This CL adds the GainCurveApplier (GCA). It owns a
FixedDigitalLevelEstimator (LE) and an InterpolatedGainCurve
(IGC). The GCA uses the LE to compute the input signal level, looks up
a gain from IGC and applies it on the signal.
The other IGC and LE submodules were added in previous CLs [1] and
[2].
This CL also turns on AGC2 in the APM fuzzer.
[1] https://webrtc-review.googlesource.com/c/src/+/51920
[2] https://webrtc-review.googlesource.com/c/src/+/52381
Bug: webrtc:7949
Change-Id: Idb10cc3ca9d6d2e4ac5824cc3391ed8aa680f6cd
Reviewed-on: https://webrtc-review.googlesource.com/54361
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22103}
Targets containing files in api/audio are moved from api/BUILD.gn to
api/audio/BUILD.gn.
Bug: webrtc:8844
Change-Id: Ib7ea4b7eb3c2ea38ef8261a1fc5c2b4674985981
Reviewed-on: https://webrtc-review.googlesource.com/54360
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22074}
This CL adds the Level Estimator of the new gain controller. The Level
Estimator divides a 10ms input frame in kSubFramesInFrame=20 sub
frames. We take the maximal sample values in every sub frame. We then
apply attack/decay smoothing. This is the final level estimate.
The results will be used with InterpolatedGainCurve (see this CL
https://webrtc-review.googlesource.com/c/src/+/51920). For every level
estimate value, we look up a gain with
InterpolatedGainCurve::LookUpGainToApply. This gain is then applied to
the signal.
Bug: webrtc:7949
Change-Id: I2b4b3894a3e945d3dd916ce516c79abacb2b18b1
Reviewed-on: https://webrtc-review.googlesource.com/52381
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22054}
The FixedGainController (FGC) applies a fixed gain. It will also
control the limiter. The limiter will be landed over the next several
CLs.
The GainController2 is a 'private submodule' of APM. It will control
the new automatic gain controller (AGC). It controls the AGC through
Initialize() and ApplyConfig().
This CL contains
* build changes to make modules/audio_processing/agc2 an independent
target
* a new MutableFloatAudioFrame which is the audio interface between
AGC2 and APM
* move of the fixed gain application from GainController2 to
FixedGainController.
If you are a googler, there is more information in this doc:
https://docs.google.com/document/d/1RV2Doet3MZtUPAHVva61Vjo20iyd1bmmm3aR8znWpzo/edit#
Bug: webrtc:7949
Change-Id: Ief95cbbce83c3aafe54638fd2ab881c9fb8bdc3a
Reviewed-on: https://webrtc-review.googlesource.com/50440
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Oskar Sundbom <ossu@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22046}
The functions replace some existing code and will be used in the
the new AutomaticGainController.
Bug: webrtc:7949
Change-Id: I9a32132d4a4699a507b8548a2eac10972a2f3fd6
Reviewed-on: https://webrtc-review.googlesource.com/53141
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22045}
Avoid including audio_processing.h from within AEC3.
Bug: webrtc:8844
Change-Id: I02c475c2fb84e2c24eac86baac3c7edaa08bebc0
Reviewed-on: https://webrtc-review.googlesource.com/53065
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22029}
This is one of several small steps of separating APM and AEC3.
Bug: webrtc:8844
Change-Id: Ib6e518fec5f7566cab3823ab35fcede8433f8f4e
Reviewed-on: https://webrtc-review.googlesource.com/53142
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22028}
This CL resets the AEC3 realignment functionality when a significant
and persistent skew in the number of render and capture API calls is
detected.
Bug: chromium:811658,webrtc:8879
Change-Id: Ib5c727b38f427da2a7d25eac7c939a17bdaabe74
Reviewed-on: https://webrtc-review.googlesource.com/52260
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21997}
This CL adds robustness in terms of echo removal and faster recovery
in order to regain echo canceller transparency after echo path changes.
The CL does:
-Improve the adaptation rate of the linear filter.
-Increase the look-window used before the linear filter has adapted.
-Decrease the effects of missed detection of residual echo.
-Increase the safety margin before allowing the suppressor gain to
increase.
