Adds two metrics for stereo detection:
- An enum indicating whether the last 10 seconds contained persistent stereo content or not, logged every 10 seconds.
- An enum indicating whether any persistent stereo content at all has been detected, logged at the end of the AEC lifetime.
These metrics allow us to assess:
- What proportion of all audio is treated as stereo.
- What proportion of sessions encounter any significant stereo content. If this is unexpectedly high, the stereo detection code may need fine tuning.
Metrics are only logged for component lifetimes exceeding 5 seconds: This is to avoid brief AEC lifetimes due to internal resets etc within APM.
Corresponding Chrome CL for XML histogram declarations:
https://crrev.com/c/3579317
Bug: chromium:1295710
Change-Id: I93e2bf74588cf4bb2a8922dbfad079bccab01456
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258760
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36537}
During temporary stereo content when the AEC3 uses a mono reference signal, the signal is downmixed by averaging instead of using only the left channel.
Additionally, temporary stereo content is flagged as an echo path change.
Tested: Modified local build: Verified stereo mode entered / left in accordance with hysteresis and timeout thresholds. Verified temporary stereo detected during temporary stereo playout. Made an aecdump and inspected content.
Bug: chromium:1295710
Change-Id: I6bd53e615dfb3ec39bc1c73275b7d6d599ac7c57
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258481
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36504}
Even if playout audio is only very briefly stereo, the AEC will enter stereo processing mode. To save CPU and improve AEC performance, this CL adds a hysteresis period before treating playout as stereo.
The feature is enabled by default in the AEC3 config.
Bug: chromium:1295710
Change-Id: I29116ab2e7823e25a02aa3b66a1c619f1d966d9e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258479
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36503}
If playout audio is temporarily stereo, the AEC will currently enter stereo processing mode indefinitely. To save CPU and improve AEC performance, this CL adds support for falling back to mono after a period of no stereo.
The feature is enabled by default in the AEC3 config.
Bug: chromium:1295710
Change-Id: I690b5b22f8407f950bf41f3bcaa9ca0138452157
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258421
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36502}
This CL adds a component in the TS implementation to return a delayed
version of the voice probability values observed when `Suppress()` is
called. That is needed in order to temporally align the voice
probability values to the processed audio since TS adds algorithmic
delay.
Bug: webrtc:13663
Change-Id: I5041ace3939d2ce7ba084ae703428e66f1aa06be
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/255860
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36496}
The values returned by `TransientSuppressor::Initialize()` and by
`TransientSuppressor::Suppress()` are never used.
Bug: webrtc:13663
Change-Id: I20b8afb5a66f49e5ebaf132acf8bcd1c4292a5f8
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/255822
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36492}
More robust API option that allows to fully initialize TS when created.
Bug: webrtc:13663
Change-Id: I42c38612ef772eb6d0bbde49d04ea39332a0e3c7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/255821
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36490}
Adding a delay unit to be used in the APM Transient Suppressor (TS)
sub-module through which the observerd voice probabilities are
temporally aligned to the audio processed by TS, which introduces
algorithmic delay.
Bug: webrtc:13663
Change-Id: I2136c303914580851c742d8db89478a13b06dacb
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/255680
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36487}
It is now required to specify which VAD is used to compute the speech
probability passed when `TransientSuppressor::Suppress()` is called.
In this way, it is possible to adapt parameters and/or logic of a
`TransientSuppressor` implementation to the behavior of the used
VAD. This CL also adds a "no VAD" mode option, which ignores the speech
probability argument passed when `Suppress()` and always applies mild
suppression to preserve transparency.
Finally, this CL adds a field trial to choose which VAD is used by
APM for transient suppression. Wiring the RNN VAD to TS will be done
in a follow-up CL.
Bug: webrtc:13663
Change-Id: I21ed49f91875a4ee0f04db97ea87c0dbc3db7f8a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/250962
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36485}
The features have two safety fallbacks:
- multichannel config has a killswitch WebRTC-Aec3SetupSpecificDefaultConfigDefaultsKillSwitch
- stereo detection has a killswitch WebRTC-Aec3StereoContentDetectionKillSwitch
Both features are enabled by default in the AEC3 config.
Tested: Bitexact on a large number of aecdumps.
