/*
* Copyright (c) 2012 The WebRTC project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef MODULES_AUDIO_PROCESSING_INCLUDE_AUDIO_PROCESSING_H_
#define MODULES_AUDIO_PROCESSING_INCLUDE_AUDIO_PROCESSING_H_
// MSVC++ requires this to be set before any other includes to get M_PI.
#ifndef _USE_MATH_DEFINES
#define _USE_MATH_DEFINES
#endif
#include <math.h>
#include <stddef.h> // size_t
#include <stdio.h> // FILE
#include <string.h>
#include <vector>
#include "absl/types/optional.h"
#include "api/array_view.h"
#include "api/audio/echo_canceller3_config.h"
#include "api/audio/echo_control.h"
#include "api/scoped_refptr.h"
#include "modules/audio_processing/include/audio_processing_statistics.h"
#include "rtc_base/arraysize.h"
#include "rtc_base/ref_count.h"
#include "rtc_base/system/file_wrapper.h"
#include "rtc_base/system/rtc_export.h"
namespace rtc {
class TaskQueue;
} // namespace rtc
namespace webrtc {
class AecDump;
class AudioBuffer;
class StreamConfig;
class ProcessingConfig;
class EchoDetector;
class CustomAudioAnalyzer;
class CustomProcessing;
// Use to enable experimental gain control (AGC). At startup the experimental
// AGC moves the microphone volume up to `startup_min_volume` if the current
// microphone volume is set too low. The value is clamped to its operating range
// [12, 255]. Here, 255 maps to 100%.
//
// Must be provided through AudioProcessingBuilder().Create(config).
#if defined(WEBRTC_CHROMIUM_BUILD)
static constexpr int kAgcStartupMinVolume = 85;
#else
static constexpr int kAgcStartupMinVolume = 0;
#endif // defined(WEBRTC_CHROMIUM_BUILD)
static constexpr int kClippedLevelMin = 70;
// The Audio Processing Module (APM) provides a collection of voice processing
// components designed for real-time communications software.
//
// APM operates on two audio streams on a frame-by-frame basis. Frames of the
// primary stream, on which all processing is applied, are passed to
// `ProcessStream()`. Frames of the reverse direction stream are passed to
// `ProcessReverseStream()`. On the client-side, this will typically be the
// near-end (capture) and far-end (render) streams, respectively. APM should be
// placed in the signal chain as close to the audio hardware abstraction layer
// (HAL) as possible.
//
// On the server-side, the reverse stream will normally not be used, with
// processing occurring on each incoming stream.
//
// Component interfaces follow a similar pattern and are accessed through
// corresponding getters in APM. All components are disabled at create-time,
// with default settings that are recommended for most situations. New settings
// can be applied without enabling a component. Enabling a component triggers
// memory allocation and initialization to allow it to start processing the
// streams.
//
// Thread safety is provided with the following assumptions to reduce locking
// overhead:
// 1. The stream getters and setters are called from the same thread as
// ProcessStream(). More precisely, stream functions are never called
// concurrently with ProcessStream().
// 2. Parameter getters are never called concurrently with the corresponding
// setter.
//
// APM accepts only linear PCM audio data in chunks of 10 ms. The int16
// interfaces use interleaved data, while the float interfaces use deinterleaved
// data.
//
// Usage example, omitting error checking:
// AudioProcessing* apm = AudioProcessingBuilder().Create();
//
// AudioProcessing::Config config;
// config.echo_canceller.enabled = true;
// config.echo_canceller.mobile_mode = false;
//
// config.gain_controller1.enabled = true;
// config.gain_controller1.mode =
// AudioProcessing::Config::GainController1::kAdaptiveAnalog;
// config.gain_controller1.analog_level_minimum = 0;
// config.gain_controller1.analog_level_maximum = 255;
//
// config.gain_controller2.enabled = true;
//
// config.high_pass_filter.enabled = true;
//
// config.voice_detection.enabled = true;
//
// apm->ApplyConfig(config)
//
// apm->noise_reduction()->set_level(kHighSuppression);
// apm->noise_reduction()->Enable(true);
//
// // Start a voice call...
//
// // ... Render frame arrives bound for the audio HAL ...
