This CL ensures that the AEC2 is only created when needed.
The changes in the CL are bitexact when running AEC2 via
audioproc_f
Bug: webrtc:8671
Change-Id: I5f6d33e45a7031c69ac53098781635c415668e49
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129740
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27772}
The one with enum.IntFlag is not feasible. An attempt is done here:
https://webrtc-review.googlesource.com/c/src/+/133884
It requires re-writing QualityAssessment to Python3 which is too much
work for little benefit. (I tried, but couldn't get the unit-tests to
pass for both 2 and 3.)
The second one is not a real todo.
TBR=alessiob@webrtc.org
NOPRESUBMIT=True
Bug=None
NOTRY=True
Change-Id: Ia25817533cd504c30490f86e4058f0b2d59dd39c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/133908
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27715}
The acoustic echo canceler AEC2 is being deprecated. The routing for reporting these metrics as UMA stats has outlived the metrics'usefulness.
Bug: webrtc:10563
Change-Id: Ib96693dfc43e25a0cfecad7d5d2043116ca7e6b5
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/133573
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27699}
As non-linear mode uses a suppressed version of y (not e) as output, this change
uses Y2, rather than E2, as nearend spectrum when computing the suppression gains.
E2 is still used in linear mode.
This change also affects how the minimum suppression gains are calculated. The
minimum gain is now min_echo_power / weighted_residual_echo.
Bug: webrtc:10550
Change-Id: I2904c5a09dd64b06bf25eb5a37c18dab50297794
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/133023
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27629}
This CL changes the APM unittests to use AEC3 instead of
AEC2.
Bug: webrtc:8671
Change-Id: I80f88dbafb7c31696abd8b7efb5a187a9fb30d1c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129420
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27607}
This CL allows audioproc_f to overrule any runtime settings for the
pre-amplifier gain that are present in the aecdump file.
Bug: webrtc:10546
Change-Id: I74dbf8d043f59b516bf0abc80f266e965af0754d
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132558
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27598}
* This is too brittle and might clash with MSVC's M_PI. See [1].
* We only used it once (in a unit test).
* We shouldn't use PI anyway [2].
Instead, pull it from <cmath> with _USE_MATH_DEFINES,
like it's already done in the code base.
[1] https://ci.chromium.org/p/webrtc/builders/try/win_x86_msvc_rel/6844
[2] https://tauday.com/tau-manifesto
Bug: webrtc:9855
Change-Id: I7a6976240604ef367ea07478d8cb5e4020e5dfeb
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132548
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Yves Gerey <yvesg@google.com>
Cr-Commit-Position: refs/heads/master@{#27597}
This CL corrects the minimum bound for the estimated
comfort noise level.
Bug: webrtc:10533
Change-Id: I473275ffbc7bb52572315849f30e13b764109d92
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132003
Reviewed-by: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27567}
The configuration parameter filter.use_linear_filter can be used to
disable the linear filtering. Disabling the linear filter is equivalent
to runing in non-linear mode.
Bug: b/130016532
Change-Id: I8ffdf474822888b9915444bba6cc1c25ec1efe5a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132552
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27566}
This CL replaces KissFFT with PFFFT for the spectral features
computation.
Remarks:
- Extra FFT output vector copy eliminated
- Scaling and windowing merged into a single vector for efficiency
- Nyquist frequency hack to keep the iteration in
BandFeaturesExtractor::ComputeSpectralCrossCorrelation simple
Bug: webrtc:9577, webrtc:10480
Change-Id: I436563bd257f66a243f5402be270ffcf859bd184
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130221
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27536}
This CL refactors the computation of band energy and spectral
cross-correlation coefficients by moving and optimizing
the code from ComputeBandCoefficients, ComputeBandEnergies and
ComputeSpectralCrossCorrelation into a single class (named
BandFeaturesExtractor).
This change will also help replacing FFT library in the RNN VAD.
