This change disables the ERLE estimation of onsets and instead assumes
minimum ERLE. This reduces the risk of echo leaks during onsets. The
estimated ERLE was sometimes incorrect due to:
- Not enough data to train on.
- Platform noise suppression can change the echo-path.
Bug: chromium:119942,webrtc:10341
Change-Id: I1dd1c0f160489e76eb784f07e99af02f44f387ec
Reviewed-on: https://webrtc-review.googlesource.com/c/123782
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26794}
This CL removes the legacy, no longer used, render buffering code. It
also removes four unused parameters from the AEC3 config. The change
is tested for bit-exactness.
Bug: webrtc:8671
Change-Id: I2bb6cb7a1097863f228767d757d551c00593bb00
Reviewed-on: https://webrtc-review.googlesource.com/c/119701
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26399}
In this CL a warning is avoided in the reverberation decay estimator code. The change is bitexact.
Bug: chromium:921582
Change-Id: I5a91f4b5970a21ba6da7254cf7fad8c2d0bcac4b
Reviewed-on: https://webrtc-review.googlesource.com/c/118441
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26342}
This reverts commit 74ba99062c.
Reason for revert: Breaks downstream project.
Original change's description:
> AEC3: Lockless transfer of render data to the capture thread
>
> This CL implements a lockless queue that replaces SwapQueue
> in the RenderWriter. This avoid stalls when the render and
> capture threads are accessing the queue at the same time.
>
> Bug: webrtc:10205
> Change-Id: Ie7d6fcf9c80fad957e2a90537658fb730ca2ed72
> Reviewed-on: https://webrtc-review.googlesource.com/c/117643
> Reviewed-by: Per Åhgren <peah@webrtc.org>
> Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#26298}
TBR=gustaf@webrtc.org,peah@webrtc.org
Change-Id: Ie76ee8835da4e44982d181a152c9ffa19ff33e23
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:10205
Reviewed-on: https://webrtc-review.googlesource.com/c/118142
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26304}
This CL implements a lockless queue that replaces SwapQueue
in the RenderWriter. This avoid stalls when the render and
capture threads are accessing the queue at the same time.
Bug: webrtc:10205
Change-Id: Ie7d6fcf9c80fad957e2a90537658fb730ca2ed72
Reviewed-on: https://webrtc-review.googlesource.com/c/117643
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26298}
Comfort noise was generated by picking random angles on the unit circle
for each frequency band and then obtaining points on the unit circle from
{cos(a), -sin(a)}.
In order to reduce complexity, this change introduces a randomly indexed
table of 32 elements over sin(a). cos(a) is obtained by adding an offset
corresponding to pi/2 to the index. The table is pre-scaled by sqrt(2) to
avoid later multiplications.
This change reduces the computational complexity of AEC3 by ~8% with no
audible degradation.
Bug: webrtc:10189
Change-Id: I8cfe2469022fb1fe910ab3f966e55d9d499b7161
Reviewed-on: https://webrtc-review.googlesource.com/c/116787
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26209}
Fixes the ENR threshold used in the dominant nearend detection when
the kill-switch WebRTC-Aec3UseLegacyNormalSuppressorTuning is pulled.
Bug: webrtc:8671,chromium:911141
Change-Id: I30ee58009633b3a9e12eff692226baada624a049
Reviewed-on: https://webrtc-review.googlesource.com/c/112903
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25880}
Correcting a mistake in the dominant nearend detection where
the meaning of the echo-to-nearend ratio was inversed.
Bug: webrtc:8671
Change-Id: I7f56369fad1784e256150c312b6b3dafcb9d0f71
Reviewed-on: https://webrtc-review.googlesource.com/c/112136
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25818}
In this CL the analysis of the impulse response that is done in the FilterAnalyzed class is changed in order to reduce its complexity. Instead of analyzing the whole impulse response in each Update call a smaller region is analyzed. That region is changed at each Update call which implies that several calls are needed in order to analyze the complete impulse response.
Bug: webrtc:10032,chromium:909007
Change-Id: Ic58be34ba18485311c63e0fed9b6e892f9cb864c
Reviewed-on: https://webrtc-review.googlesource.com/c/111602
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25817}
This change introduces a clockdrift detector operating on the estimated
delay of the echo path delay estimator. Each time the delay estimate
changes it is compared to previous estimates. If the estimates are
slowly increasing or decreasing, clockdrift is detected.
