granularities: - segment multipart/form-data: schema: $ref: '#/components/schemas/STTRequestMultipart' encoding: file: contentType: >- audio/wav, audio/mpeg, audio/flac, audio/mp4, audio/ogg, audio/webm, audio/aac example: model: openai/whisper-1 language: en response_format: json temperature: 0 timestamp_granularities: word,segment responses: '200': description: Successfully transcribed the audio. headers: X-Request-Id: description: ID of the request log. schema: type: string X-Generation-Id: description: >- Generation ID returned by the transcription provider, when available. schema: type: string X-Provider-Request-Id: description: >- Request ID returned by the transcription provider, when available. schema: type: string X-Usage-Estimated-Cost: description: Estimated cost of the transcription request. schema: type: string content: application/json: schema: $ref: '#/components/schemas/TranscriptionResponse' examples: json: summary: Standard transcription response value: text: Hello, this is a transcription test. verbose_json: summary: Verbose transcription response value: task: transcribe language: english duration: 3.52 text: Hello, this is a transcription test. segments: - id: 0 start: 0 end: 3.52 text: Hello, this is a transcription test. '400': description: Bad request. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '401': description: Missing or invalid authentication. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '402': description: Payment required - insufficient balance. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '403': description: Forbidden - workspace is unavailable or access is denied. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '502': description: Speech-to-text provider request failed. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '503': description: Service unavailable. content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' components: schemas: STTRequestJSON: type: object required: - model - input_audio properties: model: type: string description: |- The speech-to-text model to use. The model must be one of the speech-to-text models supported by the API. example: openai/whisper-1 input_audio: $ref: '#/components/schemas/InputAudio' language: type: string minLength: 2 maxLength: 2 description: |- The language of the input audio. Must be a two-letter ISO-639-1 language code. If omitted, the model may automatically detect the language. example: en response_format: type: string enum: - json - verbose_json default: json description: |- The format of the transcription response. `json` returns the standard transcription response. `verbose_json` returns additional transcription information, such as language, duration, and timestamp information when supported by the selected model. example: json temperature: type: number format: float minimum: 0 maximum: 1 description: |- Sampling temperature used during transcription. Values must be between 0 and 1. example: 0 timestamp_granularities: type: array description: |- Specifies the timestamp granularities to include in the transcription response. Supported values are `word` and `segment`. items: type: string enum: - word - segment example: - segment STTRequestMultipart: type: object required: - model - file properties: model: type: string description: |- The speech-to-text model to use. The model must be one of the speech-to-text models supported by the API. example: openai/whisper-1 file: type: string format: binary description: |- The audio file to transcribe. The file must be provided using the `file` field. Supported audio formats are: WAV, MP3, FLAC, M4A, OGG, WebM, and AAC. language: type: string minLength: 2 maxLength: 2 description: |- The language of the input audio. Must be a two-letter ISO-639-1 language code. example: en response_format: type: string enum: - json - verbose_json default: json description: |- The format of the transcription response. `json` returns the standard transcription response. `verbose_json` returns additional transcription information. example: json temperature: type: number format: float minimum: 0 maximum: 1 description: |- Sampling temperature used during transcription. Values must be between 0 and 1. example: 0 timestamp_granularities: type: string description: |- Specifies the timestamp granularities to include in the transcription response. For multipart/form-data requests, multiple values must be provided as a comma-separated string. Supported values are `word` and `segment`. Example: `word,segment` example: word,segment TranscriptionResponse: type: object description: |- Transcription result returned by the speech-to-text model. The exact fields may vary depending on the selected `response_format` and the capabilities of the selected model. properties: text: type: string description: The transcribed text. example: Hello, this is a transcription test. task: type: string description: The transcription task. example: transcribe language: type: string description: Detected or requested language. example: english duration: type: number format: float description: Duration of the input audio in seconds. example: 3.52 segments: type: array description: |- Transcription segments with timing information. Available when supported by the selected model and when requested through `timestamp_granularities`. items: $ref: '#/components/schemas/TranscriptionSegment' ErrorResponse: type: object properties: statusCode: type: integer description: HTTP status code. error: type: string description: Error type. message: type: string description: Error message. InputAudio: type: object required: - data - format description: |- Audio data provided as a Base64-encoded string. This object is used for JSON requests. Multipart requests should provide the audio through the `file` field instead. properties: data: type: string format: byte description: |- Base64-encoded audio data. The encoded audio must use one of the supported formats: WAV, MP3, FLAC, M4A, OGG, WebM, or AAC. example: UklGRiQAAABXQVZFZm10IBAAAAABAAEA... format: type: string enum: - wav - mp3 - flac - m4a - ogg - webm - aac description: The format of the Base64-encoded audio data. example: wav TranscriptionSegment: type: object properties: id: type: integer description: Segment identifier. example: 0 start: type: number format: float description: Start time of the segment in seconds. example: 0 end: type: number format: float description: End time of the segment in seconds. example: 3.52 text: type: string description: Transcribed text for the segment. example: Hello, this is a transcription test. securitySchemes: apiKey: type: apiKey description: >- Enter the API key with the `Bearer: ` prefix, e.g. "Bearer " name: Authorization in: header ````