eden-emu/opus
20
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix quantization bug where pitch can get too low

Browse source 
Would cause unused pitch embedding vectors to be used
This commit is contained in:
Jean-Marc Valin 2021-01-10 03:15:30 -05:00
parent f0df3e82ec
commit 56d9f13efd
3 changed files with 4 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

1
dnn/lpcnet.c
View file

@ -130,6 +130,7 @@ LPCNET_EXPORT void lpcnet_synthesize(LPCNetState *lpcnet, const float *features,
float pitch_gain;
/* Matches the Python code -- the 0.1 avoids rounding issues. */
pitch = (int)floor(.1 + 50*features[36]+100);
pitch = IMIN(255, IMAX(33, pitch));
pitch_gain = lpcnet->old_gain[FEATURES_DELAY-1];
memmove(&lpcnet->old_gain[1], &lpcnet->old_gain[0], (FEATURES_DELAY-1)*sizeof(lpcnet->old_gain[0]));
lpcnet->old_gain[0] = features[PITCH_GAIN_FEATURE];