OLD | NEW |
1 /* | 1 /* |
2 * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. | 2 * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. |
3 * | 3 * |
4 * Use of this source code is governed by a BSD-style license | 4 * Use of this source code is governed by a BSD-style license |
5 * that can be found in the LICENSE file in the root of the source | 5 * that can be found in the LICENSE file in the root of the source |
6 * tree. An additional intellectual property rights grant can be found | 6 * tree. An additional intellectual property rights grant can be found |
7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
9 */ | 9 */ |
10 | 10 |
11 #include "webrtc/modules/audio_processing/audio_buffer.h" | 11 #include "webrtc/modules/audio_processing/audio_buffer.h" |
12 | 12 |
| 13 #include "webrtc/base/checks.h" |
13 #include "webrtc/common_audio/include/audio_util.h" | 14 #include "webrtc/common_audio/include/audio_util.h" |
14 #include "webrtc/common_audio/resampler/push_sinc_resampler.h" | 15 #include "webrtc/common_audio/resampler/push_sinc_resampler.h" |
15 #include "webrtc/common_audio/signal_processing/include/signal_processing_librar
y.h" | 16 #include "webrtc/common_audio/signal_processing/include/signal_processing_librar
y.h" |
16 #include "webrtc/common_audio/channel_buffer.h" | 17 #include "webrtc/common_audio/channel_buffer.h" |
17 #include "webrtc/modules/audio_processing/common.h" | 18 #include "webrtc/modules/audio_processing/common.h" |
18 | 19 |
19 namespace webrtc { | 20 namespace webrtc { |
20 namespace { | 21 namespace { |
21 | 22 |
22 const size_t kSamplesPer16kHzChannel = 160; | 23 const size_t kSamplesPer16kHzChannel = 160; |
23 const size_t kSamplesPer32kHzChannel = 320; | 24 const size_t kSamplesPer32kHzChannel = 320; |
24 const size_t kSamplesPer48kHzChannel = 480; | 25 const size_t kSamplesPer48kHzChannel = 480; |
25 | 26 |
26 int KeyboardChannelIndex(const StreamConfig& stream_config) { | 27 int KeyboardChannelIndex(const StreamConfig& stream_config) { |
27 if (!stream_config.has_keyboard()) { | 28 if (!stream_config.has_keyboard()) { |
28 assert(false); | 29 RTC_NOTREACHED(); |
29 return 0; | 30 return 0; |
30 } | 31 } |
31 | 32 |
32 return stream_config.num_channels(); | 33 return stream_config.num_channels(); |
33 } | 34 } |
34 | 35 |
35 size_t NumBandsFromSamplesPerChannel(size_t num_frames) { | 36 size_t NumBandsFromSamplesPerChannel(size_t num_frames) { |
36 size_t num_bands = 1; | 37 size_t num_bands = 1; |
37 if (num_frames == kSamplesPer32kHzChannel || | 38 if (num_frames == kSamplesPer32kHzChannel || |
38 num_frames == kSamplesPer48kHzChannel) { | 39 num_frames == kSamplesPer48kHzChannel) { |
(...skipping 15 matching lines...) Expand all Loading... |
54 num_proc_channels_(num_process_channels), | 55 num_proc_channels_(num_process_channels), |
55 output_num_frames_(output_num_frames), | 56 output_num_frames_(output_num_frames), |
56 num_channels_(num_process_channels), | 57 num_channels_(num_process_channels), |
57 num_bands_(NumBandsFromSamplesPerChannel(proc_num_frames_)), | 58 num_bands_(NumBandsFromSamplesPerChannel(proc_num_frames_)), |
58 num_split_frames_(rtc::CheckedDivExact(proc_num_frames_, num_bands_)), | 59 num_split_frames_(rtc::CheckedDivExact(proc_num_frames_, num_bands_)), |
59 mixed_low_pass_valid_(false), | 60 mixed_low_pass_valid_(false), |
60 reference_copied_(false), | 61 reference_copied_(false), |
61 activity_(AudioFrame::kVadUnknown), | 62 activity_(AudioFrame::kVadUnknown), |
