diff options
| author | Martin Storsjö <martin@martin.st> | 2012-03-02 17:03:06 +0200 |
|---|---|---|
| committer | Martin Storsjö <martin@martin.st> | 2012-03-02 18:58:19 +0200 |
| commit | b087ce2bee81db8cc5caffb8f0a4f6c7c92a30fe (patch) | |
| tree | 33d6290462f3f1f41bf18ed00403a7a176ba6697 /libavcodec/g722enc.c | |
| parent | 56bf24ad7821974bc92a2d02a120165ee61efb90 (diff) | |
g722: Fix the QMF scaling
This fixes clipping if the encoder input used the full 16 bit
input range (samples with a magnitude below 16383 worked fine).
The filtered subband samples should be 15 bit maximum, while
the code earlier produced them scaled to 16 bit.
This makes the decoder output have double the magnitude
compared to before.
The spec reference samples doesn't test the QMF at all, which
was why this part slipped past initially.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diffstat (limited to 'libavcodec/g722enc.c')
| -rw-r--r-- | libavcodec/g722enc.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/libavcodec/g722enc.c b/libavcodec/g722enc.c index a5ae0a5153..ba8ceeff86 100644 --- a/libavcodec/g722enc.c +++ b/libavcodec/g722enc.c @@ -136,8 +136,8 @@ static inline void filter_samples(G722Context *c, const int16_t *samples, c->prev_samples[c->prev_samples_pos++] = samples[0]; c->prev_samples[c->prev_samples_pos++] = samples[1]; ff_g722_apply_qmf(c->prev_samples + c->prev_samples_pos - 24, &xout1, &xout2); - *xlow = xout1 + xout2 >> 13; - *xhigh = xout1 - xout2 >> 13; + *xlow = xout1 + xout2 >> 14; + *xhigh = xout1 - xout2 >> 14; if (c->prev_samples_pos >= PREV_SAMPLES_BUF_SIZE) { memmove(c->prev_samples, c->prev_samples + c->prev_samples_pos - 22, |