Sign in

v1

latestOpenAPI 3.1.02026-07-22122760.2 KB

Detect AI-generated music in an audio file

Accepts an audio file and returns a clip-level verdict on whether the content contains AI-generated music, along with a per-window breakdown of how each segment was classified.

post/api/velma-2-ai-music-detection-batch

Response

Detection completed successfully.

filenamestring required

Name of the submitted audio file. Empty string if no filename was provided in the upload.

duration_snumber double required

Total duration of the analysed audio in seconds.

primary_verdict'ai-vocal-music' | 'ai-instrumental' | 'not-ai-music' required

Clip-level classification:

  • ai-vocal-music - AI-generated music with a detected synthetic voice (covers AI songs and AI synthetic vocal tracks).
  • ai-instrumental - AI-generated instrumental music with no detectable synthetic voice.
  • not-ai-music - the clip does not appear to contain AI-generated music.
vocal_percentagenumber double required

Clip-level average percentage of the audio that contains vocal content, averaged across all windows.

vocal_ai_percentagenumber double required

Percentage of the clip duration classified as AI-generated vocals. Computed as (seconds of windows classified as AI vocals) / (total clip seconds) * 100. A window contributes its full duration when it is classified as AI-generated vocals; zero otherwise.

vocal_ai_confidencenumber double required

Average confidence that the vocal windows contain AI-generated vocals, across all windows with vocal content. Not diluted by non-vocal windows.

instrumental_percentagenumber double required

Clip-level average percentage of the audio that contains instrumental music content, averaged across all windows.

instrumental_ai_percentagenumber double required

Percentage of non-vocal non-silent windows classified as AI-generated instrumental content (0-100 scale).

instrumental_ai_confidencenumber double required

Maximum confidence that a non-vocal window contains AI-generated instrumental content. Zero if no such window was found.

silence_percentagenumber double required

Clip-level average percentage of the audio that contains neither vocal nor instrumental content, averaged across all windows.

latency_msnumber double required

End-to-end inference time in milliseconds.

Example response

{
  "filename": "my_audio.mp3",
  "duration_s": 89.28,
  "primary_verdict": "ai-vocal-music",
  "vocal_percentage": 87.5,
  "vocal_ai_percentage": 56.5,
  "vocal_ai_confidence": 0.89,
  "instrumental_percentage": 64.3,
  "instrumental_ai_percentage": 10.5,
  "instrumental_ai_confidence": 0.95,
  "silence_percentage": 3.51,
  "windows": [
    {
      "end_time_ms": 4000,
      "vocal_percentage": 100,
      "vocal_ai_percentage": 100,
      "vocal_ai_confidence": 0.97,
      "instrumental_percentage": 79
    }
  ],
  "latency_ms": 1333
}