62 keyboard_data_(NULL), | 63 keyboard_data_(NULL), |
63 data_(new IFChannelBuffer(proc_num_frames_, num_proc_channels_)) { | 64 data_(new IFChannelBuffer(proc_num_frames_, num_proc_channels_)) { |
64 assert(input_num_frames_ > 0); | 65 RTC_DCHECK_GT(input_num_frames_, 0u); |
65 assert(proc_num_frames_ > 0); | 66 RTC_DCHECK_GT(proc_num_frames_, 0u); |
66 assert(output_num_frames_ > 0); | 67 RTC_DCHECK_GT(output_num_frames_, 0u); |
67 assert(num_input_channels_ > 0); | 68 RTC_DCHECK_GT(num_input_channels_, 0u); |
68 assert(num_proc_channels_ > 0 && num_proc_channels_ <= num_input_channels_); | 69 RTC_DCHECK_GT(num_proc_channels_, 0u); |
| 70 RTC_DCHECK_LE(num_proc_channels_, num_input_channels_); |
69 | 71 |
70 if (input_num_frames_ != proc_num_frames_ || | 72 if (input_num_frames_ != proc_num_frames_ || |
71 output_num_frames_ != proc_num_frames_) { | 73 output_num_frames_ != proc_num_frames_) { |
72 // Create an intermediate buffer for resampling. | 74 // Create an intermediate buffer for resampling. |
73 process_buffer_.reset(new ChannelBuffer<float>(proc_num_frames_, | 75 process_buffer_.reset(new ChannelBuffer<float>(proc_num_frames_, |
74 num_proc_channels_)); | 76 num_proc_channels_)); |
75 | 77 |
76 if (input_num_frames_ != proc_num_frames_) { | 78 if (input_num_frames_ != proc_num_frames_) { |
77 for (size_t i = 0; i < num_proc_channels_; ++i) { | 79 for (size_t i = 0; i < num_proc_channels_; ++i) { |
78 input_resamplers_.push_back(std::unique_ptr<PushSincResampler>( | 80 input_resamplers_.push_back(std::unique_ptr<PushSincResampler>( |
(...skipping 16 matching lines...) Expand all Loading... |
95 splitting_filter_.reset(new SplittingFilter(num_proc_channels_, | 97 splitting_filter_.reset(new SplittingFilter(num_proc_channels_, |
96 num_bands_, | 98 num_bands_, |
97 proc_num_frames_)); | 99 proc_num_frames_)); |
98 } | 100 } |
99 } | 101 } |
100 | 102 |
101 AudioBuffer::~AudioBuffer() {} | 103 AudioBuffer::~AudioBuffer() {} |
102 | 104 |
103 void AudioBuffer::CopyFrom(const float* const* data, | 105 void AudioBuffer::CopyFrom(const float* const* data, |
104 const StreamConfig& stream_config) { | 106 const StreamConfig& stream_config) { |
105 assert(stream_config.num_frames() == input_num_frames_); | 107 RTC_DCHECK_EQ(stream_config.num_frames(), input_num_frames_); |
106 assert(stream_config.num_channels() == num_input_channels_); | 108 RTC_DCHECK_EQ(stream_config.num_channels(), num_input_channels_); |
107 InitForNewData(); | 109 InitForNewData(); |
108 // Initialized lazily because there's a different condition in | 110 // Initialized lazily because there's a different condition in |
109 // DeinterleaveFrom. | 111 // DeinterleaveFrom. |
110 const bool need_to_downmix = | 112 const bool need_to_downmix = |
111 num_input_channels_ > 1 && num_proc_channels_ == 1; | 113 num_input_channels_ > 1 && num_proc_channels_ == 1; |
112 if (need_to_downmix && !input_buffer_) { | 114 if (need_to_downmix && !input_buffer_) { |
113 input_buffer_.reset( | 115 input_buffer_.reset( |
114 new IFChannelBuffer(input_num_frames_, num_proc_channels_)); | 116 new IFChannelBuffer(input_num_frames_, num_proc_channels_)); |
115 } | 117 } |
116 | 118 |
(...skipping 23 matching lines...) Expand all Loading... |
140 // Convert to the S16 range. | 142 // Convert to the S16 range. |
141 for (size_t i = 0; i < num_proc_channels_; ++i) { | 143 for (size_t i = 0; i < num_proc_channels_; ++i) { |
