mia/Aegisub
1
0
Fork 1
Code Issues Pull requests Projects Releases Wiki Activity

Have FFMS2 do the audio downmixing when possible

Browse source 
Aegisub's downmixer is terrible for anything with more than two channels
(averaging all the channels together is not even vaguely close to the
correct thing to do for 5.1), so libavresample should be far better.
This commit is contained in:
Thomas Goyne 2012-12-26 19:59:42 -08:00
parent 745b9ab2a9
commit 572a58cd90
1 changed files with 16 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
aegisub/src/audio_provider_ffmpegsource.cpp
View file

@ -173,6 +173,22 @@ void FFmpegSourceAudioProvider::LoadAudio(wxString filename) {
default:
throw agi::AudioProviderOpenError("unknown or unsupported sample format", 0);
}
#if FFMS_VERSION >= ((2 << 24) | (17 << 16) | (4 << 8) | 0)
if (channels > 1 || bytes_per_sample != 2) {
std::unique_ptr<FFMS_ResampleOptions, decltype(&FFMS_DestroyResampleOptions)>
opt(FFMS_CreateResampleOptions(AudioSource), FFMS_DestroyResampleOptions);
opt->ChannelLayout = FFMS_CH_FRONT_CENTER;
opt->SampleFormat = FFMS_FMT_S16;
// Might fail if FFMS2 wasn't built with libavresample
if (!FFMS_SetOutputFormatA(AudioSource, opt.get(), nullptr)) {
channels = 1;
bytes_per_sample = 2;
float_samples = false;
}
}
#endif
}
void FFmpegSourceAudioProvider::FillBuffer(void *Buf, int64_t Start, int64_t Count) const {