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

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}
GetStream · Apr 20, 2016 · eb3603b · eb3603b
1 parent 0a2c054
commit eb3603b
Show file tree

Hide file tree

Showing 5 changed files with 20 additions and 13 deletions.
diff --git a/data/audio_processing/output_data_fixed.pb b/data/audio_processing/output_data_fixed.pb
diff --git a/data/audio_processing/output_data_float.pb b/data/audio_processing/output_data_float.pb
diff --git a/data/audio_processing/output_data_mac.pb b/data/audio_processing/output_data_mac.pb
diff --git a/webrtc/modules/audio_processing/audio_processing_impl.cc b/webrtc/modules/audio_processing/audio_processing_impl.cc
@@ -366,18 +366,20 @@ int AudioProcessingImpl::InitializeLocked(const ProcessingConfig& config) {
       std::min(formats_.api_format.input_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 = kSampleRate16kHz;
+  int rev_proc_rate = ClosestHigherNativeRate(std::min(
+      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) {
-    // ...the forward stream is at 8 kHz.
     rev_proc_rate = kSampleRate8kHz;
   } else {
-    if (formats_.api_format.reverse_input_stream().sample_rate_hz() ==
-        kSampleRate32kHz) {
-      // ...or the input is at 32 kHz, in which case we use the splitting
-      // filter rather than the resampler.
-      rev_proc_rate = kSampleRate32kHz;
-    }
+    rev_proc_rate = std::max(rev_proc_rate, static_cast<int>(kSampleRate16kHz));
   }
 
   // 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 {
   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 {
-  return formats_.rev_proc_format.sample_rate_hz() == kSampleRate32kHz &&
+  return is_multi_band(formats_.rev_proc_format.sample_rate_hz()) &&
          (is_rev_processed() ||
           public_submodules_->echo_cancellation
               ->is_enabled_render_side_query() ||

diff --git a/webrtc/modules/audio_processing/audio_processing_unittest.cc b/webrtc/modules/audio_processing/audio_processing_unittest.cc
@@ -54,7 +54,12 @@ bool write_ref_data = false;
 const google::protobuf::int32 kChannels[] = {1, 2};
 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};
+#endif
 
 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, 16000, 32000, 25, 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)));
 
 #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, 16000, 32000, 25, 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)));
 #endif