RepoMirrors
/
ffmpeg
mirror of https://git.ffmpeg.org/ffmpeg.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

libvorbis: use float input instead of s16 

libvorbis takes float input, so we can just deinterleave/reorder the input
as-is instead of also converting.
This commit is contained in:
Justin Ruggles 2012-02-28 19:55:10 -05:00
parent f15c4281dc
commit c5063e0348
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
libavcodec/libvorbis.c
View File

@ -236,7 +236,7 @@ static int oggvorbis_encode_frame(AVCodecContext *avctx, unsigned char *packets,
{
OggVorbisContext *s = avctx->priv_data;
ogg_packet op;
signed short *audio = data;
float *audio = data;
int pkt_size;
/* send samples to libvorbis */
@ -251,7 +251,7 @@ static int oggvorbis_encode_frame(AVCodecContext *avctx, unsigned char *packets,
int co = (channels > 8) ? c :
ff_vorbis_encoding_channel_layout_offsets[channels - 1][c];
for (i = 0; i < samples; i++)
buffer[c][i] = audio[i * channels + co] / 32768.f;
buffer[c][i] = audio[i * channels + co];
}
vorbis_analysis_wrote(&s->vd, samples);
} else {
@ -315,7 +315,7 @@ AVCodec ff_libvorbis_encoder = {
.encode = oggvorbis_encode_frame,
.close = oggvorbis_encode_close,
.capabilities = CODEC_CAP_DELAY,
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16,
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLT,
AV_SAMPLE_FMT_NONE },
.long_name = NULL_IF_CONFIG_SMALL("libvorbis Vorbis"),
.priv_class = &class,