TakeV/dxx-rebirth
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

Add back old SDL_mixer-based resampler

Browse source 
Add the SDL_mixer-based resampler as a compile-time alternative, which
can be chosen by setting `-DDXX_FEATURE_INTERNAL_RESAMPLER=0`.  Keep the
new internal resampler as the default.
This commit is contained in:
Kp 2023-01-07 22:17:31 +00:00
parent 9e3ebdb1f7
commit d9241c2d69
1 changed files with 66 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

72
similar/arch/sdl/digi_mixer.cpp
View file

@ -47,6 +47,10 @@
#define MIX_OUTPUT_FORMAT AUDIO_S16
#define MIX_OUTPUT_CHANNELS 2
#ifndef DXX_FEATURE_INTERNAL_RESAMPLER
#define DXX_FEATURE_INTERNAL_RESAMPLER 1
#endif
#if !((defined(__APPLE__) && defined(__MACH__)) || defined(macintosh))
#define SOUND_BUFFER_SIZE 2048
#else
@ -157,6 +161,7 @@ void digi_mixer_close() {
namespace {
#if DXX_FEATURE_INTERNAL_RESAMPLER
/*
* Blackman windowed-sinc filter coefficients at 1/4 bandwidth of upsampled
* frequency. Chosen for linear phase and approximates ~10th order IIR
@ -278,6 +283,7 @@ static std::unique_ptr<Uint8[]> convert_audio(const std::span<const uint8_t> inp
filter_fir(upsample(input, upsampledLen, upFactor), coeffs),
outsize, chFactor);
}
#endif
}
@ -293,7 +299,8 @@ namespace {
*/
static void mixdigi_convert_sound(const unsigned i)
{
if (SoundChunks[i].abuf)
auto &sci = SoundChunks[i];
if (sci.abuf)
//proceed only if not converted yet
return;
const auto data = GameSounds[i].span();
@ -303,6 +310,9 @@ static void mixdigi_convert_sound(const unsigned i)
out_freq = digi_sample_rate;
out_channels = MIX_OUTPUT_CHANNELS;
const auto freq = GameSounds[i].freq;
#if !DXX_FEATURE_INTERNAL_RESAMPLER
const Uint16 out_format = MIX_OUTPUT_FORMAT;
#endif
#elif defined(DXX_BUILD_DESCENT_II)
Uint16 out_format;
Mix_QuerySpec(&out_freq, &out_format, &out_channels); // get current output settings
@ -311,17 +321,67 @@ static void mixdigi_convert_sound(const unsigned i)
if (!data.empty())
{
#if DXX_FEATURE_INTERNAL_RESAMPLER
// Create output memory
int upFactor = out_freq / freq; // Should be integer, 2 or 4
int formatFactor = 2; // U8 -> S16 is two bytes
int convertedSize = data.size() * upFactor * out_channels * formatFactor;
auto cvtbuf = convert_audio(data, convertedSize, upFactor, out_channels);
SoundChunks[i].abuf = cvtbuf.release();
SoundChunks[i].alen = convertedSize;
SoundChunks[i].allocated = 1;
SoundChunks[i].volume = 128; // Max volume = 128
#else
SDL_AudioCVT cvt;
if (SDL_BuildAudioCVT(&cvt, AUDIO_U8, 1, freq, out_format, out_channels, out_freq) == -1)
{
con_printf(CON_URGENT, "%s:%u: SDL_BuildAudioCVT failed: sound=%i dlen=%" DXX_PRI_size_type " freq=%i out_format=%i out_channels=%i out_freq=%i", __FILE__, __LINE__, i, data.size(), freq, out_format, out_channels, out_freq);
return;
}
if (cvt.len_mult < 1)
{
con_printf(CON_URGENT, "%s:%u: SDL_BuildAudioCVT requested invalid length multiplier: sound=%i dlen=%" DXX_PRI_size_type " freq=%i out_format=%i out_channels=%i out_freq=%i len_mult=%i", __FILE__, __LINE__, i, data.size(), freq, out_format, out_channels, out_freq, cvt.len_mult);
return;
}
const std::size_t workingSize = data.size() * cvt.len_mult;
auto cvtbuf = std::make_unique<uint8_t[]>(workingSize);
cvt.buf = cvtbuf.get();
cvt.len = data.size();
memcpy(cvt.buf, data.data(), data.size());
if (SDL_ConvertAudio(&cvt))
{
con_printf(CON_URGENT, "%s:%u: SDL_ConvertAudio failed: sound=%i dlen=%" DXX_PRI_size_type " freq=%i out_format=%i out_channels=%i out_freq=%i", __FILE__, __LINE__, i, data.size(), freq, out_format, out_channels, out_freq);
return;
}
const std::size_t convertedSize = cvt.len_cvt;
if (convertedSize < workingSize)
{
/* The final sound required less space to store than
* SDL_BuildAudioCVT requested for an intermediate buffer.
* Allocate a new buffer just large enough for the final sound,
* copy the staging buffer into it, and use that new buffer as the
* long term storage.
*/
auto outbuf = std::make_unique<uint8_t[]>(convertedSize);
memcpy(outbuf.get(), cvt.buf, convertedSize);
cvtbuf = std::move(outbuf);
}
else
{
/* The final sound required as much (or more) space than was
* requested. If it required more, there was likely memory
* corruption, and that would be a bug in SDL audio conversion.
* Therefore, assume that this path is for when the requested space
* was exactly correct. No memory can be recovered with an extra
* copy, so transfer the staging buffer to the output structure.
*
* It is not a bug that there are no statements in this `else`
* block. The block exists to communicate the scope of this
* comment.
*/
}
#endif
sci.abuf = cvtbuf.release();
sci.alen = convertedSize;
sci.allocated = 1;
sci.volume = 128; // Max volume = 128
}
}