// apm->ProcessReverseStream(render_frame);
//
// // ... Capture frame arrives from the audio HAL ...
// // Call required set_stream_ functions.
// apm->set_stream_delay_ms(delay_ms);
// apm->set_stream_analog_level(analog_level);
//
// apm->ProcessStream(capture_frame);
//
// // Call required stream_ functions.
// analog_level = apm->recommended_stream_analog_level();
// has_voice = apm->stream_has_voice();
//
// // Repeat render and capture processing for the duration of the call...
// // Start a new call...
// apm->Initialize();
//
// // Close the application...
// delete apm;
//
class RTC_EXPORT AudioProcessing : public rtc::RefCountInterface {
public:
// The struct below constitutes the new parameter scheme for the audio
// processing. It is being introduced gradually and until it is fully
// introduced, it is prone to change.
// TODO(peah): Remove this comment once the new config scheme is fully rolled
// out.
//
// The parameters and behavior of the audio processing module are controlled
// by changing the default values in the AudioProcessing::Config struct.
// The config is applied by passing the struct to the ApplyConfig method.
//
// This config is intended to be used during setup, and to enable/disable
// top-level processing effects. Use during processing may cause undesired
// submodule resets, affecting the audio quality. Use the RuntimeSetting
// construct for runtime configuration.
struct RTC_EXPORT Config {
// Sets the properties of the audio processing pipeline.
struct RTC_EXPORT Pipeline {
// Maximum allowed processing rate used internally. May only be set to
// 32000 or 48000 and any differing values will be treated as 48000.
int maximum_internal_processing_rate = 48000;
// Allow multi-channel processing of render audio.
bool multi_channel_render = false;
// Allow multi-channel processing of capture audio when AEC3 is active
// or a custom AEC is injected..
bool multi_channel_capture = false;
} pipeline;
// Enabled the pre-amplifier. It amplifies the capture signal
// before any other processing is done.
// TODO(webrtc:5298): Deprecate and use the pre-gain functionality in
// capture_level_adjustment instead.
struct PreAmplifier {
bool enabled = false;
float fixed_gain_factor = 1.0f;
} pre_amplifier;
// Functionality for general level adjustment in the capture pipeline. This
// should not be used together with the legacy PreAmplifier functionality.
struct CaptureLevelAdjustment {
bool operator==(const CaptureLevelAdjustment& rhs) const;
bool operator!=(const CaptureLevelAdjustment& rhs) const {
return !(*this == rhs);
}
bool enabled = false;
// The `pre_gain_factor` scales the signal before any processing is done.
float pre_gain_factor = 1.0f;
// The `post_gain_factor` scales the signal after all processing is done.
float post_gain_factor = 1.0f;
struct AnalogMicGainEmulation {
bool operator==(const AnalogMicGainEmulation& rhs) const;
bool operator!=(const AnalogMicGainEmulation& rhs) const {
return !(*this == rhs);
}
bool enabled = false;
// Initial analog gain level to use for the emulated analog gain. Must
// be in the range [0...255].
int initial_level = 255;
} analog_mic_gain_emulation;
} capture_level_adjustment;
struct HighPassFilter {
bool enabled = false;
bool apply_in_full_band = true;
} high_pass_filter;
struct EchoCanceller {
bool enabled = false;
bool mobile_mode = false;
bool export_linear_aec_output = false;
// Enforce the highpass filter to be on (has no effect for the mobile
// mode).
bool enforce_high_pass_filtering = true;
} echo_canceller;
// Enables background noise suppression.
struct NoiseSuppression {
bool enabled = false;
enum Level { kLow, kModerate, kHigh, kVeryHigh };
Level level = kModerate;
bool analyze_linear_aec_output_when_available = false;
} noise_suppression;
// Enables transient suppression.
struct TransientSuppression {
bool enabled = false;
} transient_suppression;
// Enables reporting of `voice_detected` in webrtc::AudioProcessingStats.
struct VoiceDetection {
bool enabled = false;
} voice_detection;
// Enables automatic gain control (AGC) functionality.
// The automatic gain control (AGC) component brings the signal to an
// appropriate range. This is done by applying a digital gain directly and,
// in the analog mode, prescribing an analog gain to be applied at the audio
// HAL.
