1 /*
2 * Copyright (c) 2014 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 // Based on the WAV file format documentation at
12 // https://ccrma.stanford.edu/courses/422/projects/WaveFormat/ and
13 // http://www-mmsp.ece.mcgill.ca/Documents/AudioFormats/WAVE/WAVE.html
14
15 #include "webrtc/common_audio/wav_header.h"
16
17 #include <algorithm>
18 #include <cstring>
19 #include <limits>
20 #include <string>
21
22 #include "webrtc/base/checks.h"
23 #include "webrtc/common_audio/include/audio_util.h"
24
25 namespace webrtc {
26 namespace {
27
28 struct ChunkHeader {
29 uint32_t ID;
30 uint32_t Size;
31 };
32 static_assert(sizeof(ChunkHeader) == 8, "ChunkHeader size");
33
34 // We can't nest this definition in WavHeader, because VS2013 gives an error
35 // on sizeof(WavHeader::fmt): "error C2070: 'unknown': illegal sizeof operand".
36 struct FmtSubchunk {
37 ChunkHeader header;
38 uint16_t AudioFormat;
39 uint16_t NumChannels;
40 uint32_t SampleRate;
41 uint32_t ByteRate;
42 uint16_t BlockAlign;
43 uint16_t BitsPerSample;
44 };
45 static_assert(sizeof(FmtSubchunk) == 24, "FmtSubchunk size");
46 const uint32_t kFmtSubchunkSize = sizeof(FmtSubchunk) - sizeof(ChunkHeader);
47
48 struct WavHeader {
49 struct {
50 ChunkHeader header;
51 uint32_t Format;
52 } riff;
53 FmtSubchunk fmt;
54 struct {
55 ChunkHeader header;
56 } data;
57 };
58 static_assert(sizeof(WavHeader) == kWavHeaderSize, "no padding in header");
59
60 } // namespace
61
CheckWavParameters(int num_channels,int sample_rate,WavFormat format,int bytes_per_sample,uint32_t num_samples)62 bool CheckWavParameters(int num_channels,
63 int sample_rate,
64 WavFormat format,
65 int bytes_per_sample,
66 uint32_t num_samples) {
67 // num_channels, sample_rate, and bytes_per_sample must be positive, must fit
68 // in their respective fields, and their product must fit in the 32-bit
69 // ByteRate field.
70 if (num_channels <= 0 || sample_rate <= 0 || bytes_per_sample <= 0)
71 return false;
72 if (static_cast<uint64_t>(sample_rate) > std::numeric_limits<uint32_t>::max())
73 return false;
74 if (static_cast<uint64_t>(num_channels) >
75 std::numeric_limits<uint16_t>::max())
76 return false;
77 if (static_cast<uint64_t>(bytes_per_sample) * 8 >
78 std::numeric_limits<uint16_t>::max())
79 return false;
80 if (static_cast<uint64_t>(sample_rate) * num_channels * bytes_per_sample >
81 std::numeric_limits<uint32_t>::max())
82 return false;
83
84 // format and bytes_per_sample must agree.
85 switch (format) {
86 case kWavFormatPcm:
87 // Other values may be OK, but for now we're conservative:
88 if (bytes_per_sample != 1 && bytes_per_sample != 2)
89 return false;
90 break;
91 case kWavFormatALaw:
92 case kWavFormatMuLaw:
93 if (bytes_per_sample != 1)
94 return false;
95 break;
96 default:
97 return false;
98 }
99
100 // The number of bytes in the file, not counting the first ChunkHeader, must
101 // be less than 2^32; otherwise, the ChunkSize field overflows.
102 const uint32_t max_samples =
103 (std::numeric_limits<uint32_t>::max()
104 - (kWavHeaderSize - sizeof(ChunkHeader))) /
105 bytes_per_sample;
106 if (num_samples > max_samples)
107 return false;
108
109 // Each channel must have the same number of samples.