Bug: webrtc:10480
Change-Id: I6cefa23e8f3bc8de6eb09d3ea434699d5e19124e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129726
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27535}
Semi-automatically created with:
git grep -l " testing::" | xargs sed -i "s/ testing::/ ::testing::/g"
git grep -l "(testing::" | xargs sed -i "s/(testing::/(::testing::/g"
git cl format
After this, two .cc files failed to compile and I have fixed them
manually.
Bug: webrtc:10523
Change-Id: I4741d3bcedc831b6c5fdc04485678617eb4ce031
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132018
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27526}
This change makes it possible to disable AEC3's render delay
controller and delay estimator, and instead rely on an external
delay estimator. The delay is communicated via SetAudioBufferDelay.
When the feature is enabled, no echo removal will be performed
until the first delay is provided.
The delay is
Bug: b/130016532
Change-Id: I16643109d78d770ff1d2713cf247b0b9cce1bc1c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/131327
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27467}
This CL extends, and partly corrects, the benchmarking
code in audioproc_f to provide statistics for the API
call durations in audioproc_f
Bug: chromium:939791
Change-Id: I4c26c4bb3782335f13dd3e21e6f861842539ea62
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129260
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27443}
This change removes the (unused) ability of EchoRemover overriding
the delay of the RenderDelayController. The change is tested for
bit-exactness.
Bug: webrtc:8671
Change-Id: I188ef740f1437de64ffe236d07a7dcd4128192c2
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130518
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27414}
And delete corresponding dependencies on :webrtc_common. After this
change, common_types.h is included directly only from code in the
following directories:
api/
api/video/
api/video_codecs/
common_video/libyuv/include/
media/base/
modules/remote_bitrate_estimator/
modules/rtp_rtcp/source/
modules/video_coding/codecs/vp9/
There remains plenty of indirect dependencies on the types declared in
common_types.h, but the fewer direct dependencies should make it
easier to find the proper place for each type.
Bug: webrtc:5876
Change-Id: I93e8f214025ecb613c19fdec2015bd3f96c59aae
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130501
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27376}
This change prevents FilterAnalyzer from accessing memory out-of-bounds
when the filter is resized.
Bug: chromium:946439
Change-Id: I7e2392c8b1ff0ff55566c663bf6d7a40d7754501
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129928
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27318}
The pointer-to-submodule interfaces are being removed.
This CL:
1) introduces AudioProcessing::Config::GainController1 with most config,
2) adds functions to APM for setting and getting analog gain,
3) creates a temporary GainControlConfigProxy to support the transition
to the new config.
4) Moves the lock references in GainControlForExperimentalAgc and
GainControlImpl into the GainControlConfigProxy, as it becomes the
sole AGC object with functionality exposed to the client.
Bug: webrtc:9947, webrtc:9878
Change-Id: Ic31e15e9bb26d6497a92b77874e0b6cab21ff2b2
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/126485
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Fredrik Solenberg <solenberg@webrtc.org>
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27316}
In this CL bounds are added to the index used for accessing the filter frequency response. That vector has always a capacity in memory equal to the final number of blocks of the main filter. However, at the initial part of the call or after an echo path change, a transition phase is started and a filter with a lower number of blocks is used and, therefore, its size is lower than that capacity during that transition phase.
Bug: webrtc:10463
Change-Id: I6ebfdea43047a3fa993a27f2c52bb3024df84ffe
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128777
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27264}
This CL replaces death tests placed inside a loop with a parametric test.
A better option is to mock Pffft::IsValidFftSize and test CreatePffftWrapper
when the former returns false. However, that would require to define an
interface for the PFFFT wrapper.
Bug: webrtc:10426
Change-Id: I3c49f1b271c8bf0099a4846014bef021676ef3e0
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128862
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27260}
This CL removes parameters for AEC3 which are no longer used. To reflect
that change, one of the parameters also is renamed
Bug: chromium:941949,webrtc:8671
Change-Id: I26609b396fa14ecb7523eebfe531a1338718103b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/127780
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27128}
This change reduces the risk of echo due to noise in the headroom
of the linear filter.