// Recommended to be enabled on the client-side.
struct RTC_EXPORT GainController1 {
bool operator==(const GainController1& rhs) const;
bool operator!=(const GainController1& rhs) const {
return !(*this == rhs);
}
bool enabled = false;
enum Mode {
// Adaptive mode intended for use if an analog volume control is
// available on the capture device. It will require the user to provide
// coupling between the OS mixer controls and AGC through the
// stream_analog_level() functions.
// It consists of an analog gain prescription for the audio device and a
// digital compression stage.
kAdaptiveAnalog,
// Adaptive mode intended for situations in which an analog volume
// control is unavailable. It operates in a similar fashion to the
// adaptive analog mode, but with scaling instead applied in the digital
// domain. As with the analog mode, it additionally uses a digital
// compression stage.
kAdaptiveDigital,
// Fixed mode which enables only the digital compression stage also used
// by the two adaptive modes.
// It is distinguished from the adaptive modes by considering only a
// short time-window of the input signal. It applies a fixed gain
// through most of the input level range, and compresses (gradually
// reduces gain with increasing level) the input signal at higher
// levels. This mode is preferred on embedded devices where the capture
// signal level is predictable, so that a known gain can be applied.
kFixedDigital
};
Mode mode = kAdaptiveAnalog;
// Sets the target peak level (or envelope) of the AGC in dBFs (decibels
// from digital full-scale). The convention is to use positive values. For
// instance, passing in a value of 3 corresponds to -3 dBFs, or a target
// level 3 dB below full-scale. Limited to [0, 31].
int target_level_dbfs = 3;
// Sets the maximum gain the digital compression stage may apply, in dB. A
// higher number corresponds to greater compression, while a value of 0
// will leave the signal uncompressed. Limited to [0, 90].
// For updates after APM setup, use a RuntimeSetting instead.
int compression_gain_db = 9;
// When enabled, the compression stage will hard limit the signal to the
// target level. Otherwise, the signal will be compressed but not limited
// above the target level.
bool enable_limiter = true;
// Enables the analog gain controller functionality.
struct AnalogGainController {
bool enabled = true;
int startup_min_volume = kAgcStartupMinVolume;
// Lowest analog microphone level that will be applied in response to
// clipping.
int clipped_level_min = kClippedLevelMin;
bool enable_digital_adaptive = true;
// Amount the microphone level is lowered with every clipping event.
// Limited to (0, 255].
int clipped_level_step = 15;
// Proportion of clipped samples required to declare a clipping event.
// Limited to (0.f, 1.f).
float clipped_ratio_threshold = 0.1f;
// Time in frames to wait after a clipping event before checking again.
// Limited to values higher than 0.
int clipped_wait_frames = 300;
// Enables clipping prediction functionality.
struct ClippingPredictor {
bool enabled = false;
enum Mode {
// Clipping event prediction mode with fixed step estimation.
kClippingEventPrediction,
// Clipped peak estimation mode with adaptive step estimation.
kAdaptiveStepClippingPeakPrediction,
// Clipped peak estimation mode with fixed step estimation.
kFixedStepClippingPeakPrediction,
};
Mode mode = kClippingEventPrediction;
// Number of frames in the sliding analysis window.
int window_length = 5;
// Number of frames in the sliding reference window.
int reference_window_length = 5;
// Reference window delay (unit: number of frames).
int reference_window_delay = 5;
// Clipping prediction threshold (dBFS).
float clipping_threshold = -1.0f;
// Crest factor drop threshold (dB).
float crest_factor_margin = 3.0f;
// If true, the recommended clipped level step is used to modify the
// analog gain. Otherwise, the predictor runs without affecting the
// analog gain.
bool use_predicted_step = true;
} clipping_predictor;
} analog_gain_controller;
} gain_controller1;
// Enables the next generation AGC functionality. This feature replaces the
// standard methods of gain control in the previous AGC. Enabling this
// submodule enables an adaptive digital AGC followed by a limiter. By
// setting `fixed_gain_db`, the limiter can be turned into a compressor that
// first applies a fixed gain. The adaptive digital AGC can be turned off by
// setting |adaptive_digital_mode=false|.
