Simplismart Speech to Text API

openapi: 3.1.0
info:
  title: DeepSeek R1 Distil Qwen-32B Chat Completion Speech to Text API
  description: High-capacity LLM for complex tasks
  version: 1.0.0
servers:
- url: https://api.simplismart.live
  description: DeepSeek-R1-Distill-Qwen-32B Proxy Server
security:
- BearerAuth: []
tags:
- name: Speech to Text
paths:
  /model/v2/infer/whisper:
    post:
      tags:
      - Speech to Text
      summary: Transcribe or translate audio using Whisper Large V2
      description: Process audio files for transcription or translation with advanced options
      operationId: transcribeAudioV2
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/WhisperV2Request'
            example:
              audio_data: base64_encoded_audio_content
              language: en
              task: transcribe
              beam_size: 5
              best_of: 5
              word_timestamps: 1
              diarization: 0
              streaming: 0
              batch_size: 24
              length_penalty: 1
              patience: 1
              vad_onset: 0.5
              vad_offset: 0.363
      responses:
        '200':
          description: Successful transcription
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/WhisperV2Response'
        '400':
          description: Invalid request parameters
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ErrorResponse'
        '401':
          description: Unauthorized - Invalid token
        '500':
          description: Internal server error
  /model/infer/whisper:
    post:
      tags:
      - Speech to Text
      summary: Transcribe or translate audio using Whisper Large V3
      description: 'Process audio files for transcription or translation with enhanced language support.

        Supports multiple audio formats and provides detailed word-level timestamps and speaker diarization.