142 FloatToFloatS16(data_ptr[i], | 144 FloatToFloatS16(data_ptr[i], |
143 proc_num_frames_, | 145 proc_num_frames_, |
144 data_->fbuf()->channels()[i]); | 146 data_->fbuf()->channels()[i]); |
145 } | 147 } |
146 } | 148 } |
147 | 149 |
148 void AudioBuffer::CopyTo(const StreamConfig& stream_config, | 150 void AudioBuffer::CopyTo(const StreamConfig& stream_config, |
149 float* const* data) { | 151 float* const* data) { |
150 assert(stream_config.num_frames() == output_num_frames_); | 152 RTC_DCHECK_EQ(stream_config.num_frames(), output_num_frames_); |
151 assert(stream_config.num_channels() == num_channels_ || num_channels_ == 1); | 153 RTC_DCHECK(stream_config.num_channels() == num_channels_ || |
| 154 num_channels_ == 1); |
152 | 155 |
153 // Convert to the float range. | 156 // Convert to the float range. |
154 float* const* data_ptr = data; | 157 float* const* data_ptr = data; |
155 if (output_num_frames_ != proc_num_frames_) { | 158 if (output_num_frames_ != proc_num_frames_) { |
156 // Convert to an intermediate buffer for subsequent resampling. | 159 // Convert to an intermediate buffer for subsequent resampling. |
157 data_ptr = process_buffer_->channels(); | 160 data_ptr = process_buffer_->channels(); |
158 } | 161 } |
159 for (size_t i = 0; i < num_channels_; ++i) { | 162 for (size_t i = 0; i < num_channels_; ++i) { |
160 FloatS16ToFloat(data_->fbuf()->channels()[i], | 163 FloatS16ToFloat(data_->fbuf()->channels()[i], |
161 proc_num_frames_, | 164 proc_num_frames_, |
(...skipping 205 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
367 // We don't resample the keyboard channel. | 370 // We don't resample the keyboard channel. |
368 return input_num_frames_; | 371 return input_num_frames_; |
369 } | 372 } |
370 | 373 |
371 size_t AudioBuffer::num_bands() const { | 374 size_t AudioBuffer::num_bands() const { |
372 return num_bands_; | 375 return num_bands_; |
373 } | 376 } |
374 | 377 |
375 // The resampler is only for supporting 48kHz to 16kHz in the reverse stream. | 378 // The resampler is only for supporting 48kHz to 16kHz in the reverse stream. |
376 void AudioBuffer::DeinterleaveFrom(AudioFrame* frame) { | 379 void AudioBuffer::DeinterleaveFrom(AudioFrame* frame) { |
377 assert(frame->num_channels_ == num_input_channels_); | 380 RTC_DCHECK_EQ(frame->num_channels_, num_input_channels_); |
378 assert(frame->samples_per_channel_ == input_num_frames_); | 381 RTC_DCHECK_EQ(frame->samples_per_channel_, input_num_frames_); |
379 InitForNewData(); | 382 InitForNewData(); |
380 // Initialized lazily because there's a different condition in CopyFrom. | 383 // Initialized lazily because there's a different condition in CopyFrom. |
381 if ((input_num_frames_ != proc_num_frames_) && !input_buffer_) { | 384 if ((input_num_frames_ != proc_num_frames_) && !input_buffer_) { |
382 input_buffer_.reset( | 385 input_buffer_.reset( |
383 new IFChannelBuffer(input_num_frames_, num_proc_channels_)); | 386 new IFChannelBuffer(input_num_frames_, num_proc_channels_)); |
384 } | 387 } |
385 activity_ = frame->vad_activity_; | 388 activity_ = frame->vad_activity_; |
386 | 389 |
387 int16_t* const* deinterleaved; | 390 int16_t* const* deinterleaved; |
388 if (input_num_frames_ == proc_num_frames_) { | 391 if (input_num_frames_ == proc_num_frames_) { |
389 deinterleaved = data_->ibuf()->channels(); | 392 deinterleaved = data_->ibuf()->channels(); |
390 } else { | 393 } else { |
391 deinterleaved = input_buffer_->ibuf()->channels(); | 394 deinterleaved = input_buffer_->ibuf()->channels(); |
