This CL adds robustness to avoid the AEC3 transparent mode to be
incorrectly activated when
-there is strong near-end noise
-there is only low-level nearend activity.
Bug: webrtc:9256,chromium:841193
Change-Id: I26c2759d163914eb85dc3d863da8acbf28cbb88d
Reviewed-on: https://webrtc-review.googlesource.com/75511
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23191}
This CL corrects the usage of the estimated echo path gain to not be
hardcoded to 1. In order to retain the tuned behavior, the CL for now
maintains the former behavior in the code.
Bug: webrtc:9255,chromium:851187
Change-Id: I7f91c72e476680a8a854c22b74b1771fae446110
Reviewed-on: https://webrtc-review.googlesource.com/75510
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23190}
This CL ensures that the external audio buffer delay is correctly used
by removing the applied headroom and avoiding that the delay estimation
feedback fromt the echo remover overrules the external delay
information.
Bug: webrtc:9241,chromium:839860
Change-Id: I53cc78ace34a71994ab24a3b552f29979e2aae78
Reviewed-on: https://webrtc-review.googlesource.com/75513
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23189}
During pitch search in the RNN VAD, we calculate auto
correlation. Before this CL, we computed kNumInvertedLags12kHz=147 dot
products of vectors with kBufSize12kHz-kMaxPitch12kHz=240
elements. This was the most time consuming step of the new VAD.
This CL makes the computation happen in frequency domain. Profiling
shows a 3x speed increase. In future, we can try using a more efficient
FFT and to reduce the FFT length to some of e.g. 400, 405, 432.
# For minimal Clang plugin check change.
TBR: kwiberg@webrtc.org
Bug: webrtc:9076
Change-Id: I688251a415869d53175a37f390f441d4e035d954
Reviewed-on: https://webrtc-review.googlesource.com/73366
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23171}
This CL adds helper functions to be used for the spectral features
computation. Namely, it includes the following:
- band boundaries (frequency to FFT coeffcient index)
- band energy coefficients
- log band energy coefficients
- fixed size DCT table and computation
Bug: webrtc:9076
Change-Id: I03a8799b226d986bc1e37cefd0c3039f94b5592a
Reviewed-on: https://webrtc-review.googlesource.com/73687
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Minyue Li <minyue@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23170}
BandAnalysisFft class that wraps the FFT library, makes it easy to change
FFT library, applies windowing function and owns the FFT input buffer.
Bug: webrtc:9076
Change-Id: I9e7ed587ae263b906e04a66bf8c06eaae64daf19
Reviewed-on: https://webrtc-review.googlesource.com/72900
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23150}
This CL softens the effect of the AEC3 transparent mode to also handle
headsets that leak low-level echoes in a nonlinear way.
This is handled by reintroducing the limit in the echo path gain for the
nonlinear mode. Due to recent improvements in echo suppressor behavior
this is now possible to do with a limited impact on the near-end speech.
Bug: webrtc:9246,chromium:840347
Change-Id: I0ca5157160d1884ba93b962323b56016756986d3
Reviewed-on: https://webrtc-review.googlesource.com/74703
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23145}
Adding a build target for the bi-qaud filter to make it available for
the RNN VAD of AGC2. Also adding a unit test to test the computation
both in-place and not in-place while comparing the produced output to
that of scipy.signal.
Bug: webrtc:9076
Change-Id: I16176a477ee4b81bb1e090c4906c3a9948ad2772
Reviewed-on: https://webrtc-review.googlesource.com/74220
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23141}
Fixes a confusion of time units (milliseconds vs blocks) of externally
reported audio delay. This fix reduces the risk of echo in the beginning
of a call.
Bug: webrtc:9241,chromium:839860
Change-Id: I534cc15d6b215a5881ae46759f573a56871170a3
Reviewed-on: https://webrtc-review.googlesource.com/74589
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23128}
This CL contains changes to the echo suppressor that improves the
transparency of AEC3.