110 if (num_samples % num_channels != 0)
111 return false;
112
113 return true;
114 }
115
116 #ifdef WEBRTC_ARCH_LITTLE_ENDIAN
WriteLE16(uint16_t * f,uint16_t x)117 static inline void WriteLE16(uint16_t* f, uint16_t x) { *f = x; }
WriteLE32(uint32_t * f,uint32_t x)118 static inline void WriteLE32(uint32_t* f, uint32_t x) { *f = x; }
WriteFourCC(uint32_t * f,char a,char b,char c,char d)119 static inline void WriteFourCC(uint32_t* f, char a, char b, char c, char d) {
120 *f = static_cast<uint32_t>(a)
121 | static_cast<uint32_t>(b) << 8
122 | static_cast<uint32_t>(c) << 16
123 | static_cast<uint32_t>(d) << 24;
124 }
125
ReadLE16(uint16_t x)126 static inline uint16_t ReadLE16(uint16_t x) { return x; }
ReadLE32(uint32_t x)127 static inline uint32_t ReadLE32(uint32_t x) { return x; }
ReadFourCC(uint32_t x)128 static inline std::string ReadFourCC(uint32_t x) {
129 return std::string(reinterpret_cast<char*>(&x), 4);
130 }
131 #else
132 #error "Write be-to-le conversion functions"
133 #endif
134
RiffChunkSize(uint32_t bytes_in_payload)135 static inline uint32_t RiffChunkSize(uint32_t bytes_in_payload) {
136 return bytes_in_payload + kWavHeaderSize - sizeof(ChunkHeader);
137 }
138
ByteRate(int num_channels,int sample_rate,int bytes_per_sample)139 static inline uint32_t ByteRate(int num_channels, int sample_rate,
140 int bytes_per_sample) {
141 return static_cast<uint32_t>(num_channels) * sample_rate * bytes_per_sample;
142 }
143
BlockAlign(int num_channels,int bytes_per_sample)144 static inline uint16_t BlockAlign(int num_channels, int bytes_per_sample) {
145 return num_channels * bytes_per_sample;
146 }
147
WriteWavHeader(uint8_t * buf,int num_channels,int sample_rate,WavFormat format,int bytes_per_sample,uint32_t num_samples)148 void WriteWavHeader(uint8_t* buf,
149 int num_channels,
150 int sample_rate,
151 WavFormat format,
152 int bytes_per_sample,
153 uint32_t num_samples) {
154 RTC_CHECK(CheckWavParameters(num_channels, sample_rate, format,
155 bytes_per_sample, num_samples));
156
157 WavHeader header;
158 const uint32_t bytes_in_payload = bytes_per_sample * num_samples;
159
160 WriteFourCC(&header.riff.header.ID, 'R', 'I', 'F', 'F');
161 WriteLE32(&header.riff.header.Size, RiffChunkSize(bytes_in_payload));
162 WriteFourCC(&header.riff.Format, 'W', 'A', 'V', 'E');
163
164 WriteFourCC(&header.fmt.header.ID, 'f', 'm', 't', ' ');
165 WriteLE32(&header.fmt.header.Size, kFmtSubchunkSize);
166 WriteLE16(&header.fmt.AudioFormat, format);
167 WriteLE16(&header.fmt.NumChannels, num_channels);
168 WriteLE32(&header.fmt.SampleRate, sample_rate);
169 WriteLE32(&header.fmt.ByteRate, ByteRate(num_channels, sample_rate,
170 bytes_per_sample));
171 WriteLE16(&header.fmt.BlockAlign, BlockAlign(num_channels, bytes_per_sample));
172 WriteLE16(&header.fmt.BitsPerSample, 8 * bytes_per_sample);
173
174 WriteFourCC(&header.data.header.ID, 'd', 'a', 't', 'a');
175 WriteLE32(&header.data.header.Size, bytes_in_payload);
176
177 // Do an extra copy rather than writing everything to buf directly, since buf
178 // might not be correctly aligned.
179 memcpy(buf, &header, kWavHeaderSize);
180 }
181
ReadWavHeader(ReadableWav * readable,int * num_channels,int * sample_rate,WavFormat * format,int * bytes_per_sample,uint32_t * num_samples)182 bool ReadWavHeader(ReadableWav* readable,
183 int* num_channels,
184 int* sample_rate,
185 WavFormat* format,
186 int* bytes_per_sample,
187 uint32_t* num_samples) {
188 WavHeader header;
189 if (readable->Read(&header, kWavHeaderSize - sizeof(header.data)) !=
190 kWavHeaderSize - sizeof(header.data))
191 return false;
192
193 const uint32_t fmt_size = ReadLE32(header.fmt.header.Size);
194 if (fmt_size != kFmtSubchunkSize) {
195 // There is an optional two-byte extension field permitted to be present
196 // with PCM, but which must be zero.
197 int16_t ext_size;
198 if (kFmtSubchunkSize + sizeof(ext_size) != fmt_size)
199 return false;
200 if (readable->Read(&ext_size, sizeof(ext_size)) != sizeof(ext_size))
201 return false;
202 if (ext_size != 0)
203 return false;
204 }
205 if (readable->Read(&header.data, sizeof(header.data)) != sizeof(header.data))
206 return false;
207
208 // Parse needed fields.
209 *format = static_cast<WavFormat>(ReadLE16(header.fmt.AudioFormat));
210 *num_channels = ReadLE16(header.fmt.NumChannels);
211 *sample_rate = ReadLE32(header.fmt.SampleRate);
212 *bytes_per_sample = ReadLE16(header.fmt.BitsPerSample) / 8;
213 const uint32_t bytes_in_payload = ReadLE32(header.data.header.Size);
214 if (*bytes_per_sample <= 0)
215 return false;
216 *num_samples = bytes_in_payload / *bytes_per_sample;
217
218 // Sanity check remaining fields.
219 if (ReadFourCC(header.riff.header.ID) != "RIFF")
220 return false;
221 if (ReadFourCC(header.riff.Format) != "WAVE")
222 return false;
223 if (ReadFourCC(header.fmt.header.ID) != "fmt ")
224 return false;
225 if (ReadFourCC(header.data.header.ID) != "data")
226 return false;
227
228 if (ReadLE32(header.riff.header.Size) < RiffChunkSize(bytes_in_payload))
229 return false;
230 if (ReadLE32(header.fmt.ByteRate) !=
231 ByteRate(*num_channels, *sample_rate, *bytes_per_sample))
232 return false;
233 if (ReadLE16(header.fmt.BlockAlign) !=
234 BlockAlign(*num_channels, *bytes_per_sample))
235 return false;
236
237 return CheckWavParameters(*num_channels, *sample_rate, *format,
238 *bytes_per_sample, *num_samples);
239 }
240
241
242 } // namespace webrtc
243