| /* |
| * Copyright (c) 2025 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. |
| */ |
| #include "modules/audio_processing/capture_mixer/channel_content_remixer.h" |
| |
| #include <algorithm> |
| #include <cstddef> |
| #include <span> |
| |
| #include "rtc_base/checks.h" |
| |
| namespace webrtc { |
| |
| ChannelContentRemixer::ChannelContentRemixer(size_t num_samples_per_channel, |
| size_t num_frames_for_crossfade) |
| : num_samples_per_channel_(num_samples_per_channel), |
| num_samples_for_crossfade_(num_samples_per_channel * |
| num_frames_for_crossfade), |
| one_by_num_samples_for_crossfade_(1.0f / num_samples_for_crossfade_), |
| mixing_from_(StereoMixingVariant::kUseAverage), |
| mixing_to_(StereoMixingVariant::kUseAverage) {} |
| |
| bool ChannelContentRemixer::Mix(size_t num_output_channels, |
| StereoMixingVariant mixing_variant, |
| std::span<float> channel0, |
| std::span<float> channel1) { |
| RTC_DCHECK_EQ(channel0.size(), num_samples_per_channel_); |
| RTC_DCHECK_EQ(channel1.size(), num_samples_per_channel_); |
| |
| // Only allow a new target mixing, and a new target number of output channels, |
| // if the previous crossfade was completed. |
| if (IsCrossfadeCompleted()) { |
| mixing_from_ = mixing_to_; |
| mixing_to_ = mixing_variant; |
| num_output_channels_ = num_output_channels; |
| } |
| |
| switch (mixing_to_) { |
| case StereoMixingVariant::kUseBothChannels: { |
| switch (mixing_from_) { |
| case StereoMixingVariant::kUseBothChannels: { |
| // Mixing from: kUseBothChannels. |
| // Mixing to: kUseBothChannels. |
| |
| // No remixing needed. |
| break; |
| } |
| case StereoMixingVariant::kUseChannel0: { |
| // Mixing from: kUseChannel0. |
| // Mixing to: kUseBothChannels. |
| |
| if (num_output_channels_ == 2) { |
| // Crossfade channel 1 from using content in channel 0 to using |
| // content in channel 1. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel0, |
| /*crossfade_to=*/channel1, |
| /*destination=*/channel1, crossfade_sample_counter_); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel1: { |
| // Mixing from: kUseChannel1. |
| // Mixing to: kUseBothChannels. |
| |
| // Crossfade channel 0 from using content in channel 1 to using |
| // content in channel 0. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel1, |
| /*crossfade_to=*/channel0, |
| /*destination=*/channel0, crossfade_sample_counter_); |
| |
| break; |
| } |
| case StereoMixingVariant::kUseAverage: { |
| // Mixing from: kUseAverage. |
| // Mixing to: kUseBothChannels. |
| |
| if (num_output_channels_ == 1) { |
| // Crossfade channel0 from using the channel average to using |
| // its original content. |
| CrossFadeFromAverageInToChannel0(/*crossfade_to=*/channel0, |
| channel0, channel1, |
| crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels from using the channel average to using |
| // their original content. |
| CrossFadeFromAverageToBothChannels(channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel0: { |
| switch (mixing_from_) { |
| case StereoMixingVariant::kUseBothChannels: { |
| // Mixing from: kUseBothChannels. |
| // Mixing to: kUseChannel0. |
| |
| if (num_output_channels_ == 2) { |
| // Crossfade channel 1 from using content in channel 1 to using |
| // content in channel 0. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel1, |
| /*crossfade_to=*/channel0, |
| /*destination=*/channel1, crossfade_sample_counter_); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel0: { |
| // Mixing from: kUseChannel0. |
| // Mixing to: kUseChannel0. |
| |
| if (num_output_channels_ == 2) { |
| // Copy content of channel 0 into channel 1, no cross-fading needed. |
| CopyChannelContent(/*source=*/channel0, /*destination=*/channel1); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel1: { |
| // Mixing from: kUseChannel1. |
| // Mixing to: kUseChannel0. |
| |
| // Crossfade channel 0 from using content in channel 1 to using |
| // content in channel 0. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel1, |
| /*crossfade_to=*/channel0, |
| /*destination=*/channel0, crossfade_sample_counter_); |
| if (num_output_channels_ == 2) { |
| // Crossfade channel 1 from using content in channel 1 to using |
| // content in channel 0. |