- The comfort noise level is used as masker and the masking threshold is
increased.
- Suppression gains are allowed to increase more rapidly.
- Suppression gains decrease slower in the lower frequencies after strong
nearend.
Change-Id: I7adf31ed90b0e007072191f40439f27c3b0bccf2
Bug: webrtc:9230,chromium:839379
Reviewed-on: https://webrtc-review.googlesource.com/73680
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23115}
The audio processing code is using parts of the iSAC codec to do voice
activity detection (VAD), but it's undesirable for it to pull in the
entire iSAC codec as a dependency. So this CL factors out the parts of
iSAC that's needed for VAD to a separate build target.
Bug: webrtc:8396
Change-Id: I884e25d8fd0bc815fca664352b0573b4b173880e
Reviewed-on: https://webrtc-review.googlesource.com/69640
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23110}
RNN implementation for the AGC2 VAD that includes a fully connected
layer and a gated recurrent unit layer.
Bug: webrtc:9076
Change-Id: Ibb8b0b4e9213f09eb9dbe118bbdc94d7e8e4f91b
Reviewed-on: https://webrtc-review.googlesource.com/72060
Reviewed-by: Patrik Höglund <phoglund@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Ivo Creusen <ivoc@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23101}
Functions to estimate pitch period and gain.
Bug: webrtc:9076
Change-Id: Icfe9430dcae11bdb96165c5bfe6e2b1d3bf848ab
Reviewed-on: https://webrtc-review.googlesource.com/70382
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23066}
This CL removes the updating of the buffered data used to to pad the
64 sample blocks to 128 samples FFTs. As that padding was used
incorrectly in one place this resolves an important issue.
Bug: webrtc:9159,chromium:833801,webrtc:9206
Change-Id: Ie6830878ebec6130b61d4e7e3169357f2e253073
Reviewed-on: https://webrtc-review.googlesource.com/73240
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23059}
This CL changes the way the suppressor gain is computed in AEC3 in that
the FFTs used are padded with data and windowed with a Hanning-style
window.
This gives better FFT accuracy, an behavior matching the suppressor
gain application, and also results in one less FFT operation.
Bug: webrtc:9204,chromium:837563
Change-Id: I612676c389cb76a3130966a9b596ff3f44d21863
Reviewed-on: https://webrtc-review.googlesource.com/73141
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23057}
If the adaptive gain is too low, we raise it slowly and only during
speech.
The CL gives better behavior at the start of a call. If the gain is too
high, the fixed-digital limits it. The gain is also quickly reduced by
the AdaptiveGainApplier.
Bug: webrtc:7494
Change-Id: I683f1e3e463cddec2d91f6c7f15c73e744430034
Reviewed-on: https://webrtc-review.googlesource.com/71484
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23053}
This reverts commit b04e5cae08.
Reason for revert: The reason for the revert is that some scenarios were detected where this caused the delay estimation to occur too slowly.
Original change's description:
> Making the delay estimator more robust to noisy nearends and low echoes
>
> This CL reduces the delay estimator step size to make it react better in
> scenarios where the environment is noisy, or the echo level is fairly
> low.
>
> Bug: webrtc:9177,chromium:835281
> Change-Id: I482d898c91eddc497e1284ee500d26df21a0574a
> Reviewed-on: https://webrtc-review.googlesource.com/71486
> Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
> Commit-Queue: Per Åhgren <peah@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#22990}
TBR=gustaf@webrtc.org,peah@webrtc.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: webrtc:9177, chromium:835281
Change-Id: I33e09ebfed8ad8330419e554f482c956608befce
Reviewed-on: https://webrtc-review.googlesource.com/72843
Reviewed-by: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Oleh Prypin <oprypin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23042}
This CL makes sure that the coherence-based gains are affected by the
upper gain limit during call start-up and after resets.
