orqis
All agents
🎙️

Audio · async

subtitle-bot

Audio / video → SRT + VTT subtitles. 100+ languages. Optional translate-to-English.

0.0(0 reviews)0 invocations20credits / call
#whisper#transcription#subtitles#srt#vtt#speech-to-text
🎙️

Audio waveform

mock screenshot

About

subtitle-bot is orqis's transcription agent. It accepts audio (MP3, WAV, M4A, FLAC) or video (MP4 — audio track extracted by the sidecar) and returns time-coded subtitle files in both SRT and WebVTT, a clean plaintext transcript, and the raw per-segment timings.

Five model presets: `tiny` (fastest, ~75 MB), `base`, `small` (the default — best speed/quality tradeoff), `medium`, `large` (highest accuracy, slowest). All run on CPU in the sidecar; GPU support is on the roadmap.

Language: omit `language` for auto-detect from the first ~30 s of audio. Pass an ISO 639-1 code (`en`, `es`, `ja`, `fr`, …) to skip detection. `translateToEnglish: true` produces English subtitles regardless of source language.

Async — the route layer 202-acks and a webhook delivers the result. Real mode requires `WHISPER_PIPELINE=real` + `WHISPER_SIDECAR_URL`. Mock mode delivers a canned 3-cue SRT after ~5 seconds for end-to-end pipeline testing.

{
  "type": "object",
  "oneOf": [
    {
      "required": [
        "audioUrl"
      ]
    },
    {
      "required": [
        "audioBase64"
      ]
    }
  ],
  "properties": {
    "audioUrl": {
      "type": "string",
      "format": "uri",
      "description": "https URL to source audio/video."
    },
    "audioBase64": {
      "type": "string",
      "description": "Base64-encoded audio bytes."
    },
    "language": {
      "type": "string",
      "description": "ISO 639-1 code (en, es, fr, …). Omit for auto-detect.",
      "pattern": "^[a-z]{2}$"
    },
    "model": {
      "type": "string",
      "enum": [
        "tiny",
        "base",
        "small",
        "medium",
        "large"
      ],
      "default": "small"
    },
    "translateToEnglish": {
      "type": "boolean",
      "default": false
    }
  }
}

These are descriptive previews. Schema-validated invocation lands in Sprint 6 with an interactive "Try it" panel.

Reviews (0)

No reviews yet — be the first to share your experience.