struct RTC_EXPORT GainController2 {
bool operator==(const GainController2& rhs) const;
bool operator!=(const GainController2& rhs) const {
return !(*this == rhs);
}
bool enabled = false;
struct FixedDigital {
float gain_db = 0.0f;
} fixed_digital;
struct RTC_EXPORT AdaptiveDigital {
bool operator==(const AdaptiveDigital& rhs) const;
bool operator!=(const AdaptiveDigital& rhs) const {
return !(*this == rhs);
}
bool enabled = false;
// When true, the adaptive digital controller runs but the signal is not
// modified.
bool dry_run = false;
float headroom_db = 6.0f;
// TODO(bugs.webrtc.org/7494): Consider removing and inferring from
// `max_output_noise_level_dbfs`.
float max_gain_db = 30.0f;
float initial_gain_db = 8.0f;
int vad_reset_period_ms = 1500;
int adjacent_speech_frames_threshold = 12;
float max_gain_change_db_per_second = 3.0f;
float max_output_noise_level_dbfs = -50.0f;
} adaptive_digital;
} gain_controller2;
// TODO(bugs.webrtc.org/11539): Deprecated. Delete this flag. Replaced by
// injectable submodule.
struct ResidualEchoDetector {
bool enabled = false;
} residual_echo_detector;
std::string ToString() const;
};
// TODO(mgraczyk): Remove once all methods that use ChannelLayout are gone.
enum ChannelLayout {
kMono,
// Left, right.
kStereo,
// Mono, keyboard, and mic.
kMonoAndKeyboard,
// Left, right, keyboard, and mic.
kStereoAndKeyboard
};
// Specifies the properties of a setting to be passed to AudioProcessing at
// runtime.
class RuntimeSetting {
public:
enum class Type {
kNotSpecified,
kCapturePreGain,
kCaptureCompressionGain,
kCaptureFixedPostGain,
kPlayoutVolumeChange,
kCustomRenderProcessingRuntimeSetting,
kPlayoutAudioDeviceChange,
kCapturePostGain,
kCaptureOutputUsed
};
// Play-out audio device properties.
struct PlayoutAudioDeviceInfo {
int id; // Identifies the audio device.
int max_volume; // Maximum play-out volume.
};
RuntimeSetting() : type_(Type::kNotSpecified), value_(0.0f) {}
~RuntimeSetting() = default;
static RuntimeSetting CreateCapturePreGain(float gain) {
return {Type::kCapturePreGain, gain};
}
static RuntimeSetting CreateCapturePostGain(float gain) {
return {Type::kCapturePostGain, gain};
}
// Corresponds to Config::GainController1::compression_gain_db, but for
// runtime configuration.
static RuntimeSetting CreateCompressionGainDb(int gain_db) {
RTC_DCHECK_GE(gain_db, 0);
RTC_DCHECK_LE(gain_db, 90);
return {Type::kCaptureCompressionGain, static_cast<float>(gain_db)};
}
// Corresponds to Config::GainController2::fixed_digital::gain_db, but for
// runtime configuration.
static RuntimeSetting CreateCaptureFixedPostGain(float gain_db) {
RTC_DCHECK_GE(gain_db, 0.0f);
RTC_DCHECK_LE(gain_db, 90.0f);
return {Type::kCaptureFixedPostGain, gain_db};
}
// Creates a runtime setting to notify play-out (aka render) audio device
// changes.
static RuntimeSetting CreatePlayoutAudioDeviceChange(
PlayoutAudioDeviceInfo audio_device) {
return {Type::kPlayoutAudioDeviceChange, audio_device};
}
// Creates a runtime setting to notify play-out (aka render) volume changes.
// `volume` is the unnormalized volume, the maximum of which
static RuntimeSetting CreatePlayoutVolumeChange(int volume) {
return {Type::kPlayoutVolumeChange, volume};
}
static RuntimeSetting CreateCustomRenderSetting(float payload) {
return {Type::kCustomRenderProcessingRuntimeSetting, payload};
}
static RuntimeSetting CreateCaptureOutputUsedSetting(
bool capture_output_used) {
return {Type::kCaptureOutputUsed, capture_output_used};
}
Type type() const { return type_; }
// Getters do not return a value but instead modify the argument to protect
// from implicit casting.