Bug: chromium:804873,webrtc:8788
Change-Id: I28eedc4c8d0a4f0bc7b79c02d6d59bf00fddd566
Reviewed-on: https://webrtc-review.googlesource.com/48721
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21917}
WHAT: made a BUILD.gn with library and tests in the Audio Processing
Module Voice Activity Detector directory. Updated depending
code. Fixed a Clang warning.
WHY: to make it possible for a target to depend on just the VAD and
not the whole APM. There are other benefits:
* Sometimes faster compilation.
* The VAD takes up 28000 bytes of libjingle_peerconnection_so.so. Making
a peerconnection shared object file without the VAD has to be done in
steps. The first step is a custom target for the VAD. Hence this Cl.
Change-Id: Iea0207a0b5979db26baaf46b24beaefbb1c431af
BUG: webrtc:5716, webrtc:7494
Reviewed-on: https://webrtc-review.googlesource.com/47521
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Oleh Prypin <oprypin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21893}
Due to the growing number of arguments, these functions are being replaced by the AudioProcessingBuilder class.
Bug: webrtc:8668
Change-Id: Ic3936fbd47d92eac22a857a678dca5fd8c029d8b
Reviewed-on: https://webrtc-review.googlesource.com/46241
Commit-Queue: Ivo Creusen <ivoc@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21826}
If a WebRTC build target requires gmock it has to include
test/gmock.h and just depend on //test:test_support.
Unfortunately //testtest_support was a leaky abstraction because it
wasn't propagating the correct -I compiler flag. To make everything
work, all the targets that use gmock started also to depend on
//testing/gmock (even if they were not including any gmock header
directly).
This CL makes //testtest_support propagate the include path up in the
dependency chain so it is possible to remove unused dependencies.
Note: all_dependent_configs should probably be used in the original
gmock target. There is an ongoing discussion about it. This CL solves
the problem on WebRTC side and it is forward compatible.
TBR=phoglund@webrtc.org
Bug: webrtc:8603
Change-Id: If08daf2ce9a6431a6e881a236743b4ec33b59ea7
Reviewed-on: https://webrtc-review.googlesource.com/44340
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Oleh Prypin <oprypin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21776}
The AecMobile struct contains a ::farendOld field. It's type is 'short [2][80]'.
The field was initialized by
memset(&aecm->farendOld[0][0], 0, 160);
But sizeof(short) is not guaranteed to be 1. This causes use of
unititialized memory on some platforms. According to MSAN, it can
affect the output of the echo canceller.
The issue was found by the MSAN fuzzer.
This change initializes the array properly.
Bug: chromium:805396
Change-Id: Ibcaca2185cfa153e8fd826e9addfc04d7b65e417
Reviewed-on: https://webrtc-review.googlesource.com/43860
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21764}
A coherence vector cohxd is computed in
WebRtcAec_ComputeCoherence. The coherence values should theoretically
be 0 <= x <= 1. Due to the way they are computed that is not always
the case.
The coherence values are used to update an error signal
estimate hNl in webrtc::EchoSuppression. 'hNl[i]' should contain an
error magnitude for frequency 'i'.
The error magnitudes are used as a basis for exponentiation. If a
magnitude is negative, the result is NaN.
The NaNs will then spread to the output signal.
This change caps the hNl values at 0. I considered capping the
coherence values at 1. The coherence values are calculated differently
for MIPS, NEON and SSE. Therefore it's simpler to cap the hNl values
instead.
The issue was found by the AudioProcessing fuzzer.
Bug: chromium:804634
Change-Id: I8ebaa441d77c3f79d9c194a850cb2b9eed1c2024
Reviewed-on: https://webrtc-review.googlesource.com/43740
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21761}
This change
* replaces a left shift with multiplication, because the shiftee can
be negative.
* replaces a right shift (a >> b) with the expression (b >= 32 ? 0 : a >> b)
because a is a 32-bit value, and b can be >= 32.
cppreference quote relating to the second change:
"In any case, if the value of the right operand is
negative or is greater or equal to the number of bits in the promoted
left operand, the behavior is undefined."
Bug: chromium:805832 chromium:803078
Change-Id: I67db0c3fedb0af197b2205d424414a84f8fde474
Reviewed-on: https://webrtc-review.googlesource.com/43761
Reviewed-by: Oskar Sundbom <ossu@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21760}
The faster AEC3 alignment introduced recently may in
cases cause the alignment (and the AEC3) to repeatedly
reset. This CL avoids these resets by handling buffer
issues (which are triggering the resets) separately
during the initial coarse alignment phase.
