summaryrefslogtreecommitdiffstats
path: root/dom/media/AudioConverter.h
blob: c8cbbb794981978da4d0457d2a54b33ea1dffeaf (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
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#if !defined(AudioConverter_h)
#  define AudioConverter_h

#  include "MediaInfo.h"

// Forward declaration
typedef struct SpeexResamplerState_ SpeexResamplerState;

namespace mozilla {

template <AudioConfig::SampleFormat T>
struct AudioDataBufferTypeChooser;
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_U8> {
  typedef uint8_t Type;
};
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_S16> {
  typedef int16_t Type;
};
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_S24LSB> {
  typedef int32_t Type;
};
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_S24> {
  typedef int32_t Type;
};
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_S32> {
  typedef int32_t Type;
};
template <>
struct AudioDataBufferTypeChooser<AudioConfig::FORMAT_FLT> {
  typedef float Type;
};

// 'Value' is the type used externally to deal with stored value.
// AudioDataBuffer can perform conversion between different SampleFormat
// content.
template <AudioConfig::SampleFormat Format,
          typename Value = typename AudioDataBufferTypeChooser<Format>::Type>
class AudioDataBuffer {
 public:
  AudioDataBuffer() = default;
  AudioDataBuffer(Value* aBuffer, size_t aLength) : mBuffer(aBuffer, aLength) {}
  explicit AudioDataBuffer(const AudioDataBuffer& aOther)
      : mBuffer(aOther.mBuffer) {}
  AudioDataBuffer(AudioDataBuffer&& aOther)
      : mBuffer(std::move(aOther.mBuffer)) {}
  template <AudioConfig::SampleFormat OtherFormat, typename OtherValue>
  explicit AudioDataBuffer(
      const AudioDataBuffer<OtherFormat, OtherValue>& other) {
    // TODO: Convert from different type, may use asm routines.
    MOZ_CRASH("Conversion not implemented yet");
  }

  // A u8, s16 and float aligned buffer can only be treated as
  // FORMAT_U8, FORMAT_S16 and FORMAT_FLT respectively.
  // So allow them as copy and move constructors.
  explicit AudioDataBuffer(const AlignedByteBuffer& aBuffer)
      : mBuffer(aBuffer) {
    static_assert(Format == AudioConfig::FORMAT_U8,
                  "Conversion not implemented yet");
  }
  explicit AudioDataBuffer(const AlignedShortBuffer& aBuffer)
      : mBuffer(aBuffer) {
    static_assert(Format == AudioConfig::FORMAT_S16,
                  "Conversion not implemented yet");
  }
  explicit AudioDataBuffer(const AlignedFloatBuffer& aBuffer)
      : mBuffer(aBuffer) {
    static_assert(Format == AudioConfig::FORMAT_FLT,
                  "Conversion not implemented yet");
  }
  explicit AudioDataBuffer(AlignedByteBuffer&& aBuffer)
      : mBuffer(std::move(aBuffer)) {
    static_assert(Format == AudioConfig::FORMAT_U8,
                  "Conversion not implemented yet");
  }
  explicit AudioDataBuffer(AlignedShortBuffer&& aBuffer)
      : mBuffer(std::move(aBuffer)) {
    static_assert(Format == AudioConfig::FORMAT_S16,
                  "Conversion not implemented yet");
  }
  explicit AudioDataBuffer(AlignedFloatBuffer&& aBuffer)
      : mBuffer(std::move(aBuffer)) {
    static_assert(Format == AudioConfig::FORMAT_FLT,
                  "Conversion not implemented yet");
  }
  AudioDataBuffer& operator=(AudioDataBuffer&& aOther) {
    mBuffer = std::move(aOther.mBuffer);
    return *this;
  }
  AudioDataBuffer& operator=(const AudioDataBuffer& aOther) {
    mBuffer = aOther.mBuffer;
    return *this;
  }

  Value* Data() const { return mBuffer.Data(); }
  size_t Length() const { return mBuffer.Length(); }
  size_t Size() const { return mBuffer.Size(); }
  AlignedBuffer<Value> Forget() {
    // Correct type -> Just give values as-is.
    return std::move(mBuffer);
  }

 private:
  AlignedBuffer<Value> mBuffer;
};

typedef AudioDataBuffer<AudioConfig::FORMAT_DEFAULT> AudioSampleBuffer;

class AudioConverter {
 public:
  AudioConverter(const AudioConfig& aIn, const AudioConfig& aOut);
  ~AudioConverter();

  // Convert the AudioDataBuffer.
  // Conversion will be done in place if possible. Otherwise a new buffer will
  // be returned.
  // Providing an empty buffer and resampling is expected, the resampler
  // will be drained.
  template <AudioConfig::SampleFormat Format, typename Value>
  AudioDataBuffer<Format, Value> Process(
      AudioDataBuffer<Format, Value>&& aBuffer) {
    MOZ_DIAGNOSTIC_ASSERT(mIn.Format() == mOut.Format() &&
                          mIn.Format() == Format);
    AudioDataBuffer<Format, Value> buffer = std::move(aBuffer);
    if (CanWorkInPlace()) {
      AlignedBuffer<Value> temp = buffer.Forget();
      Process(temp, temp.Data(), SamplesInToFrames(temp.Length()));
      return AudioDataBuffer<Format, Value>(std::move(temp));
    }
    return Process(buffer);
  }

  template <AudioConfig::SampleFormat Format, typename Value>
  AudioDataBuffer<Format, Value> Process(
      const AudioDataBuffer<Format, Value>& aBuffer) {
    MOZ_DIAGNOSTIC_ASSERT(mIn.Format() == mOut.Format() &&
                          mIn.Format() == Format);
    // Perform the downmixing / reordering in temporary buffer.
    size_t frames = SamplesInToFrames(aBuffer.Length());
    AlignedBuffer<Value> temp1;
    if (!temp1.SetLength(FramesOutToSamples(frames))) {
      return AudioDataBuffer<Format, Value>(std::move(temp1));
    }
    frames = ProcessInternal(temp1.Data(), aBuffer.Data(), frames);
    if (mIn.Rate() == mOut.Rate()) {
      MOZ_ALWAYS_TRUE(temp1.SetLength(FramesOutToSamples(frames)));
      return AudioDataBuffer<Format, Value>(std::move(temp1));
    }