Bug: webrtc:9159,chromium:833801
Change-Id: I93fdd173b6e11ea861d0e01e12c048ec0a91db70
Reviewed-on: https://webrtc-review.googlesource.com/72841
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23039}
This CL is created from a work initiated at https://webrtc-review.googlesource.com/c/src/+/61160
The purpose of this work is to improve the performance of the echo canceler (AEC3) when the farend signal contains stationary noises:
- An stationarity estimator of the farend signal has been added for detecting the portions of the farend signal that are pure noise.
- When the echo canceler deals with a portion of the signal that contains basically noise, the echo suppressor is able to back-off and avoid the fading of the nearend speech.
Change-Id: Id4b87fc59f4765bf1fca36d1cab39a49aabe104a
Bug: webrtc:9193,chromium:836790
Reviewed-on: https://webrtc-review.googlesource.com/64141
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23024}
The code that attenuates narrow banded echo peaks in low frequencies
is removed as it affects transparency negatively.
Bug: webrtc:9192,chromium:836729
Change-Id: Ib90ce6a3db0a75e8d69bdca432e1f8f8bfbbd988
Reviewed-on: https://webrtc-review.googlesource.com/72380
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23022}
This CL overrides the power-based suppressor gain decision with
a coherence based descision for the cases when that indicates a
higher suppressor gain.
Bug: webrtc:9159,chromium:833801
Change-Id: I0e7d82ac1b8c70ffe9d45907559bb14b1b849d71
Reviewed-on: https://webrtc-review.googlesource.com/71660
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22997}
This CL reduces the delay estimator step size to make it react better in
scenarios where the environment is noisy, or the echo level is fairly
low.
Bug: webrtc:9177,chromium:835281
Change-Id: I482d898c91eddc497e1284ee500d26df21a0574a
Reviewed-on: https://webrtc-review.googlesource.com/71486
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22990}
Only specially taggged targets may transitively depend on poisonous
targets. We first apply it to audio codecs.
This makes it much clearer exactly what parts of the code still have
dependencies on the audio codecs (and we want to eventually get rid of
pretty much all of them).
Bug: webrtc:8396, webrtc:9121
Change-Id: Iba5c2e806c702b5cfe881022674705f647896d43
Reviewed-on: https://webrtc-review.googlesource.com/69520
Commit-Queue: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Patrik Höglund <phoglund@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22979}
- No need to have a unique ptr for the swap queue
- Remove default case from the switch in
AudioProcessingImpl::HandleRuntimeSettings()
Bug: webrtc:9138
Change-Id: I346ba1db6510b5caa637510298b67ead07197b81
Reviewed-on: https://webrtc-review.googlesource.com/71164
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22958}
Functions to estimate the inverse filter via LPC and compute the LP
residual applying the inverse filter.
This CL also includes test utilities, in particular BinaryFileReader,
used to read chunks of data and optionally cast them on the fly, and
Create*Reader() functions to read resource files available at test
time.
Bug: webrtc:9076
Change-Id: Ia4793b8ad6a63cb3089ed11ddad89d1aa0b840f6
Reviewed-on: https://webrtc-review.googlesource.com/70244
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22946}
Adding a data structure to cache the results of pair-wise comparisons
between items stored in a ring buffer. This is used to avoid recomputing
the pair-wise comparison every time that a new item is added in a ring
buffer.
Bug: webrtc:9076
Change-Id: I88fb67a80bd3fd8497764dc7ae7e0a577c06b20f
Reviewed-on: https://webrtc-review.googlesource.com/70162
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22942}
Ring buffer template for a finite number of arrays of given type and size.
Bug: webrtc:9076
Change-Id: Ia6c2065b0013f4a00f693966641f9aebe09f6f5c
Reviewed-on: https://webrtc-review.googlesource.com/70161
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22939}
- protobuf library
- file_wrapper.h
These appear to have been left behind during the AecDump refactoring.
