RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-27 05:57:45 +00:00
Code Issues Activity Actions

LibAudio: Prevent overflows during prediction

Browse source 
Saturating arithmetic leads to less screwed up audio in these cases.
This commit is contained in:
kleines Filmröllchen 2023-08-19 20:01:37 +02:00 • committed by Andrew Kaster
parent 0347d04289
commit b432674923
2 changed files with 30 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
Userland/Libraries/LibAudio/FlacLoader.cpp
View file

@ -797,16 +797,16 @@ ErrorOr<void, LoaderError> FlacLoaderPlugin::decode_custom_lpc(Vector<i64>& deco
// approximate the waveform with the predictor
for (size_t i = subframe.order; i < m_current_frame->sample_count; ++i) {
// (see below)
i64 sample = 0;
Checked<i64> sample = 0;
for (size_t t = 0; t < subframe.order; ++t) {
// It's really important that we compute in 64-bit land here.
// Even though FLAC operates at a maximum bit depth of 32 bits, modern encoders use super-large coefficients for maximum compression.
// These will easily overflow 32 bits and cause strange white noise that abruptly stops intermittently (at the end of a frame).
// The simple fix of course is to do intermediate computations in 64 bits.
// The simple fix of course is to do intermediate computations in 64 bits, but we additionally use saturating arithmetic.
// These considerations are not in the original FLAC spec, but have been added to the IETF standard: https://datatracker.ietf.org/doc/html/draft-ietf-cellar-flac-03#appendix-A.3
sample += static_cast<i64>(coefficients[t]) * static_cast<i64>(decoded[i - t - 1]);
sample.saturating_add(Checked<i64>::saturating_mul(static_cast<i64>(coefficients[t]), static_cast<i64>(decoded[i - t - 1])));
}
decoded[i] += sample >> lpc_shift;
decoded[i] += sample.value() >> lpc_shift;
}
return {};