Bug: chromium:1295710
Change-Id: I340cdc9140dacd4ca22d0911eb9f732b6cf8b226
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258129
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36482}
Apart from making the construction more straightforward, this change allows recreating the BlockProcessor at runtime. This is used to change parameterization at runtime in an upcoming CL [1].
[1] https://webrtc-review.googlesource.com/c/src/+/258129
Tested: Bitexact on a large number of aecdumps.
Bug: chromium:1295710
Change-Id: I2e0275c5c97044cb4370042633266b193c06b960
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/258100
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36473}
Motivation: never used.
Follow-up CL of https://webrtc-review.googlesource.com/c/src/+/250680.
Tested on a custom chromium build by making a video call and with
keyboard activity. The expected logs from the APM transient suppressor
sub-module were shown.
Bug: webrtc:11226
Change-Id: I4186994412dd8ba2e71ed8f9dcc9cf8f8e40fbd3
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/250667
Reviewed-by: Niels Moller <nisse@webrtc.org>
Reviewed-by: Per Kjellander <perkj@webrtc.org>
Reviewed-by: Björn Terelius <terelius@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36300}
Add a flag to override the key pressed state when simulating APM.
The current behavior changes as follows:
- Wav files simulation: instead of simulating continuous key press
events only if the transient suppressor (TS) sub-module is active,
allow to simulate the events regardless of whether TS is used;
the default key pressed state is used if the command line flag is
unspecified, otherwise it is overridden (either always false or
always true)
- AEC dump simulation: instead of simulating continuous key press
events when `--ts 2` is specified, allow to simulate the events
regardless of whether TS is used; the state recorded in the AEC
dump is used if the command line flag is unspecified, otherwise
it is overridden (either always false or always true)
- The `--ts 2` option (continuous key events) is now equivalent to
`--ts 1`.
Bug: webrtc:13663
Change-Id: I5ebe96283db73ee235ec2b2795d91d4e241a3527
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/256003
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36247}
Increase the history size of clipping_predictor_evaluator_. Use one-sample
accuracy in clipping detection for the evaluator.
Bug: webrtc:12774
Change-Id: I8c1bbfe69fe55af73ce14992e49ef7295b3ce926
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/241602
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Hanna Silen <silen@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36218}
- missing negation causes the opposite behavior when
`analog_agc_disable_digital_adaptive` is used
- flag replaced with `analog_agc_use_digital_adaptive_controller`
which is less error-prone
Bug: webrtc:7494
Change-Id: If9e0ba4fc9e539c73269faf9096ca782620dac6d
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/251322
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36113}
Also test the field trial with valid parameter and non-empty suffix.
Bug: webrtc:7494
Change-Id: I3d871b41dd71c951ac56e180b3c09cda4c3627d4
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/251441
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#36031}
This enum is no longer needed. Also moving the last piece of code from
common.h to audio_processing_impl.h, allowing to delete common.h.
Bug: chromium:1271981, b/217349489
Change-Id: If115336c36d6d7b5845a903e421c18aebfe434ee
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/251242
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35946}
When the minimum mic level is overridden via the field trial named
WebRTC-Audio-AgcMinMicLevelExperiment, AGC1 can still lower the gain
beyond the minimum value (namely, when clipping is observed).
This CL changes the behavior of the field trial. When specified, the
override always applies and therefore the mic level is guaranteed to
never become lower than what the field trial specifies.
Tested: RTC call in Chromium with and without --force-fieldtrials="
WebRTC-Audio-AgcMinMicLevelExperiment/Enabled-255"
Bug: chromium:1275566
Change-Id: I42ff45add54c11084f5ca6a2b95887c627c3c3aa
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/250141
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35914}
- Switch from ptr+size to rtc::ArrayView
- Remove `AgcManagerDirect::sample_rate_hz_` since it's always 16 kHz
- Stop passing nullptr in agc_manager_direct_unittest.cc when
`AgcManagerDirect::Process()` is called
- Allow to correctly run the tests added in the child CL (see [1])
[1] https://webrtc-review.googlesource.com/c/src/+/250141
Bug: webrtc:7494
Change-Id: I0292d7038d6510ca7c58af32b6003a1e4b121b00
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/250541
Reviewed-by: Hanna Silen <silen@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35910}
The 6-parameter Initialize method is removed. The has_keyboard parameter
in the StreamConfig constructor is removed together with the underlying
member and helper functions.
