zqz/platform-external-webrtc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

InputVolumeController: Make speech_probability non-optional

Browse Source 
Make the argument speech_probability non-optional in
InputVolumeController::Process() and
MonoInputVolumeController::Process().

Additional clean-up: Remove the flag enabled in the
config. Add unit tests for MonoInputVolumeController.

Bug: webrtc:7494
Change-Id: Ie28af77dc628bf71d09ce1ff033d39031f77a21e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/283700
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Hanna Silen <silen@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#38710}
This commit is contained in:
Hanna Silen
2022-11-22 15:00:58 +01:00
committed by WebRTC LUCI CQ
parent 310e0624aa
commit 27fed4513f
5 changed files with 568 additions and 267 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
modules/audio_processing/agc2/input_volume_controller.cc
View File

@ -187,14 +187,13 @@ void MonoInputVolumeController::Initialize() {
is_first_frame_ = true;
}
// A speech segment is considered active if at least
// A speeh segment is considered active if at least
// `update_input_volume_wait_frames_` new frames have been processed since the
// previous update and the ratio of non-silence frames (i.e., frames with a
// non-empty `speech_probability` value above `speech_probability_threshold_`)
// is at least `speech_ratio_threshold_`.
void MonoInputVolumeController::Process(
absl::optional<int> rms_error_dbfs,
absl::optional<float> speech_probability) {
// `speech_probability` higher than `speech_probability_threshold_`) is at least
// `speech_ratio_threshold_`.
void MonoInputVolumeController::Process(absl::optional<int> rms_error_dbfs,
float speech_probability) {
if (check_volume_on_next_process_) {
check_volume_on_next_process_ = false;
// We have to wait until the first process call to check the volume,
@ -203,8 +202,7 @@ void MonoInputVolumeController::Process(
}
// Count frames with a high speech probability as speech.
if (speech_probability.has_value() &&
*speech_probability >= speech_probability_threshold_) {
if (speech_probability >= speech_probability_threshold_) {
++speech_frames_since_update_input_volume_;
}
@ -364,8 +362,7 @@ void MonoInputVolumeController::UpdateInputVolume(int rms_error_dbfs) {
InputVolumeController::InputVolumeController(int num_capture_channels,
const Config& config)
: analog_controller_enabled_(config.enabled),
num_capture_channels_(num_capture_channels),
: num_capture_channels_(num_capture_channels),
min_mic_level_override_(GetMinMicLevelOverride()),
capture_output_used_(true),
clipped_level_step_(config.clipped_level_step),
@ -384,9 +381,8 @@ InputVolumeController::InputVolumeController(int num_capture_channels,
target_range_max_dbfs_(config.target_range_max_dbfs),
target_range_min_dbfs_(config.target_range_min_dbfs),
channel_controllers_(num_capture_channels) {
RTC_LOG(LS_INFO) << "[agc] analog controller enabled: "
<< (analog_controller_enabled_ ? "yes" : "no");
const int min_mic_level = min_mic_level_override_.value_or(kMinMicLevel);
RTC_LOG(LS_INFO) << "[agc] Input volume controller enabled";
RTC_LOG(LS_INFO) << "[agc] Min mic level: " << min_mic_level
<< " (overridden: "
<< (min_mic_level_override_.has_value() ? "yes" : "no")
@ -475,10 +471,12 @@ void InputVolumeController::AnalyzePreProcess(const AudioBuffer& audio_buffer) {
}
}
}
if (clipping_detected) {
RTC_DLOG(LS_INFO) << "[agc] Clipping detected. clipped_ratio="
<< clipped_ratio;
}
int step = clipped_level_step_;
if (clipping_predicted) {
predicted_step = std::max(predicted_step, clipped_level_step_);
@ -487,6 +485,7 @@ void InputVolumeController::AnalyzePreProcess(const AudioBuffer& audio_buffer) {
step = predicted_step;
}
}
if (clipping_detected ||
(clipping_predicted && use_clipping_predictor_step_)) {
for (auto& state_ch : channel_controllers_) {
@ -497,10 +496,11 @@ void InputVolumeController::AnalyzePreProcess(const AudioBuffer& audio_buffer) {
clipping_predictor_->Reset();
}
}
AggregateChannelLevels();
}
void InputVolumeController::Process(absl::optional<float> speech_probability,
void InputVolumeController::Process(float speech_probability,
absl::optional<float> speech_level_dbfs) {
AggregateChannelLevels();
@ -509,7 +509,7 @@ void InputVolumeController::Process(absl::optional<float> speech_probability,
}
absl::optional<int> rms_error_dbfs;
if (speech_probability.has_value() && speech_level_dbfs.has_value()) {
if (speech_level_dbfs.has_value()) {
// Compute the error for all frames (both speech and non-speech frames).
rms_error_dbfs = GetSpeechLevelErrorDb(
*speech_level_dbfs, target_range_min_dbfs_, target_range_max_dbfs_);
@ -527,14 +527,11 @@ void InputVolumeController::HandleCaptureOutputUsedChange(
for (auto& controller : channel_controllers_) {
controller->HandleCaptureOutputUsedChange(capture_output_used);
}
capture_output_used_ = capture_output_used;
}
void InputVolumeController::set_stream_analog_level(int level) {
if (!analog_controller_enabled_) {
recommended_input_volume_ = level;
}
for (auto& controller : channel_controllers_) {
controller->set_stream_analog_level(level);
}
@ -559,9 +556,7 @@ void InputVolumeController::AggregateChannelLevels() {
std::max(new_recommended_input_volume, *min_mic_level_override_);
}
if (analog_controller_enabled_) {
recommended_input_volume_ = new_recommended_input_volume;
}
recommended_input_volume_ = new_recommended_input_volume;
}
} // namespace webrtc