Four different patterns are considered clockdrift:
- k, k+1, k+2, k+3
- k, k+2, k+1, k+3
- k, k-1, k-2, k-3
- k, k-2, k-1, k-3
A delay estimate history matching the three last elements in one of the
patterns is considered probable clockdrift. Matching all four elements
is considered verified clockdrift.
If the delay is constant for some time after clockdrift is detected the
clockdrift detector will revert to no detected clockdrift.
The level of clockdrift is reported via an UMA histogram.
Bug: webrtc:10014
Change-Id: I1cce4d593e101a8b3fa99df6935e59b4243cb97a
Reviewed-on: https://webrtc-review.googlesource.com/c/111381
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25758}
In this CL a more precise estimation of the Erle is introduced. This is done by creating different estimators that are specialized in different regions of the linear filter. An estimation of which regions were used for generating the current echo estimate is performed and used for selecting the right Erle estimator.
Bug: webrtc:9961
Change-Id: Iba6eb24596c067c3c66d40df590be379d3e1bb7b
Reviewed-on: https://webrtc-review.googlesource.com/c/109400
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25707}
The analysis and synthesis windowing cause loss of power when
cross-fading the noise where frames are completely uncorrelated
(generated with random phase).
This CL also removes duplicate code and enables platform specific
optimizations for ARM in the comfort noise generation.
Bug: webrtc:9967,chromium:902262
Change-Id: Iffd59b301876442079d4a5f2c7fac55a3522397c
Reviewed-on: https://webrtc-review.googlesource.com/c/109581
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25526}
This change corrects the computation of the comfort noise gains.
Previously the comfort noise gain of band k, CG_k, was computed
from suppression gain of band k, SG_k, as:
CG_k = 1 - SG_k
But since the two signals are uncorrelated (the comfort noise
is randomly generated), the correct gain to maintain power is:
CG_k = sqrt(1 - SG_k^2).
Bug: webrtc:9967,chromium:902262
Change-Id: I393495742163d5e658bca4ab2f7a5067ab15af01
Reviewed-on: https://webrtc-review.googlesource.com/c/109580
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25525}
This CL utilizes the existing, but unused, ability to set
different histogram thresholds for early and late delay
estimation. It does so by tuning the parameters for these.
On top of that, some corrections are added to correctly
handle resets and the use of the hysteresis thresholds.
Bug: webrtc:19886,chromium:896334
Change-Id: I950ac107c124541af8f02b4403f477dda71cc1a1
Reviewed-on: https://webrtc-review.googlesource.com/c/106706
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25443}
This CL allows control over the dominant nearend functionality so that
it is not active during the initial phase, when estimates are less
certain.
Bug: webrtc:9906,chromium:898273
Change-Id: I5f61dac806ec3b1ebc1a3ec72f0a16d07a67f14a
Reviewed-on: https://webrtc-review.googlesource.com/c/107632
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25326}
This CL is the result of running include-what-you-use tool on part
of the code base (audio target and dependencies) plus manual fixes.
bug: webrtc:8311
Change-Id: I277d281ce943c3ecc1bd45fd8d83055931743604
Reviewed-on: https://webrtc-review.googlesource.com/c/106280
Commit-Queue: Yves Gerey <yvesg@google.com>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Patrik Höglund <phoglund@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25311}
This change makes the dominant nearend detection more accurate.
- The hangover is increased not leave nearend state between words.
- The SNR requirement is increased to not enter nearend state without
speech activity.
- An early exit mechanism has been added to leave nearend state quickly
when the echo is strong.
Bug: chromium:897701,webrtc:9897
Change-Id: I9e0f3e6ecb80eee1c0c917d4835f110555f74acf
Reviewed-on: https://webrtc-review.googlesource.com/c/107347
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25299}
In this CL we change the signal that controls the updates of the ERLE estimator. Until now, the render signal was used which is not optimum for reverberant signals. In this CL, a reverberation has been added to the the render signal and this new signal has been used for controlling when to update the ERLE estimator.
Bug: webrtc:9873
Change-Id: I0ebea3fc208f97aa237af015ba543015d49ed978
Reviewed-on: https://webrtc-review.googlesource.com/c/105660
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25285}
With this CL, the main and shadow filters are no longer fully reset to
0 as the delay changes. This allows for more robust echo removal for
some scenarios.
