v1

latestOpenAPI 3.1.02026-07-22122760.2 KB

Detect synthetic voice in an audio file

Accepts a single audio file and returns a per-frame synthetic-voice classification across the clip as a synchronous JSON response. Recommended audio length: 4-60 seconds; files shorter than 0.5 seconds are rejected with a 422.

post/api/velma-2-synthetic-voice-detection-batch

Response

Detection completed successfully.

filenamestring nullable required

Original filename of the uploaded audio.

duration_msinteger required

Total duration of the audio in milliseconds.

Example response

{
  "filename": "recording.wav",
  "frames": [
    {
      "end_time_ms": 3000,
      "verdict": "synthetic",
      "confidence": 0.9732
    }
  ],
  "duration_ms": 34500
}