After this CL, APM no longer depends on zlib by default! :)
Bug: webrtc:9139
Change-Id: I12a8df2a17a575515b9c07165825f0879c4e15eb
Reviewed-on: https://webrtc-review.googlesource.com/70762
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Commit-Queue: Fredrik Solenberg <solenberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22923}
The SequenceBuffer class template implements a linear buffer with a Push
operation that is used to add a fixed size chunk of new samples into the
buffer. Its properties are its size and the size of the chunks that are
pushed. It is used to implement the pitch buffer in the RNN VAD feature
extractor, for which a ring buffer would be a painful choice.
Bug: webrtc:9076
Change-Id: I4767bf06d5a414dbed724a96ea4186ef013a1e30
Reviewed-on: https://webrtc-review.googlesource.com/70204
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22919}
This CL adds support for using any externally reported audio buffer
delay to set the initial alignment in AEC3 which is used before the
AEC has been able to detect the delay.
Bug: chromium:834182,webrtc:9163
Change-Id: Ic71355f69b7c4d5815b78e49987043441e7908fb
Reviewed-on: https://webrtc-review.googlesource.com/70580
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22917}
This CL increases the transparency in AEC3 during regions of low level
echo. What is done is:
-Low-level echoes are smoothly weighted so as to be deemed less
disturbing.
-The time-domain masking effect of the nearend speech is increased for
all frequencies.
-A separate, even more increased, time-domain masking effect is
introduced for lower frequencies.
-The intra-band masking is reduced to reduce the risk of echo leakage.
-The limiting of maximum gain due to filter-bank dynamics is removed
as the usecase for it could no longer be identified.
Bug: webrtc:9159,cromium:833801
Change-Id: I289b92919763124d6c5e5ede19e9a5917877c654
Reviewed-on: https://webrtc-review.googlesource.com/70421
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22915}
This reverts commit 8628f5bb7c.
Reason for revert: iOS buildbot failing
Original change's description:
> AGC2 RNN VAD: initial build targets
>
> rnn_vad_tool is an executable that reads a wav file of any sample rate
> compatible with 10 ms frames that are resampled and, when the VAD is
> fully landed, will process the resampled frames to compute the VAD
> probability.
>
> To avoid mac, win and ios trybot failures, to_be_removed.h/.cc have
> been added and will be removed as soon as the :lib target includes
> code that leads to a non-empty static lib file on those platforms.
>
> Bug: webrtc:9076
> Change-Id: I810c08acfa1adf2029e3baac2adda3045ae5214a
> Reviewed-on: https://webrtc-review.googlesource.com/70202
> Reviewed-by: Alex Loiko <aleloi@webrtc.org>
> Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#22898}
TBR=alessiob@webrtc.org,aleloi@webrtc.org
Change-Id: Ic6014dde78b0ef371804c52608145ba8acdd9c97
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:9076
Reviewed-on: https://webrtc-review.googlesource.com/70144
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22899}
rnn_vad_tool is an executable that reads a wav file of any sample rate
compatible with 10 ms frames that are resampled and, when the VAD is
fully landed, will process the resampled frames to compute the VAD
probability.
To avoid mac, win and ios trybot failures, to_be_removed.h/.cc have
been added and will be removed as soon as the :lib target includes
code that leads to a non-empty static lib file on those platforms.
Bug: webrtc:9076
Change-Id: I810c08acfa1adf2029e3baac2adda3045ae5214a
Reviewed-on: https://webrtc-review.googlesource.com/70202
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22898}
Commit bbf21a3fd6 ("Remove dependencies on
modules:module_api from AudioProcessing") causes the build to fail with
libstdc++ due to several files using memcpy(3) or memset(3) while relying on
string.h being included implicitly by other headers.
Bug: webrtc:9139
Change-Id: Ib73284962f8694d8bed0551968265bfd13cab967
Reviewed-on: https://webrtc-review.googlesource.com/70180
Reviewed-by: Fredrik Solenberg <solenberg@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Raphael Kubo da Costa (rakuco) <raphael.kubo.da.costa@intel.com>
Cr-Commit-Position: refs/heads/master@{#22895}
