summaryrefslogtreecommitdiffstats
path: root/media/audio/audio_util.cc
blob: 4551f5749d4d9e25cc3cfe2c050d0d2c71ef4489 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
// Copyright (c) 2012 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

// Software adjust volume of samples, allows each audio stream its own
// volume without impacting master volume for chrome and other applications.

// Implemented as templates to allow 8, 16 and 32 bit implementations.
// 8 bit is unsigned and biased by 128.

// TODO(vrk): This file has been running pretty wild and free, and it's likely
// that a lot of the functions can be simplified and made more elegant. Revisit
// after other audio cleanup is done. (crbug.com/120319)

#include "media/audio/audio_util.h"

#include <algorithm>
#include <limits>

#include "base/basictypes.h"
#include "base/command_line.h"
#include "base/logging.h"
#include "base/string_number_conversions.h"
#include "base/time.h"
#include "media/audio/audio_parameters.h"
#include "media/base/audio_bus.h"
#include "media/base/media_switches.h"

#if defined(OS_MACOSX)
#include "media/audio/mac/audio_low_latency_input_mac.h"
#include "media/audio/mac/audio_low_latency_output_mac.h"
#elif defined(OS_WIN)
#include "base/win/windows_version.h"
#include "media/audio/audio_manager_base.h"
#include "media/audio/win/audio_low_latency_input_win.h"
#include "media/audio/win/audio_low_latency_output_win.h"
#include "media/audio/win/core_audio_util_win.h"
#include "media/base/limits.h"
#endif

namespace media {

// Returns user buffer size as specified on the command line or 0 if no buffer
// size has been specified.
static int GetUserBufferSize() {
  const CommandLine* cmd_line = CommandLine::ForCurrentProcess();
  int buffer_size = 0;
  std::string buffer_size_str(cmd_line->GetSwitchValueASCII(
      switches::kAudioBufferSize));
  if (base::StringToInt(buffer_size_str, &buffer_size) && buffer_size > 0) {
    return buffer_size;
  }

  return 0;
}

// TODO(fbarchard): Convert to intrinsics for better efficiency.
template<class Fixed>
static int ScaleChannel(int channel, int volume) {
  return static_cast<int>((static_cast<Fixed>(channel) * volume) >> 16);
}

template<class Format, class Fixed, int bias>
static void AdjustVolume(Format* buf_out,
                         int sample_count,
                         int fixed_volume) {
  for (int i = 0; i < sample_count; ++i) {
    buf_out[i] = static_cast<Format>(ScaleChannel<Fixed>(buf_out[i] - bias,
                                                         fixed_volume) + bias);
  }
}

static const int kChannel_L = 0;
static const int kChannel_R = 1;
static const int kChannel_C = 2;

template<class Fixed, int min_value, int max_value>
static int AddSaturated(int val, int adder) {
  Fixed sum = static_cast<Fixed>(val) + static_cast<Fixed>(adder);
  if (sum > max_value)
    return max_value;
  if (sum < min_value)
    return min_value;
  return static_cast<int>(sum);
}

// AdjustVolume() does an in place audio sample change.
bool AdjustVolume(void* buf,
                  size_t buflen,
                  int channels,
                  int bytes_per_sample,
                  float volume) {
  DCHECK(buf);
  if (volume < 0.0f || volume > 1.0f)
    return false;
  if (volume == 1.0f) {
    return true;
  } else if (volume == 0.0f) {
    memset(buf, 0, buflen);
    return true;
  }
  if (channels > 0 && channels <= 8 && bytes_per_sample > 0) {
    int sample_count = buflen / bytes_per_sample;
    const int fixed_volume = static_cast<int>(volume * 65536);
    if (bytes_per_sample == 1) {
      AdjustVolume<uint8, int32, 128>(reinterpret_cast<uint8*>(buf),
                                      sample_count,
                                      fixed_volume);
      return true;
    } else if (bytes_per_sample == 2) {
      AdjustVolume<int16, int32, 0>(reinterpret_cast<int16*>(buf),
                                    sample_count,
                                    fixed_volume);
      return true;
    } else if (bytes_per_sample == 4) {
      AdjustVolume<int32, int64, 0>(reinterpret_cast<int32*>(buf),
                                    sample_count,
                                    fixed_volume);
      return true;
    }
  }
  return false;
}

