kdeps-components-transcribe_audio

version: 1.1.1
requires:
    schemas/v1: ""
    tools/transcribe_audio: ""
files:
    component.yaml: |
        apiVersion: kdeps.io/v1
        kind: Component
        metadata:
          name: transcribe_audio
          description: Transcribe audio or video through a Whisper API with the kdeps API
          version: "1.1.1"
        interface:
          inputs:
            - name: file
              type: string
              required: true
              description: Absolute path to the audio/video file (mp3, mp4, mpeg, mpga, m4a, wav, webm)
            - name: backend
              type: string
              default: openai
              description: 'Provider: openai, groq, or local (a self-hosted OpenAI-compatible Whisper server)'
            - name: model
              type: string
              default: whisper-1
              description: Transcription model (groq uses whisper-large-v3)
            - name: language
              type: string
              default: ""
              description: ISO-639-1 code, e.g. en; empty = auto-detect
            - name: baseURL
              type: string
              description: Overrides the backend API base URL (the .../v1 part)
            - name: prompt
              type: string
              description: Spelling hints for names and terms
            - name: responseFormat
              type: string
              default: text
              description: text, json, verbose_json, srt or vtt
            - name: temperature
              type: string
              description: Sampling temperature 0-1
          # body is the transcribed text. The HTTP envelope stays inside the call.
          output: "{{ output('call').body }}"
        resources:
          - actionId: call
            name: Transcribe
            httpClient:
              method: POST
              # In-process host. The HTTP executor calls the handler without dialing.
              url: http://kdeps.api/apis/tools.kdeps.io/v1/tools/transcribe_audio/call
              headers:
                Content-Type: application/json
              failOnError: true
              timeout: 300s
              data:
                file: "{{ get('transcribe_audio.file') }}"
                backend: "{{ get('transcribe_audio.backend') }}"
                model: "{{ get('transcribe_audio.model') }}"
                language: "{{ get('transcribe_audio.language') }}"
                baseURL: "{{ get('transcribe_audio.baseURL') }}"
                prompt: "{{ get('transcribe_audio.prompt') }}"
                responseFormat: "{{ get('transcribe_audio.responseFormat') }}"
                temperature: "{{ get('transcribe_audio.temperature') }}"