| CopyChannelContent(/*source=*/channel0, |
| /*destination=*/channel1); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseAverage: { |
| // Mixing from: kUseAverage. |
| // Mixing to: kUseChannel0. |
| |
| if (num_output_channels_ == 1) { |
| // Crossfade both channels from using the channel average to using |
| // the content in channel 0. |
| CrossFadeFromAverageInToChannel0(/*crossfade_to=*/channel0, |
| channel0, channel1, |
| crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels from using the channel average to using |
| // the content in channel 0. |
| CrossFadeFromAverageToSingleChannelContent( |
| /*crossfade_to=*/channel0, channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel1: { |
| switch (mixing_from_) { |
| case StereoMixingVariant::kUseBothChannels: { |
| // Mixing from: kUseBothChannels. |
| // Mixing to: kUseChannel1. |
| |
| // Crossfade channel 0 from using content in channel 1 to using |
| // content in channel 1. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel0, |
| /*crossfade_to=*/channel1, |
| /*destination=*/channel0, crossfade_sample_counter_); |
| break; |
| } |
| case StereoMixingVariant::kUseChannel0: { |
| // Mixing from: kUseChannel0. |
| // Mixing to: kUseChannel1. |
| |
| // Crossfade channel 0 from using content in channel 0 to using |
| // content in channel 1. |
| CrossFadeFromSingleChannelToSingleChannel( |
| /*crossfade_from=*/channel0, |
| /*crossfade_to=*/channel1, |
| /*destination=*/channel0, crossfade_sample_counter_); |
| if (num_output_channels_ == 2) { |
| // Crossfade channel 1 from using content in channel 0 to using |
| // content in channel 1. |
| CopyChannelContent(/*source=*/channel0, |
| /*destination=*/channel1); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel1: { |
| // Mixing from: kUseChannel1. |
| // Mixing to: kUseChannel1. |
| |
| // Copy content of channel 1 into channel 0, no cross-fading needed. |
| CopyChannelContent(/*source=*/channel1, |
| /*destination=*/channel0); |
| break; |
| } |
| case StereoMixingVariant::kUseAverage: { |
| // Mixing from: kUseAverage. |
| // Mixing to: kUseChannel1. |
| |
| if (num_output_channels_ == 1) { |
| // Crossfade channel 0 from using the channel average to using the |
| // content in channel 1. |
| CrossFadeFromAverageInToChannel0(/*crossfade_to=*/channel1, |
| channel0, channel1, |
| crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels from using the channel average to using |
| // the content in channel 1. |
| CrossFadeFromAverageToSingleChannelContent( |
| /*crossfade_to=*/channel1, channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseAverage: { |
| switch (mixing_from_) { |
| case StereoMixingVariant::kUseBothChannels: { |
| // Mixing from: kUseBothChannels. |
| // Mixing to: kUseAverage. |
| if (num_output_channels_ == 1) { |
| // Crossfade channel 0 to using the channel average. |
| CrossFadeChannel0ToAverage(/*crossfade_from=*/channel0, channel0, |
| channel1, crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels to using the channel average. |
| CrossFadeFromBothChannelsToAverage(channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel0: { |
| // Mixing from: kUseChannel0. |
| // Mixing to: kUseAverage. |
| if (num_output_channels_ == 1) { |
| // Crossfade channel 0 to using the channel average. |
| CrossFadeChannel0ToAverage(/*crossfade_from=*/channel0, channel0, |
| channel1, crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels to using the content from channel 0 to |
| // using the channel average. |
| CrossFadeFromSingleChannelContentToAverage( |
| /*crossfade_from=*/channel0, channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseChannel1: { |
| // Mixing from: kUseChannel1. |
| // Mixing to: kUseAverage. |
| if (num_output_channels_ == 1) { |
| // Crossfade channel 0 to using the channel average. |
| CrossFadeChannel0ToAverage(/*crossfade_from=*/channel1, channel0, |
| channel1, crossfade_sample_counter_); |
| } else { |
| // Crossfade both channels to using the content from channel 1 to |
| // using the channel average. |
| CrossFadeFromSingleChannelContentToAverage( |
| /*crossfade_from=*/channel1, channel0, channel1, |
| crossfade_sample_counter_); |
| } |
| break; |
| } |