void GetFloat(float* value) const {
RTC_DCHECK(value);
*value = value_.float_value;
}
void GetInt(int* value) const {
RTC_DCHECK(value);
*value = value_.int_value;
}
void GetBool(bool* value) const {
RTC_DCHECK(value);
*value = value_.bool_value;
}
void GetPlayoutAudioDeviceInfo(PlayoutAudioDeviceInfo* value) const {
RTC_DCHECK(value);
*value = value_.playout_audio_device_info;
}
private:
RuntimeSetting(Type id, float value) : type_(id), value_(value) {}
RuntimeSetting(Type id, int value) : type_(id), value_(value) {}
RuntimeSetting(Type id, PlayoutAudioDeviceInfo value)
: type_(id), value_(value) {}
Type type_;
union U {
U() {}
U(int value) : int_value(value) {}
U(float value) : float_value(value) {}
U(PlayoutAudioDeviceInfo value) : playout_audio_device_info(value) {}
float float_value;
int int_value;
bool bool_value;
PlayoutAudioDeviceInfo playout_audio_device_info;
} value_;
};
~AudioProcessing() override {}
// Initializes internal states, while retaining all user settings. This
// should be called before beginning to process a new audio stream. However,
// it is not necessary to call before processing the first stream after
// creation.
//
// It is also not necessary to call if the audio parameters (sample
// rate and number of channels) have changed. Passing updated parameters
// directly to `ProcessStream()` and `ProcessReverseStream()` is permissible.
// If the parameters are known at init-time though, they may be provided.
// TODO(webrtc:5298): Change to return void.
virtual int Initialize() = 0;
// The int16 interfaces require:
// - only `NativeRate`s be used
// - that the input, output and reverse rates must match
// - that `processing_config.output_stream()` matches
// `processing_config.input_stream()`.
//
// The float interfaces accept arbitrary rates and support differing input and
// output layouts, but the output must have either one channel or the same
// number of channels as the input.
virtual int Initialize(const ProcessingConfig& processing_config) = 0;
// TODO(peah): This method is a temporary solution used to take control
// over the parameters in the audio processing module and is likely to change.
virtual void ApplyConfig(const Config& config) = 0;
// TODO(ajm): Only intended for internal use. Make private and friend the
// necessary classes?
virtual int proc_sample_rate_hz() const = 0;
virtual int proc_split_sample_rate_hz() const = 0;
virtual size_t num_input_channels() const = 0;
virtual size_t num_proc_channels() const = 0;
virtual size_t num_output_channels() const = 0;
virtual size_t num_reverse_channels() const = 0;
// Set to true when the output of AudioProcessing will be muted or in some
// other way not used. Ideally, the captured audio would still be processed,
// but some components may change behavior based on this information.
// Default false. This method takes a lock. To achieve this in a lock-less
// manner the PostRuntimeSetting can instead be used.
virtual void set_output_will_be_muted(bool muted) = 0;
// Enqueues a runtime setting.
virtual void SetRuntimeSetting(RuntimeSetting setting) = 0;
// Enqueues a runtime setting. Returns a bool indicating whether the
// enqueueing was successfull.
virtual bool PostRuntimeSetting(RuntimeSetting setting) = 0;
// Accepts and produces a 10 ms frame interleaved 16 bit integer audio as
// specified in `input_config` and `output_config`. `src` and `dest` may use
// the same memory, if desired.
virtual int ProcessStream(const int16_t* const src,
const StreamConfig& input_config,
const StreamConfig& output_config,
int16_t* const dest) = 0;
// Accepts deinterleaved float audio with the range [-1, 1]. Each element of
// `src` points to a channel buffer, arranged according to `input_stream`. At
// output, the channels will be arranged according to `output_stream` in
// `dest`.
