mirror of
https://gitlab.freedesktop.org/gstreamer/gstreamer.git
synced 2024-12-27 10:40:34 +00:00
directsoundsrc: Use latency-time and buffer-time settings
Earlier, the plugin was ignoring those settings and blindly setting buffer-time to 2 seconds and latency-time to 200ms, which forced all pipelines to have a minimum latency of 200ms + sink latency. The values of segsize and segtotal were also not derived correctly. Now we obey these values, and you can get close to the previous behaviour by setting buffer-time and latency-time manually. Note that they are set in microseconds. As a consequence, when we haven't received enough data from the device, we now sleep for a time proportional to the data remaining. However, Directsound is a deprecated API so it maintains its own software ringbuffer which updates at arbitrary intervals. Hence we might have to wait a full segsize to get the last 10% of data. To avoid tight loops, we clamp our sleep floor at 10ms. In my testing, this keeps the wakeups not-too-high (proportional to the latency-time set on the source). Further improvements should be made by fixing the WASAPI audio source plugin instead of this. Directsound is deprecated and as the comments explain, it is impossible to get low latency, decent quality, or good performance from it. Based on a patch by Sebastian Dröge <sebastian@centricular.com> https://bugzilla.gnome.org/show_bug.cgi?id=781249
This commit is contained in:
Nirbheek Chauhan
Sebastian Dröge
parent
6f38a0c6e1
commit
1a8610f8f8
2 changed files with 94 additions and 49 deletions
|
|
@ -74,6 +74,7 @@
|
|||
#include <windows.h>
|
||||
#include <dsound.h>
|
||||
#include <mmsystem.h>
|
||||
#include <stdio.h>
|
||||
|
||||
GST_DEBUG_CATEGORY_STATIC (directsoundsrc_debug);
|
||||
#define GST_CAT_DEFAULT directsoundsrc_debug
|
||||
|
|
@ -530,12 +531,45 @@ gst_directsound_src_prepare (GstAudioSrc * asrc, GstAudioRingBufferSpec * spec)
|
|||
if (wfx.wBitsPerSample != 16 && wfx.wBitsPerSample != 8)
|
||||
goto dodgy_width;
|
||||
|
||||
/* Set the buffer size to two seconds.
|
||||
This should never reached.
|
||||
*/
|
||||
dsoundsrc->buffer_size = wfx.nAvgBytesPerSec * 2;
|
||||
GST_INFO_OBJECT (asrc, "latency time: %" G_GUINT64_FORMAT " - buffer time: %"
|
||||
G_GUINT64_FORMAT, spec->latency_time, spec->buffer_time);
|
||||
|
||||
GST_DEBUG_OBJECT (asrc, "Buffer size: %d", dsoundsrc->buffer_size);
|
||||
/* Buffer-time should always be >= 2*latency */
|
||||
if (spec->buffer_time < spec->latency_time * 2) {
|
||||
spec->buffer_time = spec->latency_time * 2;
|
||||
GST_WARNING ("buffer-time was less than 2*latency-time, clamping");
|
||||
}
|
||||
|
||||
/* Set the buffer size from our configured buffer time (in microsecs) */
|
||||
dsoundsrc->buffer_size =
|
||||
gst_util_uint64_scale_int (spec->buffer_time, wfx.nAvgBytesPerSec,
|
||||
GST_SECOND / GST_USECOND);
|
||||
|
||||
