zdoom/fluidsynth
0
0
Fork 0
mirror of https://github.com/ZDoom/fluidsynth.git synced 2025-04-08 08:41:21 +00:00
Code Issues Projects Releases Packages Wiki Activity

Fix broken voice mixing (#582)

Browse source 
This fixes a regression introduced in 907ec27a9e

When rendering a voice, there are 3 cases to consider: silent, playing,
and finished. When optimizing away the memset, I incorrectly assumed that
a voice cannot switch between playing and silence (like crazy) while
rendering FLUID_MIXER_MAX_BUFFERS_DEFAULT. Apparently this does not
hold true, esp. when rendering at sample rates ~96kHz.
This commit is contained in:
Tom M 2019-10-27 16:45:07 +01:00 committed by GitHub
parent 68c7989d0d
commit 38b929e501
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 38 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

55
src/rvoice/fluid_rvoice_mixer.c
View file

@ -360,14 +360,14 @@ get_dest_buf(fluid_rvoice_buffers_t *buffers, int index,
}
/**
* Mix data down to buffers
* Mix samples down from internal dsp_buf to output buffers
*
* @param buffers Destination buffer(s)
* @param dsp_buf Mono sample source
* @param start_block Block to start mixing at
* @param start_block starting sample in dsp_buf
* @param sample_count number of samples to mix following \c start_block
* @param dest_bufs Array of buffers to mixdown to
* @param dest_bufcount Length of dest_bufs
* @param dest_bufcount Length of dest_bufs (i.e count of buffers)
*/
static void
fluid_rvoice_buffers_mix(fluid_rvoice_buffers_t *buffers,
@ -375,9 +375,11 @@ fluid_rvoice_buffers_mix(fluid_rvoice_buffers_t *buffers,
int start_block, int sample_count,
fluid_real_t **dest_bufs, int dest_bufcount)
{
/* buffers count to mixdown to */
int bufcount = buffers->count;
int i, dsp_i;
/* if there is nothing to mix, return immediatly */
if(sample_count <= 0 || dest_bufcount <= 0)
{
return;
@ -386,6 +388,7 @@ fluid_rvoice_buffers_mix(fluid_rvoice_buffers_t *buffers,
FLUID_ASSERT((uintptr_t)dsp_buf % FLUID_DEFAULT_ALIGNMENT == 0);
FLUID_ASSERT((uintptr_t)(&dsp_buf[start_block * FLUID_BUFSIZE]) % FLUID_DEFAULT_ALIGNMENT == 0);
/* mixdown for each buffer */
for(i = 0; i < bufcount; i++)
{
fluid_real_t *FLUID_RESTRICT buf = get_dest_buf(buffers, i, dest_bufs, dest_bufcount);
@ -398,11 +401,17 @@ fluid_rvoice_buffers_mix(fluid_rvoice_buffers_t *buffers,
FLUID_ASSERT((uintptr_t)buf % FLUID_DEFAULT_ALIGNMENT == 0);
/* mixdown sample_count samples in the current buffer buf
Note, that this loop could be unrolled by FLUID_BUFSIZE elements */
#pragma omp simd aligned(dsp_buf,buf:FLUID_DEFAULT_ALIGNMENT)
for(dsp_i = (start_block * FLUID_BUFSIZE); dsp_i < sample_count; dsp_i++)
for(dsp_i = 0; dsp_i < sample_count; dsp_i++)
{
buf[dsp_i] += amp * dsp_buf[dsp_i];
// Index by blocks (not by samples) to let the compiler know that we always start accessing
// buf and dsp_buf at the FLUID_BUFSIZE*sizeof(fluid_real_t) byte boundary and never somewhere
// in between.
// A good compiler should understand: Aha, so I don't need to add a peel loop when vectorizing
// this loop. Great.
buf[start_block * FLUID_BUFSIZE + dsp_i] += amp * dsp_buf[start_block * FLUID_BUFSIZE + dsp_i];
}
}
}
@ -417,30 +426,42 @@ fluid_mixer_buffers_render_one(fluid_mixer_buffers_t *buffers,
fluid_rvoice_t *rvoice, fluid_real_t **dest_bufs,
unsigned int dest_bufcount, fluid_real_t *src_buf, int blockcount)
{
int i, total_samples = 0, start_block = 0;
int i, total_samples = 0, last_block_mixed = 0;
for(i = 0; i < blockcount; i++)
{
/* render one block in src_buf */
int s = fluid_rvoice_write(rvoice, &src_buf[FLUID_BUFSIZE * i]);
if(s == -1)
{
start_block += s;
s = FLUID_BUFSIZE;
/* the voice is silent, mix back all the previously rendered sound */
fluid_rvoice_buffers_mix(&rvoice->buffers, src_buf, last_block_mixed,
total_samples - (last_block_mixed*FLUID_BUFSIZE),
dest_bufs, dest_bufcount);
last_block_mixed = i+1; /* future block start index to mix from */
total_samples += FLUID_BUFSIZE; /* accumulate samples count rendered */
}
total_samples += s;
if(s < FLUID_BUFSIZE)
else
{
break;
/* the voice wasn't quiet. Some samples have been rendered [0..FLUID_BUFSIZE] */
total_samples += s;
if(s < FLUID_BUFSIZE)
{
/* voice has finished */
break;
}
}
}
fluid_rvoice_buffers_mix(&rvoice->buffers, src_buf, -start_block, total_samples - ((-start_block)*FLUID_BUFSIZE), dest_bufs, dest_bufcount);
/* Now mix the remaining blocks from last_block_mixed to total_sample */
fluid_rvoice_buffers_mix(&rvoice->buffers, src_buf, last_block_mixed,
total_samples - (last_block_mixed*FLUID_BUFSIZE),
dest_bufs, dest_bufcount);
if(total_samples < blockcount * FLUID_BUFSIZE)
{
/* voice has finished */
fluid_finish_rvoice(buffers, rvoice);
}
}
@ -602,7 +623,7 @@ fluid_mixer_buffers_zero(fluid_mixer_buffers_t *buffers, int current_blockcount)
static int
fluid_mixer_buffers_init(fluid_mixer_buffers_t *buffers, fluid_rvoice_mixer_t *mixer)
{
const int samplecount = FLUID_BUFSIZE * FLUID_MIXER_MAX_BUFFERS_DEFAULT;
static const int samplecount = FLUID_BUFSIZE * FLUID_MIXER_MAX_BUFFERS_DEFAULT;
buffers->mixer = mixer;
buffers->buf_count = mixer->buffers.buf_count;