| case StereoMixingVariant::kUseAverage: { |
| // Mixing from: kUseAverage. |
| // Mixing to: kUseAverage. |
| if (num_output_channels_ == 1) { |
| // Simply store the average into channel 0, no crossfade needed. |
| StoreChannelAverageIntoChannel0(channel0, channel1); |
| } else { |
| // Simply store the average into both channels, no crossfade needed. |
| StoreChannelAverageIntoBothChannels(channel0, channel1); |
| } |
| break; |
| } |
| } |
| break; |
| } |
| } |
| |
| return IsCrossfadeCompleted(); |
| } |
| |
| bool ChannelContentRemixer::IsCrossfadeCompleted() { |
| if (crossfade_sample_counter_ == num_samples_for_crossfade_) { |
| crossfade_sample_counter_ = 0; |
| } |
| return crossfade_sample_counter_ == 0; |
| } |
| |
| void ChannelContentRemixer::CopyChannelContent( |
| std::span<const float> source, |
| std::span<float> destination) const { |
| std::copy(source.begin(), source.end(), destination.begin()); |
| } |
| |
| void ChannelContentRemixer::StoreChannelAverageIntoBothChannels( |
| std::span<float> channel0, |
| std::span<float> channel1) const { |
| for (size_t k = 0; k < channel0.size(); ++k) { |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = average; |
| channel1[k] = average; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromSingleChannelToSingleChannel( |
| std::span<const float> crossfade_from, |
| std::span<const float> crossfade_to, |
| std::span<float> destination, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < destination.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| destination[k] = |
| (1.0f - scaling) * crossfade_from[k] + scaling * crossfade_to[k]; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromSingleChannelContentToAverage( |
| std::span<const float> crossfade_from, |
| std::span<float> channel0, |
| std::span<float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| float sample = (1.0f - scaling) * crossfade_from[k] + scaling * average; |
| |
| channel0[k] = sample; |
| channel1[k] = sample; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromAverageToSingleChannelContent( |
| std::span<const float> crossfade_to, |
| std::span<float> channel0, |
| std::span<float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| float sample = (1.0f - scaling) * average + scaling * crossfade_to[k]; |
| channel0[k] = sample; |
| channel1[k] = sample; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromAverageToBothChannels( |
| std::span<float> channel0, |
| std::span<float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float scaled_average = |
| (1.0f - scaling) * (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = scaled_average + scaling * channel0[k]; |
| channel1[k] = scaled_average + scaling * channel1[k]; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromBothChannelsToAverage( |
| std::span<float> channel0, |
| std::span<float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float scaled_average = scaling * (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = (1.0f - scaling) * channel0[k] + scaled_average; |
| channel1[k] = (1.0f - scaling) * channel1[k] + scaled_average; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeFromAverageInToChannel0( |
| std::span<const float> crossfade_to, |
| std::span<float> channel0, |
| std::span<float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = (1.0f - scaling) * average + scaling * crossfade_to[k]; |
| } |
| } |
| |
| void ChannelContentRemixer::CrossFadeChannel0ToAverage( |
| std::span<const float> crossfade_from, |
| std::span<float> channel0, |
| std::span<const float> channel1, |
| size_t& crossfade_sample_counter) const { |
| for (size_t k = 0; k < channel0.size(); ++k, ++crossfade_sample_counter) { |
| const float scaling = |
| crossfade_sample_counter * one_by_num_samples_for_crossfade_; |
| |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = (1.0f - scaling) * crossfade_from[k] + scaling * average; |
| } |
| } |
| |
| void ChannelContentRemixer::StoreChannelAverageIntoChannel0( |
| std::span<float> channel0, |
| std::span<const float> channel1) const { |
| for (size_t k = 0; k < channel0.size(); ++k) { |
| float average = (channel0[k] + channel1[k]) * 0.5f; |
| channel0[k] = average; |
| } |
| } |
| |
| } // namespace webrtc |