GST_INFO_OBJECT (asrc, "Buffer size: %d", dsoundsrc->buffer_size);
|
||||
|
||||
spec->segsize =
|
||||
gst_util_uint64_scale (spec->latency_time, wfx.nAvgBytesPerSec,
|
||||
GST_SECOND / GST_USECOND);
|
||||
|
||||
/* Sanitized segsize */
|
||||
if (spec->segsize < GST_AUDIO_INFO_BPF (&spec->info))
|
||||
spec->segsize = GST_AUDIO_INFO_BPF (&spec->info);
|
||||
else if (spec->segsize % GST_AUDIO_INFO_BPF (&spec->info) != 0)
|
||||
spec->segsize =
|
||||
((spec->segsize + GST_AUDIO_INFO_BPF (&spec->info) -
|
||||
1) / GST_AUDIO_INFO_BPF (&spec->info)) *
|
||||
GST_AUDIO_INFO_BPF (&spec->info);
|
||||
spec->segtotal = dsoundsrc->buffer_size / spec->segsize;
|
||||
/* The device usually takes time = 1-2 segments to start producing buffers */
|
||||
spec->seglatency = spec->segtotal + 2;
|
||||
|
||||
/* Fetch and set the actual latency time that will be used */
|
||||
dsoundsrc->latency_time =
|
||||
gst_util_uint64_scale (spec->segsize, GST_SECOND / GST_USECOND,
|
||||
GST_AUDIO_INFO_BPF (&spec->info) * GST_AUDIO_INFO_RATE (&spec->info));
|
||||
|
||||
GST_INFO_OBJECT (asrc, "actual latency time: %" G_GUINT64_FORMAT,
|
||||
spec->latency_time);
|
||||
|
||||
/* Init secondary buffer desciption */
|
||||
memset (&descSecondary, 0, sizeof (DSCBUFFERDESC));
|
||||
|
|
@ -555,33 +589,7 @@ gst_directsound_src_prepare (GstAudioSrc * asrc, GstAudioRingBufferSpec * spec)
|
|||
|
||||
dsoundsrc->bytes_per_sample = GST_AUDIO_INFO_BPF (&spec->info);
|
||||
|
||||
GST_DEBUG ("latency time: %" G_GUINT64_FORMAT " - buffer time: %"
|
||||
G_GUINT64_FORMAT, spec->latency_time, spec->buffer_time);
|
||||
|
||||
/* Buffer-time should be always more than 2*latency */
|
||||
if (spec->buffer_time < spec->latency_time * 2) {
|
||||
spec->buffer_time = spec->latency_time * 2;
|
||||
GST_WARNING ("buffer-time was less than latency");
|
||||
}
|
||||
|
||||
/* Save the times */
|
||||
dsoundsrc->buffer_time = spec->buffer_time;
|
||||
dsoundsrc->latency_time = spec->latency_time;
|
||||
|
||||
dsoundsrc->latency_size = (gint) wfx.nAvgBytesPerSec *
|
||||
dsoundsrc->latency_time / 1000000.0;
|
||||
|
||||
spec->segsize = (guint) (((double) spec->buffer_time / 1000000.0) *
|
||||
wfx.nAvgBytesPerSec);
|
||||
|
||||
/* just in case */
|
||||
if (spec->segsize < 1)
|
||||
spec->segsize = 1;
|
||||
|
||||
spec->segtotal = GST_AUDIO_INFO_BPF (&spec->info) * 8 *
|
||||
(wfx.nAvgBytesPerSec / spec->segsize);
|
||||
|
||||
GST_DEBUG_OBJECT (asrc,
|
||||
GST_INFO_OBJECT (asrc,
|
||||
"bytes/sec: %lu, buffer size: %d, segsize: %d, segtotal: %d",
|
||||
wfx.nAvgBytesPerSec, dsoundsrc->buffer_size, spec->segsize,
|
||||
spec->segtotal);
|
||||
|
|
@ -589,7 +597,7 @@ gst_directsound_src_prepare (GstAudioSrc * asrc, GstAudioRingBufferSpec * spec)
|
|||
/* Not read anything yet */
|
||||
dsoundsrc->current_circular_offset = 0;
|
||||
|
||||
GST_DEBUG_OBJECT (asrc, "channels: %d, rate: %d, bytes_per_sample: %d"
|
||||