Bug: chromium:1271981, b/217349489
Change-Id: I7259a114a395f74f735a9c06510c0fc0f0f008e3
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/250221
Reviewed-by: Sam Zackrisson <saza@google.com>
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Auto-Submit: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35908}
This CL removes even more top-level const from parameters in function
declarations. This change is safe because top-level const in function
declarations (not function definitions) are ignored by the compiler
and so change is just a no-op cleanup.
Bug: webrtc:13610
Change-Id: Icf6868c27b1fdb9d9915b3a7020eb34bdcf07a09
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/249989
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Ali Tofigh <alito@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35866}
This is a safe cleanup change since top-level const applied to
parameters in function declarations (that are not also
definitions) are ignored by the compiler. Hence, such changes do
not change the type of the declared functions and are simply
no-ops.
Bug: webrtc:13610
Change-Id: Ibafb92c45119a6d8bdb6f9109aa8dad6385163a9
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/249086
Reviewed-by: Niels Moller <nisse@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Ali Tofigh <alito@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35802}
Just applied a short sed script. See bug description for
the motiviation for this change.
This is the command that was used to generate the changes:
$ find . -type f \( -iname '*.cc' -o -iname '*.h' \) -print0 | \
xargs -0 sed -i -e 's/(const override)/(const, override)/'
Bug: webrtc:13090
Change-Id: Iec7d280f9d55263a972dbb3bd644ebfcd2eb38cf
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/249088
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Ali Tofigh <alito@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35801}
Changing to an index for-loop (instead of using std::max_element & std::distance) tracking even & odd elements separately allows the compiler to produce code with less pipeline stall.
Bug: None
Change-Id: Iaa3e820a3a3b61e2eb276f0dac9106c848db1891
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/240061
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Christian Schuldt <cschuldt@google.com>
Cr-Commit-Position: refs/heads/main@{#35729}
Manually unrolling the multiply-and-accumulate loop of the matched filter allows interleaving of instruction, which gives a significant saving.
Bug: None
Change-Id: Ie7a7d92bd453d81e9dd61812781a7b6d62e1f1f4
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/240321
Reviewed-by: Ivo Creusen <ivoc@webrtc.org>
Commit-Queue: Christian Schuldt <cschuldt@google.com>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35566}
Important: This change does not in any way affect echo cancellation or standardized stats. The user audio experience is unchanged. Only non-standard stats are affected. Echo return loss metrics are unchanged. Residual echo likelihood {recent max} will no longer be computed by default.
Important: The echo detector is no longer enabled by default.
API change, PSA: https://groups.google.com/g/discuss-webrtc/c/mJV5cDysBDI/m/7PTPBjVHCgAJ
This CL removes the default usage of the residual echo detector in APM.
It can now only be used via injection and the helper function webrtc::CreateEchoDetector. See how the function audio_processing_unittest.cc:CreateApm() changed, for an example.
The echo detector implementation is marked poisonous, to avoid accidental dependencies.
Some cleanup is done:
- EchoDetector::PackRenderAudioBuffer is declared in one target but is defined in another target. It is not necessary to keep in the API. It is made an implementation detail, and the echo detector input is documented in the API.
- The internal state of APM is large and difficult to track. Submodule pointers that are set permanently on construction are now appropriately marked const.
Tested:
- existing + new unit tests
- audioproc_f is bitexact on a large number of aecdumps
Bug: webrtc:11539
Change-Id: I00cc2ee112fedb06451a533409311605220064d0
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/239652
Reviewed-by: Ivo Creusen <ivoc@webrtc.org>
Reviewed-by: Per Kjellander <perkj@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35550}
Reducing pointer following. This will allow the compiler to optimize more efficiently with the "-fno-strict-aliasing" flag.
Bug: None
Change-Id: I8e2d841fa543b28c59eb08c654a2b0515ab39d69
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/241780
Reviewed-by: Sam Zackrisson <saza@webrtc.org>
Commit-Queue: Christian Schuldt <cschuldt@google.com>
Cr-Commit-Position: refs/heads/main@{#35548}
Unloading states and coefficients to local variables avoids excessive memory access when building with "-fno-strict-aliasing".
Bug: None
Change-Id: I90bf81ae794c21e9e41500c5040387cf67ebdd38
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/240320
Reviewed-by: Ivo Creusen <ivoc@webrtc.org>
Commit-Queue: Christian Schuldt <cschuldt@google.com>
Cr-Commit-Position: refs/heads/main@{#35518}