// TODO(enal): use template specialization and size-specific intrinsics.
//             Call is on the time-critical path, and by using SSE/AVX
//             instructions we can speed things up by ~4-8x, more for the case
//             when we have to adjust volume as well.
template<class Format, class Fixed, int min_value, int max_value, int bias>
static void MixStreams(Format* dst, Format* src, int count, float volume) {
  if (volume == 0.0f)
    return;
  if (volume == 1.0f) {
    // Most common case -- no need to adjust volume.
    for (int i = 0; i < count; ++i) {
      Fixed value = AddSaturated<Fixed, min_value, max_value>(dst[i] - bias,
                                                              src[i] - bias);
      dst[i] = static_cast<Format>(value + bias);
    }
  } else {
    // General case -- have to adjust volume before mixing.
    const int fixed_volume = static_cast<int>(volume * 65536);
    for (int i = 0; i < count; ++i) {
      Fixed adjusted_src = ScaleChannel<Fixed>(src[i] - bias, fixed_volume);
      Fixed value = AddSaturated<Fixed, min_value, max_value>(dst[i] - bias,
                                                              adjusted_src);
      dst[i] = static_cast<Format>(value + bias);
    }
  }
}

void MixStreams(void* dst,
                void* src,
                size_t buflen,
                int bytes_per_sample,
                float volume) {
  DCHECK(dst);
  DCHECK(src);
  DCHECK_GE(volume, 0.0f);
  DCHECK_LE(volume, 1.0f);
  switch (bytes_per_sample) {
    case 1:
      MixStreams<uint8, int32, -128, 127, 128>(static_cast<uint8*>(dst),
                                               static_cast<uint8*>(src),
                                               buflen,
                                               volume);
      break;
    case 2:
      DCHECK_EQ(0u, buflen % 2);
      MixStreams<int16, int32, -32768, 32767, 0>(static_cast<int16*>(dst),
                                                 static_cast<int16*>(src),
                                                 buflen / 2,
                                                 volume);
      break;
    case 4:
      DCHECK_EQ(0u, buflen % 4);
      MixStreams<int32, int64, 0x80000000, 0x7fffffff, 0>(
          static_cast<int32*>(dst),
          static_cast<int32*>(src),
          buflen / 4,
          volume);
      break;
    default:
      NOTREACHED() << "Illegal bytes per sample";
      break;
  }
}

int GetAudioHardwareSampleRate() {
#if defined(OS_MACOSX)
  // Hardware sample-rate on the Mac can be configured, so we must query.
  return AUAudioOutputStream::HardwareSampleRate();
#elif defined(OS_WIN)
  if (!CoreAudioUtil::IsSupported()) {
    // Fall back to Windows Wave implementation on Windows XP or lower
    // and use 48kHz as default input sample rate.
    return 48000;
  }

  // TODO(crogers): tune this rate for best possible WebAudio performance.
  // WebRTC works well at 48kHz and a buffer size of 480 samples will be used
  // for this case. Note that exclusive mode is experimental.
  const CommandLine* cmd_line = CommandLine::ForCurrentProcess();
  if (cmd_line->HasSwitch(switches::kEnableExclusiveAudio)) {
    // This sample rate will be combined with a buffer size of 256 samples
    // (see GetAudioHardwareBufferSize()), which corresponds to an output
    // delay of ~5.33ms.
    return 48000;
  }

  // Hardware sample-rate on Windows can be configured, so we must query.
  // TODO(henrika): improve possibility to specify an audio endpoint.
  // Use the default device (same as for Wave) for now to be compatible
  // or possibly remove the ERole argument completely until it is in use.
  return WASAPIAudioOutputStream::HardwareSampleRate(eConsole);
#elif defined(OS_ANDROID)
  return 16000;
#else
  // Hardware for Linux is nearly always 48KHz.
  // TODO(crogers) : return correct value in rare non-48KHz cases.
  return 48000;
#endif
}

int GetAudioInputHardwareSampleRate(const std::string& device_id) {
  // TODO(henrika): add support for device selection on all platforms.
  // Only exists on Windows today.
#if defined(OS_MACOSX)
  return AUAudioInputStream::HardwareSampleRate();
#elif defined(OS_WIN)
  if (!CoreAudioUtil::IsSupported()) {
    return 48000;
  }
  return WASAPIAudioInputStream::HardwareSampleRate(device_id);
#elif defined(OS_ANDROID)
  return 16000;
#else
  return 48000;
#endif
}

size_t GetAudioHardwareBufferSize() {
  int user_buffer_size = GetUserBufferSize();
  if (user_buffer_size)
    return user_buffer_size;

  // The sizes here were determined by experimentation and are roughly
  // the lowest value (for low latency) that still allowed glitch-free
  // audio under high loads.
  //
  // For Mac OS X and Windows the chromium audio backend uses a low-latency
  // Core Audio API, so a low buffer size is possible. For Linux, further
  // tuning may be needed.
#if defined(OS_MACOSX)
  return 128;
#elif defined(OS_WIN)
  // Buffer size to use when a proper size can't be determined from the system.
  static const int kFallbackBufferSize = 4096;

  if (!CoreAudioUtil::IsSupported()) {
    // Fall back to Windows Wave implementation on Windows XP or lower
    // and assume 48kHz as default sample rate.
    return kFallbackBufferSize;
  }

