geopti/sam-audio-base 🖼️📝✓ → 🖼️

▶️ 327 runs 📅 Feb 2026 ⚙️ Cog 0.16.11 🔗 GitHub 📄 Paper ⚖️ License
audio-to-audio music-source-separation

Performance

2.5sTypical run time
~214sCold start (first call)
327Total runs

About

A foundation model for isolating any sound in audio using text, visual, or temporal prompts

Example Output

Output

Example output

Performance Metrics

2.46s Prediction Time
214.07s Total Time
All Input Parameters 
{
  "audio": "https://replicate.delivery/pbxt/OXLmuvymOhyvr27oioBOSl5xoqb6X9C8oXDd7jSEWs3SU27N/test.m4a",
  "description": "speech",
  "span_anchors": "[]",
  "predict_spans": false,
  "output_residual": false,
  "use_span_prompting": false
}
Input Parameters
audio (required) Type: string
Input audio or video file (WAV, MP3, MP4, etc.)
description Type: stringDefault: speech
Text description of the sound to isolate. Use simple noun phrases like 'speech', 'man speaking', 'dog barking', 'piano', 'guitar playing', 'birds chirping'
span_anchors Type: stringDefault: []
[Only if use_span_prompting=True] Time ranges as JSON array. Format: [['+', start_sec, end_sec], ...]. '+' means sound present, '-' means absent. Example: [['+', 2.0, 4.0]] or [['+', 1.0, 3.0], ['-', 5.0, 6.0]]
predict_spans Type: booleanDefault: false
Auto-detect time spans where target sound occurs. Improves quality for non-ambient sounds but slower.
output_residual Type: booleanDefault: false
Also output the residual audio (everything except the target sound)
use_span_prompting Type: booleanDefault: false
Enable span prompting to specify time ranges where the target sound occurs. More precise but requires knowing timestamps.
Output Schema

Output

Type: array • Items Type: string • Items Format: uri

Version Details
Version ID
b84861ae9b787409ef92927b5a07704fda87a0a7762e9bb7b09c517357eadb53
Version Created
February 4, 2026
Run on Replicate →