392 } | 395 } |
393 if (num_proc_channels_ == 1) { | 396 if (num_proc_channels_ == 1) { |
394 // Downmix and deinterleave simultaneously. | 397 // Downmix and deinterleave simultaneously. |
395 DownmixInterleavedToMono(frame->data_, input_num_frames_, | 398 DownmixInterleavedToMono(frame->data_, input_num_frames_, |
396 num_input_channels_, deinterleaved[0]); | 399 num_input_channels_, deinterleaved[0]); |
397 } else { | 400 } else { |
398 assert(num_proc_channels_ == num_input_channels_); | 401 RTC_DCHECK_EQ(num_proc_channels_, num_input_channels_); |
399 Deinterleave(frame->data_, | 402 Deinterleave(frame->data_, |
400 input_num_frames_, | 403 input_num_frames_, |
401 num_proc_channels_, | 404 num_proc_channels_, |
402 deinterleaved); | 405 deinterleaved); |
403 } | 406 } |
404 | 407 |
405 // Resample. | 408 // Resample. |
406 if (input_num_frames_ != proc_num_frames_) { | 409 if (input_num_frames_ != proc_num_frames_) { |
407 for (size_t i = 0; i < num_proc_channels_; ++i) { | 410 for (size_t i = 0; i < num_proc_channels_; ++i) { |
408 input_resamplers_[i]->Resample(input_buffer_->fbuf_const()->channels()[i], | 411 input_resamplers_[i]->Resample(input_buffer_->fbuf_const()->channels()[i], |
409 input_num_frames_, | 412 input_num_frames_, |
410 data_->fbuf()->channels()[i], | 413 data_->fbuf()->channels()[i], |
411 proc_num_frames_); | 414 proc_num_frames_); |
412 } | 415 } |
413 } | 416 } |
414 } | 417 } |
415 | 418 |
416 void AudioBuffer::InterleaveTo(AudioFrame* frame, bool data_changed) { | 419 void AudioBuffer::InterleaveTo(AudioFrame* frame, bool data_changed) { |
417 frame->vad_activity_ = activity_; | 420 frame->vad_activity_ = activity_; |
418 if (!data_changed) { | 421 if (!data_changed) { |
419 return; | 422 return; |
420 } | 423 } |
421 | 424 |
422 assert(frame->num_channels_ == num_channels_ || num_channels_ == 1); | 425 RTC_DCHECK(frame->num_channels_ == num_channels_ || num_channels_ == 1); |
423 assert(frame->samples_per_channel_ == output_num_frames_); | 426 RTC_DCHECK_EQ(frame->samples_per_channel_, output_num_frames_); |
424 | 427 |
425 // Resample if necessary. | 428 // Resample if necessary. |
426 IFChannelBuffer* data_ptr = data_.get(); | 429 IFChannelBuffer* data_ptr = data_.get(); |
427 if (proc_num_frames_ != output_num_frames_) { | 430 if (proc_num_frames_ != output_num_frames_) { |
428 if (!output_buffer_) { | 431 if (!output_buffer_) { |
429 output_buffer_.reset( | 432 output_buffer_.reset( |
430 new IFChannelBuffer(output_num_frames_, num_channels_)); | 433 new IFChannelBuffer(output_num_frames_, num_channels_)); |
431 } | 434 } |
432 for (size_t i = 0; i < num_channels_; ++i) { | 435 for (size_t i = 0; i < num_channels_; ++i) { |
433 output_resamplers_[i]->Resample( | 436 output_resamplers_[i]->Resample( |
(...skipping 30 matching lines...) Expand all Loading... |
464 | 467 |
465 void AudioBuffer::SplitIntoFrequencyBands() { | 468 void AudioBuffer::SplitIntoFrequencyBands() { |
466 splitting_filter_->Analysis(data_.get(), split_data_.get()); | 469 splitting_filter_->Analysis(data_.get(), split_data_.get()); |
467 } | 470 } |
468 | 471 |
469 void AudioBuffer::MergeFrequencyBands() { | 472 void AudioBuffer::MergeFrequencyBands() { |
470 splitting_filter_->Synthesis(split_data_.get(), data_.get()); | 473 splitting_filter_->Synthesis(split_data_.get(), data_.get()); |
471 } | 474 } |
472 | 475 |
473 } // namespace webrtc | 476 } // namespace webrtc |
OLD | NEW |