Changes:
- Use shorter delay headroom
- Delay headroom is specified in samples (not blocks)
- No hysteresis limit when delay is reduced
Bug: chromium:119942,webrtc:10341
Change-Id: I708e80e26d541dff8ca04b6da2d346a1d59cbfcb
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/126420
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27126}
This adds a flag to audioproc_f to generate a custom call order
file from an AEC dump. This file can be used to get more realism
when simulating with wav-files.
Bug: webrtc:10393
Change-Id: I245533d18affaab2f6cef53138332d7d83c71822
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/126782
Commit-Queue: Ivo Creusen <ivoc@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27104}
Since there is no way to enable/disable these diagnostics at runtime,
this CL moves the suppression into the rtc_* templates in order to
remove the need to explicitly add the snippet of code needed to
suppress it (currently copy/pasted in 144 locations).
The diagnostic that causes the most problems is the one about "complex
class/struct explicit ctor/dtor" [1] because WebRTC doesn't find
it useful enough.
Other diagnostics are good (for example the one that warns about
using "virtual" instead of "override", but that will be covered by
this clang-tidy check [2]) while others are Chromium related so
they have never triggered.
[1] - https://cs.chromium.org/chromium/src/tools/clang/plugins/FindBadConstructsConsumer.cpp?l=147-167&rcl=b4bebe1aa15dba7ca5fcc6456a81a55665327c3a
[2] - https://clang.llvm.org/extra/clang-tidy/checks/modernize-use-override.html
Bug: webrtc:163
Change-Id: Icbf27efa5b369100a31e6a32df1a0913729b3b34
Reviewed-on: https://webrtc-review.googlesource.com/c/125088
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26918}
This CL adds a temporary flag for specifying that the legacy AEC2 should
be used.
Bug: webrtc:10366
Change-Id: Ie3edaa1560cdc1282b62242beb67aa6fee7f2841
Reviewed-on: https://webrtc-review.googlesource.com/c/124980
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26891}
Import proto_library.gni when rtc_enable_protobuf is true instead of when
build_with_mozilla is false.
Makes it maybe easier to reason about the intention (e.g. intention is to not
compile any protobuf in, hence flag rtc_enable_protobuf)
The build file could not work if build_with_mozilla = true but
rtc_enable_protobuf = true.
Bug: webrtc:10338
Change-Id: I26e5983bd1519aa46c308b11796d518de5ef7597
Reviewed-on: https://webrtc-review.googlesource.com/c/123763
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26868}
Pretty-Fast Fast Fourier Transform is a 3rd party FFT C library meant to
replace other FFT libraries in WebRTC (see https://crbug.com/webrtc/9577).
This CL adds a WebRTC wrapper meant to be used inside the Audio Processing
Module (APM). As a first step, it only supports aligned memory allocated
via PFFFT. Support for the C++ standard library containers will be done
afterwards since it requires careful investigation and benchmarking (because
PFFFT uses SIMD optimizations).
The wrapper pre-allocates a scratch buffer to avoid VLA.
Bug: webrtc:9577
Change-Id: Ied00c3d3b1df292024f608ccf0ed1917d6e92e56
Reviewed-on: https://webrtc-review.googlesource.com/c/122563
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Max Morin <maxmorin@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26808}
In this CL we avoid the propagation of the echo control factory to the AudioProcessing instance when this is not set. That propagation was unnecessarily overriding the echo control factory that might have been already set on that AudioProcessing instance.
Change-Id: Ife8f479bc7a81c35ecf656e7d0ddfcc98981c74f
Bug: webrtc:10344
Reviewed-on: https://webrtc-review.googlesource.com/c/123765
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26802}
