1 /*
2 * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved.
3 *
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
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
10
11 #include "webrtc/voice_engine/utility.h"
12
13 #include "webrtc/common_audio/resampler/include/push_resampler.h"
14 #include "webrtc/common_audio/signal_processing/include/signal_processing_library.h"
15 #include "webrtc/common_types.h"
16 #include "webrtc/modules/interface/module_common_types.h"
17 #include "webrtc/modules/utility/interface/audio_frame_operations.h"
18 #include "webrtc/system_wrappers/interface/logging.h"
19 #include "webrtc/voice_engine/voice_engine_defines.h"
20
21 namespace webrtc {
22 namespace voe {
23
24 // TODO(ajm): There is significant overlap between RemixAndResample and
25 // ConvertToCodecFormat, but if we're to consolidate we should probably make a
26 // real converter class.
RemixAndResample(const AudioFrame & src_frame,PushResampler<int16_t> * resampler,AudioFrame * dst_frame)27 void RemixAndResample(const AudioFrame& src_frame,
28 PushResampler<int16_t>* resampler,
29 AudioFrame* dst_frame) {
30 const int16_t* audio_ptr = src_frame.data_;
31 int audio_ptr_num_channels = src_frame.num_channels_;
32 int16_t mono_audio[AudioFrame::kMaxDataSizeSamples];
33
34 // Downmix before resampling.
35 if (src_frame.num_channels_ == 2 && dst_frame->num_channels_ == 1) {
36 AudioFrameOperations::StereoToMono(src_frame.data_,
37 src_frame.samples_per_channel_,
38 mono_audio);
39 audio_ptr = mono_audio;
40 audio_ptr_num_channels = 1;
41 }
42
43 if (resampler->InitializeIfNeeded(src_frame.sample_rate_hz_,
44 dst_frame->sample_rate_hz_,
45 audio_ptr_num_channels) == -1) {
46 LOG_FERR3(LS_ERROR, InitializeIfNeeded, src_frame.sample_rate_hz_,
47 dst_frame->sample_rate_hz_, audio_ptr_num_channels);
48 assert(false);
49 }
50
51 const int src_length = src_frame.samples_per_channel_ *
52 audio_ptr_num_channels;
53 int out_length = resampler->Resample(audio_ptr, src_length, dst_frame->data_,
54 AudioFrame::kMaxDataSizeSamples);
55 if (out_length == -1) {
56 LOG_FERR3(LS_ERROR, Resample, audio_ptr, src_length, dst_frame->data_);
57 assert(false);
58 }
59 dst_frame->samples_per_channel_ = out_length / audio_ptr_num_channels;
60
61 // Upmix after resampling.
62 if (src_frame.num_channels_ == 1 && dst_frame->num_channels_ == 2) {
63 // The audio in dst_frame really is mono at this point; MonoToStereo will
64 // set this back to stereo.
65 dst_frame->num_channels_ = 1;
66 AudioFrameOperations::MonoToStereo(dst_frame);
67 }
68
69 dst_frame->timestamp_ = src_frame.timestamp_;
70 dst_frame->elapsed_time_ms_ = src_frame.elapsed_time_ms_;
71 dst_frame->ntp_time_ms_ = src_frame.ntp_time_ms_;
72 }
73
DownConvertToCodecFormat(const int16_t * src_data,int samples_per_channel,int num_channels,int sample_rate_hz,int codec_num_channels,int codec_rate_hz,int16_t * mono_buffer,PushResampler<int16_t> * resampler,AudioFrame * dst_af)74 void DownConvertToCodecFormat(const int16_t* src_data,
75 int samples_per_channel,
76 int num_channels,
77 int sample_rate_hz,
78 int codec_num_channels,
79 int codec_rate_hz,
80 int16_t* mono_buffer,
81 PushResampler<int16_t>* resampler,
82 AudioFrame* dst_af) {
83 assert(samples_per_channel <= kMaxMonoDataSizeSamples);
84 assert(num_channels == 1 || num_channels == 2);
85 assert(codec_num_channels == 1 || codec_num_channels == 2);
86 dst_af->Reset();
87
88 // Never upsample the capture signal here. This should be done at the
89 // end of the send chain.
90 int destination_rate = std::min(codec_rate_hz, sample_rate_hz);
91
92 // If no stereo codecs are in use, we downmix a stereo stream from the
93 // device early in the chain, before resampling.
94 if (num_channels == 2 && codec_num_channels == 1) {
95 AudioFrameOperations::StereoToMono(src_data, samples_per_channel,
96 mono_buffer);
97 src_data = mono_buffer;
98 num_channels = 1;
99 }
100
101 if (resampler->InitializeIfNeeded(
102 sample_rate_hz, destination_rate, num_channels) != 0) {
103 LOG_FERR3(LS_ERROR,
104 InitializeIfNeeded,
105 sample_rate_hz,
106 destination_rate,
107 num_channels);
108 assert(false);
109 }
110
111 const int in_length = samples_per_channel * num_channels;
112 int out_length = resampler->Resample(
113 src_data, in_length, dst_af->data_, AudioFrame::kMaxDataSizeSamples);
114 if (out_length == -1) {
115 LOG_FERR3(LS_ERROR, Resample, src_data, in_length, dst_af->data_);
116 assert(false);
117 }
118
119 dst_af->samples_per_channel_ = out_length / num_channels;
120 dst_af->sample_rate_hz_ = destination_rate;
121 dst_af->num_channels_ = num_channels;
122 }
123
MixWithSat(int16_t target[],int target_channel,const int16_t source[],int source_channel,int source_len)124 void MixWithSat(int16_t target[],
125 int target_channel,
126 const int16_t source[],
127 int source_channel,
128 int source_len) {
129 assert(target_channel == 1 || target_channel == 2);
130 assert(source_channel == 1 || source_channel == 2);
131
132 if (target_channel == 2 && source_channel == 1) {
133 // Convert source from mono to stereo.
134 int32_t left = 0;
135 int32_t right = 0;
136 for (int i = 0; i < source_len; ++i) {
137 left = source[i] + target[i * 2];
138 right = source[i] + target[i * 2 + 1];
139 target[i * 2] = WebRtcSpl_SatW32ToW16(left);
140 target[i * 2 + 1] = WebRtcSpl_SatW32ToW16(right);
141 }
142 } else if (target_channel == 1 && source_channel == 2) {
143 // Convert source from stereo to mono.
144 int32_t temp = 0;
145 for (int i = 0; i < source_len / 2; ++i) {
146 temp = ((source[i * 2] + source[i * 2 + 1]) >> 1) + target[i];
147 target[i] = WebRtcSpl_SatW32ToW16(temp);
148 }
149 } else {
150 int32_t temp = 0;
151 for (int i = 0; i < source_len; ++i) {
152 temp = source[i] + target[i];
153 target[i] = WebRtcSpl_SatW32ToW16(temp);
154 }
155 }
156 }
157
158 } // namespace voe
159 } // namespace webrtc
160