Find the time of each beep in a clip's sound
0:20owner-made · agentic-video gallery (owner-made)
Proof frames
Tool calls
audio_onsets
Arguments
{
"source": "$CLIP",
"band_hz": [
900,
1100
],
"min_strength": 0.7
}
Result
{
"ok": true,
"data": {
"source_path": "~/work/cards/beeps-screen-recording/clip.mp4",
"start_sec": 0,
"end_sec": 20.015604,
"sample_rate": 48000,
"hop_samples": 256,
"time_resolution_ms": 5.333,
"min_strength": 0.7,
"band_hz": {
"name": null,
"low_hz": 900,
"high_hz": 1100
},
"compare_to": null,
"compare_min_delta_db": null,
"onsets": [
{
"time_sec": 3.717333,
"strength": 1,
"band_db": {
"sub": -57.326,
"low_mid": -51.443,
"mid": -3.158,
"high": -70.593,
"air": -92.239
},
"band_energy_db": 31.186
},
{
"time_sec": 4.453333,
"strength": 1,
"band_db": {
"sub": -51.222,
"low_mid": -16.722,
"mid": -43.994,
"high": -65.362,
"air": -84.413
},
"band_energy_db": -14.25
},
{
"time_sec": 7.482667,
"strength": 1,
"band_db": {
"sub": -52.883,
"low_mid": -47.026,
"mid": -3.175,
"high": -66.135,
"air": -88.121
},
"band_energy_db": 31.168
},
{
"time_sec": 8.869333,
"strength": 1,
"band_db": {
"sub": -86.859,
"low_mid": -72.671,
"mid": -95.065,
"high": -104.602,
"air": -116.447
},
"band_energy_db": -69.742
},
{
"time_sec": 14.16,
"strength": 1,
"band_db": {
"sub": -72.461,
"low_mid": -64.192,
"mid": -91.939,
"high": -106.938,
"air": -118.364
},
"band_energy_db": -68.547
},
{
"time_sec": 19.52,
"strength": 1,
"band_db": {
"sub": -51.679,
"low_mid": -20.375,
"mid": -39.847,
"high": -56.813,
"air": -82.447
},
"band_energy_db": -10.849
}
],
"total": 6,
"merged_count": 0
}
}
Question and answer
- Question
- When does each beep sound in this video?
- Answer
- The tool reports sound onsets at 3.72, 4.45, 7.48, 8.87, 14.16, 19.52 s in the 900 to 1100 Hz band.
- Tool returned
- [3.717333,4.453333,7.482667,8.869333,14.16,19.52]
- Truth
- [1.872,3.726,5.643,7.494,9.553]
Check failedprecision 0.33 (min 1), recall 0.40 (min 1), tolerance 0.1s
How it was checked
- Truth from
- constructed
- Fixed at
Planted by this recipe before any tool ran: 5 sine beeps at 1000 Hz, 0.1 s long and amplitude 0.8, mixed over the clip's own sound (kept at 0.5 times its volume) and started at 1.872, 3.726, 5.643, 7.494, 9.553 s. The truth is these planted start times; the tolerance 0.1 s allows for the tool's frame timing.
The script mixed 5 beeps into the clip's own sound at set times, so the answer is those set times. Any other sound in the band that the tool reports is counted as wrong.
Try it yourself
Using agentic-video, find the short 1000 Hz beeps in the sound of my video and give me the time of each one in seconds. Use my own file, and look only at the 900 to 1100 Hz band.
- Get the .mcpb file and open it in Claude Desktop.
- Or:
pip install agentic-video
Clip and licence
- Licence
- owner-made
Generated for the gallery; no third-party material.