Audio Analysis
PlannedWorking now
- Browser-side upload validation (type/size)
- Measurement table that shows NOT MEASURED until a real analyzer runs
Not yet implemented
- Server decode + signal processing (ffmpeg/ebur128 or equivalent)
- Waveform viewer (wavesurfer.js) and A/B playback
- Writing audio_analysis_results with method/units/confidence
Upload (validation only)
| Duration | NOT MEASURED |
| Peak level | NOT MEASURED |
| RMS level | NOT MEASURED |
| Integrated loudness (LUFS) | NOT MEASURED |
| True peak | NOT MEASURED |
| Loudness range | NOT MEASURED |
| Stereo balance | NOT MEASURED |
| Low-frequency energy | NOT MEASURED |
| Clipping | NOT MEASURED |
| Estimated tempo | NOT MEASURED |
| Estimated sections | NOT MEASURED |
A language model is never asked to invent LUFS, True Peak or BPM. These fields only populate from a real measurement with method, units, date and confidence.