  // TODO(crogers): tune this size to best possible WebAudio performance.
  // WebRTC always uses 10ms for Windows and does not call this method.
  // Note that exclusive mode is experimental.
  const CommandLine* cmd_line = CommandLine::ForCurrentProcess();
  if (cmd_line->HasSwitch(switches::kEnableExclusiveAudio)) {
    return 256;
  }

  // TODO(henrika): remove when the --enable-webaudio-input flag is no longer
  // utilized.
  if (cmd_line->HasSwitch(switches::kEnableWebAudioInput)) {
    AudioParameters params;
    HRESULT hr = CoreAudioUtil::GetPreferredAudioParameters(eRender, eConsole,
                                                            &params);
    return FAILED(hr) ? kFallbackBufferSize : params.frames_per_buffer();
  }

  // This call must be done on a COM thread configured as MTA.
  // TODO(tommi): http://code.google.com/p/chromium/issues/detail?id=103835.
  int mixing_sample_rate =
      WASAPIAudioOutputStream::HardwareSampleRate(eConsole);

  // Windows will return a sample rate of 0 when no audio output is available
  // (i.e. via RemoteDesktop with remote audio disabled), but we should never
  // return a buffer size of zero.
  if (mixing_sample_rate == 0)
    return kFallbackBufferSize;

  // Use different buffer sizes depening on the sample rate . The existing
  // WASAPI implementation is tuned to provide the most stable callback
  // sequence using these combinations.
  if (mixing_sample_rate % 11025 == 0)
    // Use buffer size of ~10.15873 ms.
    return (112 * (mixing_sample_rate / 11025));

  if (mixing_sample_rate % 8000 == 0)
    // Use buffer size of 10ms.
    return (80 * (mixing_sample_rate / 8000));

  // Ensure we always return a buffer size which is somewhat appropriate.
  LOG(ERROR) << "Unknown sample rate " << mixing_sample_rate << " detected.";
  if (mixing_sample_rate > limits::kMinSampleRate)
    return (mixing_sample_rate / 100);
  return kFallbackBufferSize;
#else
  return 2048;
#endif
}

ChannelLayout GetAudioInputHardwareChannelLayout(const std::string& device_id) {
  // TODO(henrika): add support for device selection on all platforms.
  // Only exists on Windows today.
#if defined(OS_MACOSX)
  return CHANNEL_LAYOUT_MONO;
#elif defined(OS_WIN)
  if (!CoreAudioUtil::IsSupported()) {
    // Fall back to Windows Wave implementation on Windows XP or lower and
    // use stereo by default.
    return CHANNEL_LAYOUT_STEREO;
  }
  return WASAPIAudioInputStream::HardwareChannelCount(device_id) == 1 ?
      CHANNEL_LAYOUT_MONO : CHANNEL_LAYOUT_STEREO;
#else
  return CHANNEL_LAYOUT_STEREO;
#endif
}

// Computes a buffer size based on the given |sample_rate|. Must be used in
// conjunction with AUDIO_PCM_LINEAR.
size_t GetHighLatencyOutputBufferSize(int sample_rate) {
  int user_buffer_size = GetUserBufferSize();
  if (user_buffer_size)
    return user_buffer_size;

  // TODO(vrk/crogers): The buffer sizes that this function computes is probably
  // overly conservative. However, reducing the buffer size to 2048-8192 bytes
  // caused crbug.com/108396. This computation should be revisited while making
  // sure crbug.com/108396 doesn't happen again.

  // The minimum number of samples in a hardware packet.
  // This value is selected so that we can handle down to 5khz sample rate.
  static const size_t kMinSamplesPerHardwarePacket = 1024;

  // The maximum number of samples in a hardware packet.
  // This value is selected so that we can handle up to 192khz sample rate.
  static const size_t kMaxSamplesPerHardwarePacket = 64 * 1024;

  // This constant governs the hardware audio buffer size, this value should be
  // chosen carefully.
  // This value is selected so that we have 8192 samples for 48khz streams.
  static const size_t kMillisecondsPerHardwarePacket = 170;

  // Select the number of samples that can provide at least
  // |kMillisecondsPerHardwarePacket| worth of audio data.
  size_t samples = kMinSamplesPerHardwarePacket;
  while (samples <= kMaxSamplesPerHardwarePacket &&
         samples * base::Time::kMillisecondsPerSecond <
         sample_rate * kMillisecondsPerHardwarePacket) {
    samples *= 2;
  }
  return samples;
}

#if defined(OS_WIN)

int NumberOfWaveOutBuffers() {
  // Use 4 buffers for Vista, 3 for everyone else:
  //  - The entire Windows audio stack was rewritten for Windows Vista and wave
  //    out performance was degraded compared to XP.
  //  - The regression was fixed in Windows 7 and most configurations will work
  //    with 2, but some (e.g., some Sound Blasters) still need 3.
  //  - Some XP configurations (even multi-processor ones) also need 3.
  return (base::win::GetVersion() == base::win::VERSION_VISTA) ? 4 : 3;
}

#endif

}  // namespace media