Bug: webrtc:9879,chromium:895838
Change-Id: I859aa3df3ae41648bc8efde01ec2e2a5cb392279
Reviewed-on: https://webrtc-review.googlesource.com/c/106345
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25251}
Removing code that has no audible effect.
Bug: webrtc:8671
Change-Id: Ibd7d0d19d760ae16b09285498c2ee09b42eb5968
Reviewed-on: https://webrtc-review.googlesource.com/c/106301
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25250}
This CL introduces two related changes
1) It changes the way that the AEC3 determines whether the linear
filter is sufficiently good for its output to be used. The new scheme
achieves this much earlier than what was done in the legacy scheme.
2) It changes the way that saturated echo is and handled so that the
impact of the nearend speech is lower.
Bug: webrtc:9835,webrtc:9843,chromium:895435,chromium:895431
Change-Id: I0b493676886e2134205e9992bbe4badac7e414cc
Reviewed-on: https://webrtc-review.googlesource.com/c/104380
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25208}
This CL simplifies the buffering of render data. Instead of making assumptions
about the worst possible platform, it leverages recent improvements in
the delay estimator to quickly adapt when the conditions change.
Pros:
- No capture delay, delay is found ~200 ms faster.
- Cleaner code that makes the concept of delay more clear.
- Allows for removal of one matched filter because of the jitter headroom
removal.
Cons:
- Delay estimator needs to re-adapt when the call jitter increases.
The code can be deactivated by a kill switch. When the kill switch is
pulled the CL is bit exact.
Bug: webrtc:9726,chromium:895338
Change-Id: Ie2f9c8c5ce5b5a4510b4bdb95db2b970b57cd5d0
Reviewed-on: https://webrtc-review.googlesource.com/c/96920
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25169}
In this CL the use of the stationarity properties at init is set to true by default.
Bug: webrtc:9865, chromium:894439
Change-Id: I716ce0d792a50616dc38cc0ba6f2c702549a81cc
Reviewed-on: https://webrtc-review.googlesource.com/c/105303
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25123}
"Perfection is achieved, not when there is nothing more to add,
but when there is nothing left to take away."
This CL removes the following kill-switches from AEC3
- WebRTC-Aec3DownSamplingFactor8KillSwitch
- WebRTC-Aec3NewSuppressionKillSwitch
- WebRTC-Aec3ShadowFilterJumpstartKillSwitch
- WebRTC-Aec3SlowFilterAdaptationKillSwitch
- WebRTC-Aec3SuppressorNearendAveragingKillSwitch
It also removes code paths and configuration parameters that are no
longer in use. The list of kill-switches in the audio processing
fuzzer test is updated.
The change has been tested for bit-exactness.
Bug: webrtc:8671
Change-Id: Ie0af86a14baf853548bf9c00b2b9b3bbc32c1aaa
Reviewed-on: https://webrtc-review.googlesource.com/c/105324
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25120}
This CL removes outdated code for testing of platforms with clock-drift
Bug: webrtc:8671
Change-Id: Ie202c514609d9f3d2357107b0daf895331275797
Reviewed-on: https://webrtc-review.googlesource.com/c/105183
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25105}
This CL ensures that the default setting for the usage of stationary signal
properties is not overridden by mistake.
Bug: chromium:894243
Change-Id: I85ab65383ee82b5f3153864da7a0cede7776c146
Reviewed-on: https://webrtc-review.googlesource.com/c/105181
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25104}
This CL changes the tuning of the echo suppressor for the case when
there is echo only. The resulting effect is a slight increase of
transparency
Bug: webrtc:9844,chromium:893744
Change-Id: I5e6a867e0d03dc3a468a8f5cfa64103e001baae1
Reviewed-on: https://webrtc-review.googlesource.com/c/104760
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25075}
This CL utilizes the AEC3 ability to tailor the suppressor during
situations when the nearend dominates over the residual echo. This is
done by increasing the thresholds for transparent echo suppressor
behavior when the nearend is strong compared to the residual echo.
Bug: webrtc:9836, chromium:893744
Change-Id: Ic06569eefc7f2557b401db43b3ac84b299071294
Reviewed-on: https://webrtc-review.googlesource.com/c/104460
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25071}
This is a new attempt to reduce the filter divergence
during double-talk without regressing in clock-drift
scenarios.
- The error_floor in decreased to allow for slow adaptation
when the filter performs well.
- The leakage_diverged is increased to allow for fast adaptation
when the shadow filter performs better.
- A new parameter, error_ceil, was added to stop the filter from
adapting too fast.