GST_INFO_OBJECT (asrc, "channels: %d, rate: %d, bytes_per_sample: %d"
|
||||
" WAVEFORMATEX.nSamplesPerSec: %ld, WAVEFORMATEX.wBitsPerSample: %d,"
|
||||
" WAVEFORMATEX.nBlockAlign: %d, WAVEFORMATEX.nAvgBytesPerSec: %ld",
|
||||
GST_AUDIO_INFO_CHANNELS (&spec->info), GST_AUDIO_INFO_RATE (&spec->info),
|
||||
|
|
@ -639,6 +647,8 @@ gst_directsound_src_read (GstAudioSrc * asrc, gpointer data, guint length,
|
|||
GstClockTime * timestamp)
|
||||
{
|
||||
GstDirectSoundSrc *dsoundsrc;
|
||||
guint64 sleep_time_ms;
|
||||
guint64 slept_time_ms = 0;
|
||||
|
||||
HRESULT hRes; /* Result for windows functions */
|
||||
DWORD dwCurrentCaptureCursor = 0;
|
||||
|
|
@ -670,13 +680,26 @@ gst_directsound_src_read (GstAudioSrc * asrc, gpointer data, guint length,
|
|||
if (!(dwStatus & DSCBSTATUS_CAPTURING)) {
|
||||
hRes = IDirectSoundCaptureBuffer_Start (dsoundsrc->pDSBSecondary,
|
||||
DSCBSTART_LOOPING);
|
||||
// Sleep (dsoundsrc->latency_time/1000);
|
||||
GST_DEBUG_OBJECT (asrc, "capture started");
|
||||
GST_INFO_OBJECT (asrc, "capture started");
|
||||
}
|
||||
// calculate_buffersize:
|
||||
while (length > dwBufferSize) {
|
||||
Sleep (dsoundsrc->latency_time / 1000);
|
||||
|
||||
/* Loop till the source has produced bytes equal to or greater than @length.
|
||||
*
|
||||
* DirectSound has a notification-based API that uses Windows CreateEvent()
|
||||
* + WaitForSingleObject(), but it is completely useless for live streams.
|
||||
*
|
||||
* 1. You must schedule all events before starting capture
|
||||
* 2. The events are all fired exactly once
|
||||
* 3. You cannot schedule new events while a capture is running
|
||||
* 4. You cannot stop/schedule/start either
|
||||
*
|
||||
* This means you cannot use the API while doing live looped capture and we
|
||||
* must resort to this.
|
||||
*
|
||||
* However, this is almost as efficient as event-based capture since it's ok
|
||||
* to consistently overwait by a fixed amount; the extra bytes will just end
|
||||
* up being used in the next call, and the extra latency will be constant. */
|
||||
while (TRUE) {
|
||||
hRes =
|
||||
IDirectSoundCaptureBuffer_GetCurrentPosition (dsoundsrc->pDSBSecondary,
|
||||
&dwCurrentCaptureCursor, NULL);
|
||||
|
|
@ -686,7 +709,8 @@ gst_directsound_src_read (GstAudioSrc * asrc, gpointer data, guint length,
|
|||
return -1;
|
||||
}
|
||||
|
||||
/* calculate the buffer */
|
||||
/* calculate the size of the buffer that's been captured while accounting
|
||||
* for wrap-arounds */
|
||||
if (dwCurrentCaptureCursor < dsoundsrc->current_circular_offset) {
|
||||
dwBufferSize = dsoundsrc->buffer_size -
|
||||
(dsoundsrc->current_circular_offset - dwCurrentCaptureCursor);
|
||||
|
|
@ -694,14 +718,38 @@ gst_directsound_src_read (GstAudioSrc * asrc, gpointer data, guint length,
|
|||
dwBufferSize =
|
||||
dwCurrentCaptureCursor - dsoundsrc->current_circular_offset;
|
||||
}
|
||||
} // while (...