//
// The output must have one channel or as many channels as the input. `src`
// and `dest` may use the same memory, if desired.
virtual int ProcessStream(const float* const* src,
const StreamConfig& input_config,
const StreamConfig& output_config,
float* const* dest) = 0;
// Accepts and produces a 10 ms frame of interleaved 16 bit integer audio for
// the reverse direction audio stream as specified in `input_config` and
// `output_config`. `src` and `dest` may use the same memory, if desired.
virtual int ProcessReverseStream(const int16_t* const src,
const StreamConfig& input_config,
const StreamConfig& output_config,
int16_t* const dest) = 0;
// Accepts deinterleaved float audio with the range [-1, 1]. Each element of
// `data` points to a channel buffer, arranged according to `reverse_config`.
virtual int ProcessReverseStream(const float* const* src,
const StreamConfig& input_config,
const StreamConfig& output_config,
float* const* dest) = 0;
// Accepts deinterleaved float audio with the range [-1, 1]. Each element
// of `data` points to a channel buffer, arranged according to
// `reverse_config`.
virtual int AnalyzeReverseStream(const float* const* data,
const StreamConfig& reverse_config) = 0;
// Returns the most recently produced 10 ms of the linear AEC output at a rate
// of 16 kHz. If there is more than one capture channel, a mono representation
// of the input is returned. Returns true/false to indicate whether an output
// returned.
virtual bool GetLinearAecOutput(
rtc::ArrayView<std::array<float, 160>> linear_output) const = 0;
// This must be called prior to ProcessStream() if and only if adaptive analog
// gain control is enabled, to pass the current analog level from the audio
// HAL. Must be within the range [0, 255].
virtual void set_stream_analog_level(int level) = 0;
// When an analog mode is set, this should be called after ProcessStream()
// to obtain the recommended new analog level for the audio HAL. It is the
// user's responsibility to apply this level.
virtual int recommended_stream_analog_level() const = 0;
// This must be called if and only if echo processing is enabled.
//
// Sets the `delay` in ms between ProcessReverseStream() receiving a far-end
// frame and ProcessStream() receiving a near-end frame containing the
// corresponding echo. On the client-side this can be expressed as
// delay = (t_render - t_analyze) + (t_process - t_capture)
// where,
// - t_analyze is the time a frame is passed to ProcessReverseStream() and
// t_render is the time the first sample of the same frame is rendered by
// the audio hardware.
// - t_capture is the time the first sample of a frame is captured by the
// audio hardware and t_process is the time the same frame is passed to
// ProcessStream().
virtual int set_stream_delay_ms(int delay) = 0;
virtual int stream_delay_ms() const = 0;
// Call to signal that a key press occurred (true) or did not occur (false)
// with this chunk of audio.
virtual void set_stream_key_pressed(bool key_pressed) = 0;
// Creates and attaches an webrtc::AecDump for recording debugging
// information.
// The `worker_queue` may not be null and must outlive the created
// AecDump instance. |max_log_size_bytes == -1| means the log size
// will be unlimited. `handle` may not be null. The AecDump takes
// responsibility for `handle` and closes it in the destructor. A
// return value of true indicates that the file has been
// sucessfully opened, while a value of false indicates that
// opening the file failed.
virtual bool CreateAndAttachAecDump(const std::string& file_name,
int64_t max_log_size_bytes,
rtc::TaskQueue* worker_queue) = 0;
virtual bool CreateAndAttachAecDump(FILE* handle,
int64_t max_log_size_bytes,
rtc::TaskQueue* worker_queue) = 0;
// TODO(webrtc:5298) Deprecated variant.
// Attaches provided webrtc::AecDump for recording debugging
// information. Log file and maximum file size logic is supposed to
// be handled by implementing instance of AecDump. Calling this
// method when another AecDump is attached resets the active AecDump
// with a new one. This causes the d-tor of the earlier AecDump to
// be called. The d-tor call may block until all pending logging
// tasks are completed.
virtual void AttachAecDump(std::unique_ptr<AecDump> aec_dump) = 0;
// If no AecDump is attached, this has no effect. If an AecDump is
// attached, it's destructor is called. The d-tor may block until
// all pending logging tasks are completed.
virtual void DetachAecDump() = 0;
// Get audio processing statistics.
virtual AudioProcessingStats GetStatistics() = 0;
// TODO(webrtc:5298) Deprecated variant. The `has_remote_tracks` argument
// should be set if there are active remote tracks (this would usually be true
// during a call). If there are no remote tracks some of the stats will not be
// set by AudioProcessing, because they only make sense if there is at least
// one remote track.
virtual AudioProcessingStats GetStatistics(bool has_remote_tracks) = 0;
// Returns the last applied configuration.
virtual AudioProcessing::Config GetConfig() const = 0;
enum Error {
// Fatal errors.
kNoError = 0,
kUnspecifiedError = -1,
kCreationFailedError = -2,
kUnsupportedComponentError = -3,
kUnsupportedFunctionError = -4,
kNullPointerError = -5,
kBadParameterError = -6,
kBadSampleRateError = -7,
kBadDataLengthError = -8,
kBadNumberChannelsError = -9,
kFileError = -10,
kStreamParameterNotSetError = -11,
kNotEnabledError = -12,
// Warnings are non-fatal.