        '
      operationId: transcribeAudioV3
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/WhisperV3Request'
            example:
              audio_data: base64_encoded_audio_content
              language: en
              task: transcribe
              word_timestamps: true
              diarization: false
              streaming: false
              batch_size: 24
              length_penalty: 1
              patience: 1
              vad_onset: 0.5
              vad_offset: 0.363
      responses:
        '200':
          description: Successful transcription
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/WhisperV3Response'
        '400':
          description: Invalid request parameters
        '401':
          description: Unauthorized - Invalid token
        '500':
          description: Internal server error
components:
  schemas:
    WhisperV2Response:
      type: object
      properties:
        transcription:
          type: array
          items:
            type: string
          description: Array of transcribed text segments
        segments:
          type: array
          items:
            type: object
            properties:
              start:
                type: number
                description: Start time of segment
              end:
                type: number
                description: End time of segment
              text:
                type: string
                description: Transcribed text
              words:
                type: array
                items:
                  type: object
                  properties:
                    word:
                      type: string
                    start:
                      type: number
                    end:
                      type: number
                description: Word-level timing information
        request_time:
          type: number
          description: Total processing time in seconds
        language:
          type: string
          description: Detected or specified language
    WhisperV3Request:
      type: object
      required:
      - audio_data
      - language
      - task
      properties:
        audio_data:
          type: string
          description: 'TEXT FIELD: This is a string field (not a file upload). Provide the audio as a base64-encoded string. First convert your audio file (.mp3, .wav, .flac) to base64, then paste the resulting string here.'
          example: base64_encoded_audio_content
        language:
          type: string
          description: Language code (e.g. 'en' for English)
          example: en
        task:
          type: string
          enum:
          - transcribe
          - translate
          description: Task type - transcribe in source language or translate to English
          example: transcribe
        initial_prompt:
          type: string
          description: Optional starting text prompt for context
          example: 'Meeting transcript between John and Sarah:'
        beam_size:
          type: integer
          description: Number of parallel sequences evaluated
          minimum: 1
          maximum: 5
          default: 5
        best_of:
          type: integer
          description: Number of best sequences considered
          minimum: 1
          maximum: 5
          default: 5
        word_timestamps:
          type: boolean
          description: Include word-level timestamps
          default: false
        diarization:
          type: boolean
          description: Enable speaker diarization
          default: false
        vad_filter:
          type: boolean
          description: Enable voice activity detection filter
          default: true
        without_timestamps:
          type: boolean
          description: Exclude timestamps from output
          default: false
        streaming:
          type: boolean
          description: Enable streaming output
          default: false
        min_speakers:
          type: integer
          description: Minimum number of speakers to detect (0 for automatic)
          default: 0
          minimum: 0
        max_speakers:
          type: integer
          description: Maximum number of speakers to detect (0 for automatic)
          default: 0
          minimum: 0
        batch_size:
          type: integer
          description: Number of audio samples processed in one batch
          minimum: 0
          maximum: 24
          default: 24
        length_penalty:
          type: number
          description: Penalty for longer sequences (1.0 means no penalty)
          default: 1
          minimum: 0
        patience:
          type: number
          description: Beam search patience factor
          minimum: 0
          maximum: 1
          default: 1
        min_duration_off:
          type: number
          description: Minimum duration of silence for a break (seconds)
          default: 0
          minimum: 0
        min_duration_on:
          type: number
          description: Minimum duration for speech detection (seconds)
          default: 0
          minimum: 0
        vad_onset:
          type: number
          description: Voice activity detection onset threshold
          minimum: 0
          maximum: 1
          default: 0.5
        vad_offset:
          type: number
          description: Voice activity detection offset threshold
          minimum: 0
          maximum: 1
          default: 0.363
        pad_offset:
          type: number
          description: Additional padding at segment end (seconds)
          default: 0
          minimum: 0
        pad_onset:
          type: number
          description: Additional padding at segment start (seconds)
          default: 0
          minimum: 0
        max_duration:
          type: number
          description: Maximum duration to process (seconds)
          default: 30
          minimum: 0
    WhisperV3Response:
      type: object
      required:
      - transcription
      - request_time
      - language
      properties:
        transcription:
          type: array
          items:
            type: string
          description: Array of transcribed text segments
          example:
          - Hello, this is a test.
          - The audio quality is good.
        segments:
          type: array
          items:
            type: object
            required:
            - start
            - end
            - text
            properties:
              start:
                type: number
                description: Start time of segment (seconds)
                example: 0
              end:
                type: number
                description: End time of segment (seconds)
                example: 2.5
              text:
                type: string
                description: Transcribed text
                example: Hello, this is a test.
              words:
                type: array
                items:
                  type: object
                  required:
                  - word
                  - start
                  - end
                  properties:
                    word:
                      type: string
                      example: Hello
                    start:
                      type: number
                      example: 0
                    end:
                      type: number
                      example: 0.5
                description: Word-level timing information
        request_time:
          type: number
          description: Total processing time in seconds
          example: 2.5
        language:
          type: string
          description: Detected or specified language
          example: en
    ErrorResponse:
      type: object
      properties:
        error:
          type: string
          description: Error message
        status:
          type: integer
          description: Error status code
    WhisperV2Request:
      type: object
      required:
      - audio_data
      - language
      - task
      properties:
        audio_data:
          type: string
          description: 'TEXT FIELD: This is a string field (not a file upload). Provide the audio as a base64-encoded string. First convert your audio file (.mp3, .wav, .flac) to base64, then paste the resulting string here.'
        language:
          type: string
          description: Language code (e.g. 'en' for English)
          example: en
        task:
          type: string
          enum:
          - transcribe
          - translate
          description: Task type - transcribe in source language or translate to English
          example: transcribe
        initial_prompt:
          type: string
          description: Optional starting text prompt for context
        beam_size:
          type: integer
          description: Number of parallel sequences evaluated
          minimum: 1
          maximum: 5
          default: 5
        best_of:
          type: integer
          description: Number of best sequences considered
          minimum: 1
          maximum: 5
          default: 5
        word_timestamps:
          type: integer
          enum:
          - 0
          - 1
          description: Include word-level timestamps (0=false, 1=true)
          default: 0
        diarization:
          type: integer
          enum:
          - 0
          - 1
          description: Enable speaker diarization (0=false, 1=true)
          default: 0
        vad_filter:
          type: integer
          enum:
          - 0
          - 1
          description: Enable voice activity detection filter (0=false, 1=true)
          default: 1
        without_timestamps:
          type: integer
          enum:
          - 0
          - 1
          description: Exclude timestamps from output (0=false, 1=true)
          default: 0
        streaming:
          type: integer
          enum:
          - 0
          - 1
          description: Enable streaming output (0=false, 1=true)
          default: 0
        min_speakers:
          type: number
          description: Minimum number of speakers to detect
          default: 0
        max_speakers:
          type: number
          description: Maximum number of speakers to detect
          default: 0
        batch_size:
          type: integer
          description: Number of audio samples processed in one batch
          minimum: 0
          maximum: 24
          default: 24
        length_penalty:
          type: number
          description: Penalty for longer sequences
          default: 1
        patience:
          type: number
          description: Beam search patience factor
          minimum: 0
          maximum: 1
          default: 1
        min_duration_off:
          type: number
          description: Minimum duration of silence for a break
          default: 0
        min_duration_on:
          type: number
          description: Minimum duration for speech detection
          default: 0
        vad_onset:
          type: number
          description: Voice activity detection onset threshold
          minimum: 0
          maximum: 1
          default: 0.5
        vad_offset:
          type: number
          description: Voice activity detection offset threshold
          minimum: 0
          maximum: 1
          default: 0.363
        pad_offset:
          type: number
          description: Additional padding at segment end
          default: 0
        pad_onset:
          type: number
          description: Additional padding at segment start
          default: 0
        max_duration:
          type: number
          description: Maximum duration to process
          default: 30
  securitySchemes:
    BearerAuth:
      type: http
      scheme: bearer
      bearerFormat: JWT
      description: JWT token for authentication