AU格式解析

扩展名：au,snd

公司：Sun

        AU又称SUD，是一种简单的PCM文件被广泛应用于Sun系统。这种格式的组成由原始数据紧跟在文件头之后。

        下面是文件头数据结构，所有字段按大段字节序排列：

magic            32 bits  ".snd"data_start       32       absolute file offset of first sampledata_size        32       size of sample data, 0 means unspecifiedsample_format    32       sample format code, see belowsample_rate      32       sample rate in samples/secondchannels         32       number of channels

The header may be followed by a chunk of arbitrary data if the data_start field is greater than 24.

The sample format codes are:

 0  unspecified 1  8-bit mu-law 2  8-bit signed linear 3  16-bit signed linear 4  24-bit signed linear 5  32-bit signed linear 6  floating-point 7  double precision float 8  fragmented sampled data 9  unknown10  DSP program11  8-bit fixed-point12  16-bit fixed-point13  24-bit fixed-point14  32-bit fixed-point15  unknown16  non-audio data17  mu-law squelch18  16-bit linear with emphasis19  16-bit linear with compression20  16-bit linear with emphasis and compression21  Music Kit DSP commands22  Music Kit DSP samples23  G.721 ADPCM24  G.722 ADPCM25  G.723 ADPCM26  5-bit G.723 ADPCM27  8-bit a-law

其实ffmpeg中au.c只对1,2,3,4,5,6,7,24,27，这几种采样格式编码做了处理。由上可知AU格式的数据部分是经过压缩的PCM数据或者ADPCM数据，而且是一种有损压缩，一般压缩比在2:1左右。

关于bits_per_sample，是根据PCM类型来确定的，下面给出ffmpeg中定义的所有PCM类型，ADPCM就不给出了：

8位采样

AV_CODEC_ID_PCM_ALAW

AV_CODEC_ID_PCM_MULAW

AV_CODEC_ID_PCM_S8

AV_CODEC_ID_PCM_S8_PLANAR

AV_CODEC_ID_PCM_U8

AV_CODEC_ID_PCM_ZORK

16位采样

AV_CODEC_ID_PCM_S16BE

AV_CODEC_ID_PCM_S16BE_PLANAR

AV_CODEC_ID_PCM_S16LE                //此为WAV格式中PCM数据的类型

AV_CODEC_ID_PCM_S16LE_PLANAR

AV_CODEC_ID_PCM_U16BE

AV_CODEC_ID_PCM_U16LE

24位采样

AV_CODEC_ID_PCM_S24DAUD

AV_CODEC_ID_PCM_S24BE

AV_CODEC_ID_PCM_S24LE

AV_CODEC_ID_PCM_S24LE_PLANAR

AV_CODEC_ID_PCM_U24BE

AV_CODEC_ID_PCM_U24LE

32位采样

AV_CODEC_ID_PCM_S32BE

AV_CODEC_ID_PCM_S32LE

AV_CODEC_ID_PCM_S32LE_PLANAR

AV_CODEC_ID_PCM_U32BE

AV_CODEC_ID_PCM_U32LE

AV_CODEC_ID_PCM_F32BE

AV_CODEC_ID_PCM_F32LE

64位采样

AV_CODEC_ID_PCM_F64BE

AV_CODEC_ID_PCM_F64LE

还有一些关于ADPCM的请见ffmpeg中utils.c

可以通过已知量计算得到的量：

如PCM是AV_CODEC_ID_PCM_MULAW，即sample_format为1，

则采样位数bps = 8(bits_per_sample)，

比特率bit_rate = channels * sample_rate * bps

总播放时间duration = data_size / bit_rate

块对齐block_align = (bps * channels / 8) > 1 ?(bps * channels / 8) : 1

AU格式的seek公式如下

若为PCM数据：

pos = seektime * sample_rate * (bps / 8) *channels + data_start                  

若为ADPCM数据：

pos_tmp = seektime * byterate / blockalign                                                              //注意是byterate，即bit_rate*8 

pos = pos_tmp * blockalign + data_start                                                                  

存在的疑惑：已知AU中PCM数据是经过压缩的PCM数据或者ADPCM数据，且根据压缩方法不同其类型也各不相同，那么将它们解压缩的时候该用什么方法呢？会像ADPCM一样利用一些已知的表么？