Switched from Uint8Array to Float32Array for level monitoring
to get accurate, full-precision audio measurements.
The Problem:
- getByteTimeDomainData() uses Uint8Array (0-255)
- Byte conversion: (value - 128) / 128 has asymmetric range
- Positive peaks: (255-128)/128 = 0.992 (not full 1.0)
- Precision loss from byte quantization
- Mastered tracks with peaks at 0dBFS only showed ~50%
The Solution:
- Switched to getFloatTimeDomainData() with Float32Array
- Returns actual sample values directly in -1.0 to +1.0 range
- No conversion needed, no precision loss
- Accurate representation of audio peaks
Changes Applied:
- useMultiTrackPlayer: Float32Array with analyser.fftSize samples
- useRecording: Float32Array with analyser.fftSize samples
- Peak detection: Math.abs() on float values directly
Benefits:
✅ Full 0-100% range for properly mastered audio
✅ Higher precision (32-bit float vs 8-bit byte)
✅ Symmetric range (-1.0 to +1.0, not -1.0 to ~0.992)
✅ Accurate metering for professional audio files
Now mastered tracks with peaks at 0dBFS will correctly show
~100% on the meters instead of being capped at 50%.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Sebastian Krüger
a0ce83a654