Change-Id: Idf5e2ffda2591906da8060d03ec8ca73cdaedf53
Bug: webrtc:8798,chromium:805815
Reviewed-on: https://webrtc-review.googlesource.com/43480
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21758}
This change handles a special case in NoiseSuppression. The special
case was found by the AudioProcessing fuzzer.
A const copy of the capture audio stream is sent to
NoiseSuppression::AnalyzeCaptureAudio. Then audio undergoes processing
by e.g. the echo canceller. Then it's processed by
NoiseSuppression::ProcessCaptureAudio.
The special case is when the following conditions are all satisfied:
* All stream samples are constantly zero in the call to
AnalyzeCaptureAudio
* a processing component modifies it to be nonzero before the call to
ProcessCaptureAudio
* The array NoiseSuppressionC::magnPrevAnalyze is filled with
zeros. This holds after initialization.
In this case, there is a division by zero in WebRtcNs_ProcessCore. The
resulting NaN values pollute the output signal. They are only detected
several submodules later in the process chain. The NaN values cause
the EchoDetector to crash in debug mode.
There is special handling of the case when the signal is constant zero
in ProcessCore. This change avoids zero division by handling this
issue the same way.
Bug: chromium:803810 chromium:804634
Change-Id: I6d698dd0cd27e6d550b42085124300ce58533125
Reviewed-on: https://webrtc-review.googlesource.com/41282
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21745}
This CL increases the speech of the initial alignment in AEC3 by
loosening the requirements on the accuracy of the initial estimates.
Bug: webrtc:8784, chromium:804270
Change-Id: I86e2d97830843524090a1cf877965739f66dc058
Reviewed-on: https://webrtc-review.googlesource.com/40660
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21728}
The conversational_speech_generator tool now adjusts the level of
different speech segments.
Implementation:
The Turn and MultiEndCall::SpeakingTurn structs have an extra 'gain'
member. It's read and parsed in timing.cc and put in a Turn
struct. It's put in a SpeakingTurn struct in multiend_call.cc and read
and applied to the signal in simulator.cc
Bug: webrtc:7494
Change-Id: I9b82a896eb616c8b5ef14d41dfdfd085ef1d3fbb
Reviewed-on: https://webrtc-review.googlesource.com/26280
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21714}
This minor issue was found by the UBSAN fuzzer.
We have used the Godbolt compiler explorer to check that similar
changes produce identical compiled code.
Bug: chromium:803078
Change-Id: Ib3fa38c101d7bda53d8d39062cb2c0a55144305f
Reviewed-on: https://webrtc-review.googlesource.com/42580
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21713}
We have done changes to the Audio Processing fuzzer here
https://webrtc-review.googlesource.com/c/src/+/36500/6.
We ran the new version of the fuzzer locally. The UBSAN
detector found these (minor) issues.
We have used the Godbolt compiler explorer to check that similar
changes produce identical compiled code.
Bug: webrtc:7820
Change-Id: I9cc3b81e4be7cf691f878c37010ce105bc2f3e38
Reviewed-on: https://webrtc-review.googlesource.com/39264
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21605}
This CL generalizes the hysteresis behavior on the AEC3 delay estimator
to be two-sided and easier to configure.
Bug: webrtc:8671
Change-Id: Ife21c1511416e32eb3618c81178deefe332ac1e8
Reviewed-on: https://webrtc-review.googlesource.com/39267
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21604}
This adds a generic interface for an echo detector, and makes it possible to inject one into the audio processing module.
Bug: webrtc:8732
Change-Id: I30d97aeb829307b2ae9c4dbeb9a3e15ab7ec0912
Reviewed-on: https://webrtc-review.googlesource.com/38900
Commit-Queue: Ivo Creusen <ivoc@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21588}
This CL adds an nonwindowed spectrum of the linear filter error
to use in the NLP computation.
Bug: webrtc:8661
Change-Id: I45bc9bb3eb8eeac0c5d6adb414638eb12b635a27
Reviewed-on: https://webrtc-review.googlesource.com/38701
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#21583}