// This results when a set_stream_ parameter is out of range. Processing
// will continue, but the parameter may have been truncated.
kBadStreamParameterWarning = -13
};
// Native rates supported by the integer interfaces.
enum NativeRate {
kSampleRate8kHz = 8000,
kSampleRate16kHz = 16000,
kSampleRate32kHz = 32000,
kSampleRate48kHz = 48000
};
// TODO(kwiberg): We currently need to support a compiler (Visual C++) that
// complains if we don't explicitly state the size of the array here. Remove
// the size when that's no longer the case.
static constexpr int kNativeSampleRatesHz[4] = {
kSampleRate8kHz, kSampleRate16kHz, kSampleRate32kHz, kSampleRate48kHz};
static constexpr size_t kNumNativeSampleRates =
arraysize(kNativeSampleRatesHz);
static constexpr int kMaxNativeSampleRateHz =
kNativeSampleRatesHz[kNumNativeSampleRates - 1];
static constexpr int kChunkSizeMs = 10;
};
class RTC_EXPORT AudioProcessingBuilder {
public:
AudioProcessingBuilder();
AudioProcessingBuilder(const AudioProcessingBuilder&) = delete;
AudioProcessingBuilder& operator=(const AudioProcessingBuilder&) = delete;
~AudioProcessingBuilder();
// Sets the APM configuration.
AudioProcessingBuilder& SetConfig(const AudioProcessing::Config& config) {
config_ = config;
return *this;
}
// Sets the echo controller factory to inject when APM is created.
AudioProcessingBuilder& SetEchoControlFactory(
std::unique_ptr<EchoControlFactory> echo_control_factory) {
echo_control_factory_ = std::move(echo_control_factory);
return *this;
}
// Sets the capture post-processing sub-module to inject when APM is created.
AudioProcessingBuilder& SetCapturePostProcessing(
std::unique_ptr<CustomProcessing> capture_post_processing) {
capture_post_processing_ = std::move(capture_post_processing);
return *this;
}
// Sets the render pre-processing sub-module to inject when APM is created.
AudioProcessingBuilder& SetRenderPreProcessing(
std::unique_ptr<CustomProcessing> render_pre_processing) {
render_pre_processing_ = std::move(render_pre_processing);
return *this;
}
// Sets the echo detector to inject when APM is created.
AudioProcessingBuilder& SetEchoDetector(
rtc::scoped_refptr<EchoDetector> echo_detector) {
echo_detector_ = std::move(echo_detector);
return *this;
}
// Sets the capture analyzer sub-module to inject when APM is created.
AudioProcessingBuilder& SetCaptureAnalyzer(
std::unique_ptr<CustomAudioAnalyzer> capture_analyzer) {
capture_analyzer_ = std::move(capture_analyzer);
return *this;
}
// Creates an APM instance with the specified config or the default one if
// unspecified. Injects the specified components transferring the ownership
// to the newly created APM instance - i.e., except for the config, the
// builder is reset to its initial state.
rtc::scoped_refptr<AudioProcessing> Create();
private:
AudioProcessing::Config config_;
std::unique_ptr<EchoControlFactory> echo_control_factory_;
std::unique_ptr<CustomProcessing> capture_post_processing_;
std::unique_ptr<CustomProcessing> render_pre_processing_;
rtc::scoped_refptr<EchoDetector> echo_detector_;
std::unique_ptr<CustomAudioAnalyzer> capture_analyzer_;
};
class StreamConfig {
public:
// sample_rate_hz: The sampling rate of the stream.
// num_channels: The number of audio channels in the stream.
