Add SILK gain quantization#162
Merged
Merged
Conversation
Co-authored-by: François Allais <francois.allais@hotmail.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #162 +/- ##
==========================================
+ Coverage 89.27% 89.31% +0.04%
==========================================
Files 44 45 +1
Lines 8783 8817 +34
==========================================
+ Hits 7841 7875 +34
Misses 707 707
Partials 235 235
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
FrantaBOT
approved these changes
Jul 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
adds
quantizeGains— per-subframe log-domain gain quantization with hysteresis (silk_gains_quant): independent 6-bit gain index for the first subframe of a frame (when not delta-coded), delta index otherwise, with a double-step mechanism so large gain jumps can still be reached within the delta rangeported from
gain_quant.cthis one needed a small structural change, not just a straight extraction:
quantizeGainswas a method on*Encoder(kept the runningpreviousLogGainon the struct), but that type doesn't exist in this package yet — same blocker asfind_pitch_lags.go/encodeNLSF/quantLTPGainsbefore it. Instead of deferring the whole file (there'd be nothing left for this PR), I turned it into a plain function that takespreviousLogGainas an explicit*int32in/out parameter, mirroring the patternpitchAnalysisCorealready uses forltpCorrin this same package.encodeSubframeGains/emitGainIndices(the range-coder wiring) are what's still deferred to the orchestration piecetests replay the decoder's reconstruction math directly against the returned indices instead of a full Encoder→bitstream→Decoder round trip, same approach as the NLSF piece (#160)
Reference issue
part of the SILK encoder series (#160)