Bug: webrtc:9746,chromium:883264
Change-Id: Ie2868d2388b48412a192a004ec13f9eff34517b8
Reviewed-on: https://webrtc-review.googlesource.com/c/100460
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25063}
This CL lowers the default reverb decay to better match the standard
rooms where calls are made.
Bug: webrtc:9843
Change-Id: I46f1a629ecfdd72561829326d4fa58ede8107b6c
Reviewed-on: https://webrtc-review.googlesource.com/c/104740
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25061}
This CL introduces a major refactoring of AecState for the purpose of
simplifying further improvements to the logic in this code.
The changes have successfully been tested for bitexactness.
Bug: webrtc:8671
Change-Id: If98efde55a22c76b093089a11a0562daac7e16e6
Reviewed-on: https://webrtc-review.googlesource.com/c/102362
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24996}
In this CL the ERLE estimator is reset after a pre-amplifier gain change is communicated to APM.
Bug: webrtc:9805
Change-Id: I040f344e4607e862240250f9478d06de0d58a096
Reviewed-on: https://webrtc-review.googlesource.com/103222
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Alex Loiko <aleloi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24933}
This CL removes some deprecated build targets (and their headers)
from system_wrappers:
- field_trial_api
- field_trial_default
- metrics_api
- metrics_default
It also refreshes all the dependencies on field_trial.h and metrics.h.
A nice side effect is that it is finally possible to remove 'nogncheck'
from the following files (when it was used with field_trial_default
and metrics_default):
- sdk/objc/api/peerconnection/RTCMetricsSampleInfo+Private.h
- sdk/android/src/jni/pc/peerconnectionfactory.cc
- sdk/objc/api/peerconnection/RTCFieldTrials.mm
Bug: webrtc:9631
Change-Id: Ib621f41ef8ad0aba4fe1c1d7e749c044afc956c3
No-Try: True
Reviewed-on: https://webrtc-review.googlesource.com/100524
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24878}
This CL removes killswitches for code that has been properly tested in
experiments and is to be considered to be permanent.
The changes have been tested for bitexactness.
Bug: webrtc:8671
Change-Id: I0f9db16f377390d9dd3779096da91f3abc0fb4a5
Reviewed-on: https://webrtc-review.googlesource.com/102360
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24877}
Specially for devices with high echo path gain, even low render signal can allow the linear filter of the AEC3 to converge. However, the conditions that were used for updating the ERLE avoided to update that estimation. In this commit, we allow adapting the ERLE estimator using even low render signal but the update of the ERLE is constraint in a way that decreases are not allowed.
Bug: webrtc:9776
Change-Id: Ic4331efcc47a0b05f394cdea9a88f336292de5a1
Reviewed-on: https://webrtc-review.googlesource.com/101641
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24859}
Speeds up adaptation of the matched filter of the delay estimator by
allowing the estimated echo and the error signal (microphone minus
estimated echo) to be saturated. Only microphone saturation pauses
the filter adaptation.
Bug: webrtc:9773
Change-Id: I8b8400539fde3ee821f36a95818bece02ddd626b
Reviewed-on: https://webrtc-review.googlesource.com/101341
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24802}
Right after a volume decrease, the echo path estimate is overestimated and, as a side effect, the nearend signal is also overestimated. Due to that, the suppression gains are kept high avoiding the suppression of echoes. In this CL the neared power spectrum estimation is limited to a level given by the power spectrum or the microphone input signal. Additionally, the minimum gain that is computed inside the suppressor is also modified. Instead of using the nearend power spectrum that is now bounded, the power spectrum of the signal after the linear echo canceler is used.
Bug: webrtc:9762
Change-Id: Ia24cd2ce248f2c2ba124711b75acff3b8c5cfa9f
Reviewed-on: https://webrtc-review.googlesource.com/100720
Commit-Queue: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24796}
Replace calls to .str() which copies with .Release which moves in cases where that's safe.
This CL was generated by this command:
git grep -l 'StringBuilder' |
xargs perl -i -0 -pe "s/(rtc::StringBuilder (\S+);.*?return )\\g2.str\(\)/\$1\$2.Release\(\)/sg"
Bug: webrtc:8982
Change-Id: If4dadbeb039df010aaaa9e58da81c1971a84fe8f
Reviewed-on: https://webrtc-review.googlesource.com/100307
Commit-Queue: Jonas Olsson <jonasolsson@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24790}