    // At this point, temp1 contains the buffer reordered and downmixed.
    // If we are downsampling we can re-use it.
    AlignedBuffer<Value>* outputBuffer = &temp1;
    AlignedBuffer<Value> temp2;
    if (!frames || mOut.Rate() > mIn.Rate()) {
      // We are upsampling or about to drain, we can't work in place.
      // Allocate another temporary buffer where the upsampling will occur.
      if (!temp2.SetLength(
              FramesOutToSamples(ResampleRecipientFrames(frames)))) {
        return AudioDataBuffer<Format, Value>(std::move(temp2));
      }
      outputBuffer = &temp2;
    }
    if (!frames) {
      frames = DrainResampler(outputBuffer->Data());
    } else {
      frames = ResampleAudio(outputBuffer->Data(), temp1.Data(), frames);
    }
    MOZ_ALWAYS_TRUE(outputBuffer->SetLength(FramesOutToSamples(frames)));
    return AudioDataBuffer<Format, Value>(std::move(*outputBuffer));
  }

  // Attempt to convert the AudioDataBuffer in place.
  // Will return 0 if the conversion wasn't possible.
  template <typename Value>
  size_t Process(Value* aBuffer, size_t aFrames) {
    MOZ_DIAGNOSTIC_ASSERT(mIn.Format() == mOut.Format());
    if (!CanWorkInPlace()) {
      return 0;
    }
    size_t frames = ProcessInternal(aBuffer, aBuffer, aFrames);
    if (frames && mIn.Rate() != mOut.Rate()) {
      frames = ResampleAudio(aBuffer, aBuffer, aFrames);
    }
    return frames;
  }

  template <typename Value>
  size_t Process(AlignedBuffer<Value>& aOutBuffer, const Value* aInBuffer,
                 size_t aFrames) {
    MOZ_DIAGNOSTIC_ASSERT(mIn.Format() == mOut.Format());
    MOZ_ASSERT((aFrames && aInBuffer) || !aFrames);
    // Up/down mixing first
    if (!aOutBuffer.SetLength(FramesOutToSamples(aFrames))) {
      MOZ_ALWAYS_TRUE(aOutBuffer.SetLength(0));
      return 0;
    }
    size_t frames = ProcessInternal(aOutBuffer.Data(), aInBuffer, aFrames);
    MOZ_ASSERT(frames == aFrames);
    // Check if resampling is needed
    if (mIn.Rate() == mOut.Rate()) {
      return frames;
    }
    // Prepare output in cases of drain or up-sampling
    if ((!frames || mOut.Rate() > mIn.Rate()) &&
        !aOutBuffer.SetLength(
            FramesOutToSamples(ResampleRecipientFrames(frames)))) {
      MOZ_ALWAYS_TRUE(aOutBuffer.SetLength(0));
      return 0;
    }
    if (!frames) {
      frames = DrainResampler(aOutBuffer.Data());
    } else {
      frames = ResampleAudio(aOutBuffer.Data(), aInBuffer, frames);
    }
    // Update with the actual buffer length
    MOZ_ALWAYS_TRUE(aOutBuffer.SetLength(FramesOutToSamples(frames)));
    return frames;
  }

  bool CanWorkInPlace() const;
  bool CanReorderAudio() const {
    return mIn.Layout().MappingTable(mOut.Layout());
  }
  static bool CanConvert(const AudioConfig& aIn, const AudioConfig& aOut);

  const AudioConfig& InputConfig() const { return mIn; }
  const AudioConfig& OutputConfig() const { return mOut; }

 private:
  const AudioConfig mIn;
  const AudioConfig mOut;
  // mChannelOrderMap will be empty if we do not know how to proceed with this
  // channel layout.
  AutoTArray<uint8_t, AudioConfig::ChannelLayout::MAX_CHANNELS>
      mChannelOrderMap;
  /**
   * ProcessInternal
   * Parameters:
   * aOut  : destination buffer where converted samples will be copied
   * aIn   : source buffer
   * aSamples: number of frames in source buffer
   *
   * Return Value: number of frames converted or 0 if error
   */
  size_t ProcessInternal(void* aOut, const void* aIn, size_t aFrames);
  void ReOrderInterleavedChannels(void* aOut, const void* aIn,
                                  size_t aFrames) const;
  size_t DownmixAudio(void* aOut, const void* aIn, size_t aFrames) const;
  size_t UpmixAudio(void* aOut, const void* aIn, size_t aFrames) const;

  size_t FramesOutToSamples(size_t aFrames) const;
  size_t SamplesInToFrames(size_t aSamples) const;
  size_t FramesOutToBytes(size_t aFrames) const;

  // Resampler context.
  SpeexResamplerState* mResampler;
  size_t ResampleAudio(void* aOut, const void* aIn, size_t aFrames);
  size_t ResampleRecipientFrames(size_t aFrames) const;
  void RecreateResampler();
  size_t DrainResampler(void* aOut);
};

}  // namespace mozilla

#endif /* AudioConverter_h */