StreamConfig(int sample_rate_hz = 0, size_t num_channels = 0)
: sample_rate_hz_(sample_rate_hz),
num_channels_(num_channels),
num_frames_(calculate_frames(sample_rate_hz)) {}
void set_sample_rate_hz(int value) {
sample_rate_hz_ = value;
num_frames_ = calculate_frames(value);
}
void set_num_channels(size_t value) { num_channels_ = value; }
int sample_rate_hz() const { return sample_rate_hz_; }
// The number of channels in the stream.
size_t num_channels() const { return num_channels_; }
size_t num_frames() const { return num_frames_; }
size_t num_samples() const { return num_channels_ * num_frames_; }
bool operator==(const StreamConfig& other) const {
return sample_rate_hz_ == other.sample_rate_hz_ &&
num_channels_ == other.num_channels_;
}
bool operator!=(const StreamConfig& other) const { return !(*this == other); }
private:
static size_t calculate_frames(int sample_rate_hz) {
return static_cast<size_t>(AudioProcessing::kChunkSizeMs * sample_rate_hz /
1000);
}
int sample_rate_hz_;
size_t num_channels_;
size_t num_frames_;
};
class ProcessingConfig {
public:
enum StreamName {
kInputStream,
kOutputStream,
kReverseInputStream,
kReverseOutputStream,
kNumStreamNames,
};
const StreamConfig& input_stream() const {
return streams[StreamName::kInputStream];
}
const StreamConfig& output_stream() const {
return streams[StreamName::kOutputStream];
}
const StreamConfig& reverse_input_stream() const {
return streams[StreamName::kReverseInputStream];
}
const StreamConfig& reverse_output_stream() const {
return streams[StreamName::kReverseOutputStream];
}
StreamConfig& input_stream() { return streams[StreamName::kInputStream]; }
StreamConfig& output_stream() { return streams[StreamName::kOutputStream]; }
StreamConfig& reverse_input_stream() {
return streams[StreamName::kReverseInputStream];
}
StreamConfig& reverse_output_stream() {
return streams[StreamName::kReverseOutputStream];
}
bool operator==(const ProcessingConfig& other) const {
for (int i = 0; i < StreamName::kNumStreamNames; ++i) {
if (this->streams[i] != other.streams[i]) {
return false;
}
}
return true;
}
bool operator!=(const ProcessingConfig& other) const {
return !(*this == other);
}
StreamConfig streams[StreamName::kNumStreamNames];
};
// Experimental interface for a custom analysis submodule.
class CustomAudioAnalyzer {
public:
// (Re-) Initializes the submodule.
virtual void Initialize(int sample_rate_hz, int num_channels) = 0;
// Analyzes the given capture or render signal.
virtual void Analyze(const AudioBuffer* audio) = 0;
// Returns a string representation of the module state.
virtual std::string ToString() const = 0;
virtual ~CustomAudioAnalyzer() {}
};
// Interface for a custom processing submodule.
class CustomProcessing {
public:
// (Re-)Initializes the submodule.
virtual void Initialize(int sample_rate_hz, int num_channels) = 0;
// Processes the given capture or render signal.
virtual void Process(AudioBuffer* audio) = 0;
// Returns a string representation of the module state.
virtual std::string ToString() const = 0;
// Handles RuntimeSettings. TODO(webrtc:9262): make pure virtual
// after updating dependencies.
virtual void SetRuntimeSetting(AudioProcessing::RuntimeSetting setting);
virtual ~CustomProcessing() {}
};
// Interface for an echo detector submodule.
class EchoDetector : public rtc::RefCountInterface {
public:
// (Re-)Initializes the submodule.
virtual void Initialize(int capture_sample_rate_hz,
int num_capture_channels,
int render_sample_rate_hz,
int num_render_channels) = 0;
// Analysis (not changing) of the first channel of the render signal.
virtual void AnalyzeRenderAudio(rtc::ArrayView<const float> render_audio) = 0;
// Analysis (not changing) of the capture signal.
virtual void AnalyzeCaptureAudio(
rtc::ArrayView<const float> capture_audio) = 0;
struct Metrics {
absl::optional<double> echo_likelihood;
absl::optional<double> echo_likelihood_recent_max;
};
// Collect current metrics from the echo detector.
virtual Metrics GetMetrics() const = 0;
};
} // namespace webrtc
#endif // MODULES_AUDIO_PROCESSING_INCLUDE_AUDIO_PROCESSING_H_