mollyim/webrtc
1
0
Fork 0
mirror of https://github.com/mollyim/webrtc.git synced 2025-05-13 05:40:42 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions

Don't always downsample to 16kHz in the reverse stream in APM

Browse source 
The first approach landed here: https://codereview.webrtc.org/1773173002
But it was partially reverted, because it affected the AEC performance, here: https://codereview.webrtc.org/1867483003/
The main difference of this approach is that it doesn't use the 3-band splitting filter in the reverse stream, which seems to be the culprit of the AEC regression.
Also, the 2-band splitting filter has been used for the 32kHz case for a long time without any problem, and this is expanded in the CL to cover the 48kHz case as well.

BUG=webrtc:5725
TBR=tina.legrand@webrtc.org

Review URL: https://codereview.webrtc.org/1865633005

Cr-Commit-Position: refs/heads/master@{#12451}
This commit is contained in:
aluebs 2016-04-20 15:27:58 -07:00 committed by Commit bot
parent 0a2c054f42
commit eb3603bd5e
5 changed files with 20 additions and 13 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

BIN
data/audio_processing/output_data_fixed.pb
View file

Binary file not shown.

BIN
data/audio_processing/output_data_float.pb
View file

Binary file not shown.

BIN
data/audio_processing/output_data_mac.pb
View file

Binary file not shown.

24
webrtc/modules/audio_processing/audio_processing_impl.cc
View file

@ -366,18 +366,20 @@ int AudioProcessingImpl::InitializeLocked(const ProcessingConfig& config) {
std::min(formats_.api_format.input_stream().sample_rate_hz(), std::min(formats_.api_format.input_stream().sample_rate_hz(),
formats_.api_format.output_stream().sample_rate_hz()))); formats_.api_format.output_stream().sample_rate_hz())));
// We normally process the reverse stream at 16 kHz. Unless... int rev_proc_rate = ClosestHigherNativeRate(std::min(
int rev_proc_rate = kSampleRate16kHz; formats_.api_format.reverse_input_stream().sample_rate_hz(),
formats_.api_format.reverse_output_stream().sample_rate_hz()));
// TODO(aluebs): Remove this restriction once we figure out why the 3-band
// splitting filter degrades the AEC performance.
if (rev_proc_rate > kSampleRate32kHz) {
rev_proc_rate = is_rev_processed() ? kSampleRate32kHz : kSampleRate16kHz;
}
// If the forward sample rate is 8 kHz, the reverse stream is also processed
// at this rate.
if (capture_nonlocked_.fwd_proc_format.sample_rate_hz() == kSampleRate8kHz) { if (capture_nonlocked_.fwd_proc_format.sample_rate_hz() == kSampleRate8kHz) {
// ...the forward stream is at 8 kHz.
rev_proc_rate = kSampleRate8kHz; rev_proc_rate = kSampleRate8kHz;
} else { } else {
if (formats_.api_format.reverse_input_stream().sample_rate_hz() == rev_proc_rate = std::max(rev_proc_rate, static_cast<int>(kSampleRate16kHz));
kSampleRate32kHz) {
// ...or the input is at 32 kHz, in which case we use the splitting
// filter rather than the resampler.
rev_proc_rate = kSampleRate32kHz;
}
} }
// Always downmix the reverse stream to mono for analysis. This has been // Always downmix the reverse stream to mono for analysis. This has been
@ -1151,11 +1153,11 @@ bool AudioProcessingImpl::is_rev_processed() const {
bool AudioProcessingImpl::rev_synthesis_needed() const { bool AudioProcessingImpl::rev_synthesis_needed() const {
return (is_rev_processed() && return (is_rev_processed() &&
formats_.rev_proc_format.sample_rate_hz() == kSampleRate32kHz); is_multi_band(formats_.rev_proc_format.sample_rate_hz()));
} }
bool AudioProcessingImpl::rev_analysis_needed() const { bool AudioProcessingImpl::rev_analysis_needed() const {
return formats_.rev_proc_format.sample_rate_hz() == kSampleRate32kHz && return is_multi_band(formats_.rev_proc_format.sample_rate_hz()) &&
(is_rev_processed() || (is_rev_processed() ||
public_submodules_->echo_cancellation public_submodules_->echo_cancellation
->is_enabled_render_side_query() || ->is_enabled_render_side_query() ||

9
webrtc/modules/audio_processing/audio_processing_unittest.cc
View file

@ -54,7 +54,12 @@ bool write_ref_data = false;
const google::protobuf::int32 kChannels[] = {1, 2}; const google::protobuf::int32 kChannels[] = {1, 2};
const int kSampleRates[] = {8000, 16000, 32000, 48000}; const int kSampleRates[] = {8000, 16000, 32000, 48000};
#if defined(WEBRTC_AUDIOPROC_FIXED_PROFILE)
// Android doesn't support 48kHz.
const int kProcessSampleRates[] = {8000, 16000, 32000};
#elif defined(WEBRTC_AUDIOPROC_FLOAT_PROFILE)
const int kProcessSampleRates[] = {8000, 16000, 32000, 48000}; const int kProcessSampleRates[] = {8000, 16000, 32000, 48000};
#endif
enum StreamDirection { kForward = 0, kReverse }; enum StreamDirection { kForward = 0, kReverse };
@ -2692,7 +2697,7 @@ INSTANTIATE_TEST_CASE_P(
std::tr1::make_tuple(16000, 32000, 32000, 32000, 25, 0), std::tr1::make_tuple(16000, 32000, 32000, 32000, 25, 0),
std::tr1::make_tuple(16000, 32000, 16000, 32000, 25, 20), std::tr1::make_tuple(16000, 32000, 16000, 32000, 25, 20),
std::tr1::make_tuple(16000, 16000, 48000, 16000, 40, 20), std::tr1::make_tuple(16000, 16000, 48000, 16000, 40, 20),
std::tr1::make_tuple(16000, 16000, 32000, 16000, 50, 20), std::tr1::make_tuple(16000, 16000, 32000, 16000, 40, 20),
std::tr1::make_tuple(16000, 16000, 16000, 16000, 0, 0))); std::tr1::make_tuple(16000, 16000, 16000, 16000, 0, 0)));
#elif defined(WEBRTC_AUDIOPROC_FIXED_PROFILE) #elif defined(WEBRTC_AUDIOPROC_FIXED_PROFILE)
@ -2748,7 +2753,7 @@ INSTANTIATE_TEST_CASE_P(
std::tr1::make_tuple(16000, 32000, 32000, 32000, 25, 0), std::tr1::make_tuple(16000, 32000, 32000, 32000, 25, 0),
std::tr1::make_tuple(16000, 32000, 16000, 32000, 25, 20), std::tr1::make_tuple(16000, 32000, 16000, 32000, 25, 20),
std::tr1::make_tuple(16000, 16000, 48000, 16000, 35, 20), std::tr1::make_tuple(16000, 16000, 48000, 16000, 35, 20),
std::tr1::make_tuple(16000, 16000, 32000, 16000, 40, 20), std::tr1::make_tuple(16000, 16000, 32000, 16000, 35, 20),
std::tr1::make_tuple(16000, 16000, 16000, 16000, 0, 0))); std::tr1::make_tuple(16000, 16000, 16000, 16000, 0, 0)));
#endif #endif