|
||||
|
||||
/* Lock the buffer */
|
||||
if (dwBufferSize >= length) {
|
||||
/* Yay, we got all the data we need */
|
||||
break;
|
||||
} else {
|
||||
GST_DEBUG_OBJECT (asrc, "not enough data, got %lu (want at least %u)",
|
||||
dwBufferSize, length);
|
||||
/* If we didn't get enough data, sleep for a proportionate time */
|
||||
sleep_time_ms = gst_util_uint64_scale (dsoundsrc->latency_time,
|
||||
length - dwBufferSize, length * 1000);
|
||||
/* Make sure we don't run in a tight loop unnecessarily */
|
||||
sleep_time_ms = MAX (sleep_time_ms, 10);
|
||||
GST_DEBUG_OBJECT (asrc, "sleeping for %" G_GUINT64_FORMAT "ms",
|
||||
sleep_time_ms);
|
||||
Sleep (sleep_time_ms);
|
||||
slept_time_ms += sleep_time_ms;
|
||||
}
|
||||
}
|
||||
|
||||
GST_DEBUG_OBJECT (asrc, "Got enough data: %lu bytes (wanted at least %u), "
|
||||
"slept for %" G_GUINT64_FORMAT "ms", dwBufferSize, length, slept_time_ms);
|
||||
|
||||
/* Lock the buffer and read only the first @length bytes. Keep the rest in
|
||||
* the capture buffer for the next read. */
|
||||
hRes = IDirectSoundCaptureBuffer_Lock (dsoundsrc->pDSBSecondary,
|
||||
dsoundsrc->current_circular_offset,
|
||||
length,
|
||||
&pLockedBuffer1, &dwSizeBuffer1, &pLockedBuffer2, &dwSizeBuffer2, 0L);
|
||||
|
||||
/* NOTE: We now assume that dwSizeBuffer1 + dwSizeBuffer2 == length since the
|
||||
* API is supposed to guarantee that */
|
||||
|
||||
/* Copy buffer data to another buffer */
|
||||
if (hRes == DS_OK) {
|
||||
memcpy (data, pLockedBuffer1, dwSizeBuffer1);
|
||||
|
|
@ -720,7 +768,7 @@ gst_directsound_src_read (GstAudioSrc * asrc, gpointer data, guint length,
|
|||
|
||||
GST_DSOUND_UNLOCK (dsoundsrc);
|
||||
|
||||
/* return length (readed data size in bytes) */
|
||||
/* We always read exactly @length data */
|
||||
return length;
|
||||
}
|
||||
|
||||
|
|
@ -733,7 +781,7 @@ gst_directsound_src_delay (GstAudioSrc * asrc)
|
|||
DWORD dwBytesInQueue = 0;
|
||||
gint nNbSamplesInQueue = 0;
|
||||
|
||||
GST_DEBUG_OBJECT (asrc, "Delay");
|
||||
GST_INFO_OBJECT (asrc, "Delay");
|
||||
|
||||
dsoundsrc = GST_DIRECTSOUND_SRC (asrc);
|
||||
|
||||
|
|
@ -755,6 +803,8 @@ gst_directsound_src_delay (GstAudioSrc * asrc)
|
|||
nNbSamplesInQueue = dwBytesInQueue / dsoundsrc->bytes_per_sample;
|
||||
}
|
||||
|
||||
GST_INFO_OBJECT (asrc, "Delay is %d samples", nNbSamplesInQueue);
|
||||
|
||||
return nNbSamplesInQueue;
|
||||
}
|
||||
|
||||
|
|
@ -818,7 +868,7 @@ gst_directsound_src_mixer_find (GstDirectSoundSrc * dsoundsrc,
|
|||
if (mmres != MMSYSERR_NOERROR)
|
||||
continue;
|
||||
|
||||
mmres = mixerGetDevCaps ((UINT_PTR)dsoundsrc->mixer,
|
||||
mmres = mixerGetDevCaps ((UINT_PTR) dsoundsrc->mixer,
|
||||
mixer_caps, sizeof (MIXERCAPS));
|
||||
|
||||
if (mmres != MMSYSERR_NOERROR) {
|
||||
|
|
|
|||
|
|
@ -82,14 +82,9 @@ struct _GstDirectSoundSrc
|
|||
LPDIRECTSOUNDCAPTUREBUFFER pDSBSecondary; /*Secondaty capturebuffer*/
|
||||
DWORD current_circular_offset;
|
||||
|
||||
HANDLE rghEvent;
|
||||
DWORD notifysize;
|
||||
|
||||
guint buffer_size;
|
||||
guint latency_size;
|
||||
guint bytes_per_sample;
|
||||
|
||||
guint buffer_time;
|
||||
guint latency_time;
|
||||
|
||||
HMIXER mixer;
|
||||
|
|
|
|||
Loading…
Reference in a new issue