让终端支持播放mp3，移植mp3解码库libmad和madplay到嵌入式linux

短命女 2023-07-14 05:41 26阅读 0赞

libmad简介

官网：[https://www.underbit.com/products/mad/][https_www.underbit.com_products_mad]

MAD （libmad）是一个开源的高精度 MPEG 音频解码库，支持 MPEG-1（Layer I, Layer II 和 LayerIII（也就是 MP3）。LIBMAD 提供 24  
\-bit 的 PCM 输出，完全是定点计算，非常适合没有浮点支持的平台上使用。使用 libmad 提供的一系列 API，就可以非常简单地实现 MP3 数据解码工作。

移植涉及到的库：  
zlib-1.2.3.tar.gz  
libid3tag-0.15.1b.tar.gz  
libmad-0.15.1b.tar.gz  
\---------------------------------------------------------------------------------------------------------------  
madplay介绍：  
 madplay基于libmad的基础上做了一个播放器，该播放器除了目前不支持网络播放以为，其余功能都支持。如快进播放，seek播放，暂停，恢复等  
最后移植了一个基于libmad的应用madplay,可以直接用它来播放mp3.  
madplay-0.15.2b.tar.gz

介绍完了，当然移植不是一番风顺的，折腾了一天。中间遇到不少问题，还好，最后都一一解决了。

网上提供的代码。交叉编译，由于环境不一样，会出现各种问题。

这里记录一下过程。

首先是交叉编译zlib-1.2.3.tar.gz

用交叉编译工具编译zlib，并且把库生成到交叉编译环境的库目录下  
./configure --prefix=/home/ban/madplay/source     
修改Makefile.  
CC=arm-linux-gnueabihf-gcc  
AR=arm-linux-gnueabihf-ar rc  
RANLIB=arm-linux-gnueabihf-ranlib  
make   
make install  
安装完成后，在/home/ban/madplay/source/ 中将生产lib跟include2个文件夹。

这步一般不会有啥问题，但是，默认成功的是静态库啊， 虽然配置上是说默认配置生成动态库，但是确实没有。

实际是有的，指定--shared即可。

或者仔细查makefile,把相关的编译语句找出来，我手动调用gcc - shared -fPIC - $(OBJS)生成了.so

这里还需要注意的是,指定好自己的--prifix,因为后续的编译，好多是要依赖这个的。

接下来编译libid3tag-0.15.1b.tar.gz

这时候要注意了，如果上一步编译不过，或者没有指定--prifix, 这里就麻烦了。

由于我需要的是动态库，发现configer后，竟没带-FPIC参数，还要去改makefile才行。

./configure --host=arm-linux-gnueabihf  --disable-debugging --prefix=/home/ban/madplay/source CPPFLAGS=-I/home/ban/madplay/source/include LDFLAGS=-L/home/ban/madplay/source/lib  
make   
make install

编译libmad  
./configure --host=arm-linux-gnueabihf  --disable-debugging --prefix=/home/ban/madplay/source CPPFLAGS=-I/home/ban/madplay/source/include LDFLAGS=-L/home/ban/madplay/source/lib  
make  
make install

出现错误：  
cc1: error: unrecognized command line option “-fforce-mem”  
原因是高版本的gcc，已经将-fforce-mem去除了，解决方法：  
 sed -i '/-fforce-mem/d' configure

再执行：  
./configure --host=arm-linux-gnueabihf --prefix=/usr/local/libmad\_arm --enable-shared --enable-static --enable-fpm=arm --

with-gnu-ld=arm-linux-gnueabihf-ld --build=arm  
出现错误：  
/tmp/ccf2FxyW.s:1299: Error: selected processor does not support Thumb mode \`rsc r0,r0,\#0'  
/tmp/ccf2FxyW.s:1435: Error: selected processor does not support Thumb mode \`rsc r8,r8,\#0'  
/tmp/ccf2FxyW.s:1857: Error: selected processor does not support Thumb mode \`rsc r0,r0,\#0'  
/tmp/ccf2FxyW.s:1996: Error: selected processor does not support Thumb mode \`rsc r0,r0,\#0  
百度一下发现这是libmad的一个bug.  
解决方法是：  
vim  fixed.h  
将  
\#  define MAD\_F\_MLN(hi, lo)  \\  
    asm ("rsbs  %0, %2, \#0\\n\\t"  \\  
         "rsc   %1, %3, \#0"  \\  
         : "=r" (lo), "=r" (hi)  \\  
         : "0" (lo), "1" (hi)  \\  
         : "cc")  
改为  
\#ifdef \_\_thumb\_\_  
/\* In Thumb-2, the RSB-immediate instruction is only allowed with a zero  
operand. If needed this code can also support Thumb-1   
(simply append "s" to the end of the second two instructions). \*/  
\# define MAD\_F\_MLN(hi, lo) \\  
asm ("rsbs %0, %0, \#0\\n\\t" \\  
"       sbc %1, %1, %1\\n\\t" \\  
        "sub %1, %1, %2" \\  
        : "+&r" (lo), "=&r" (hi) \\  
        : "r" (hi) \\  
        : "cc")  
\#else /\* ! \_\_thumb\_\_ \*/  
\# define MAD\_F\_MLN(hi, lo) \\  
        asm ("rsbs %0, %2, \#0\\n\\t" \\  
        "rsc %1, %3, \#0" \\  
         : "=r" (lo), "=r" (hi) \\  
          : "=&r" (lo), "=r" (hi) \\  
          : "0" (lo), "1" (hi) \\  
          : "cc")  
\#endif /\* \_\_thumb\_\_ \*/  
再make，编译通过了！

编译madplay  
./configure --host=arm-linux-gnueabihf CC=arm-linux-gnueabihf-gcc --disable-debugging --with-alsa CPPFLAGS=-I/home/ban/madplay/source/include LDFLAGS=-L/home/ban/madplay/source/lib  
make   
make install  
完成以后把生成的可执行文件madplay下载到开发板中  
执行./madplay filename.mp3

这个需要注意的是，如果不指定--with-alsa，即便编译成功了，放到板子上也是跑不起的，提示找不到dev/dsp,这个让我折腾了好久，竟发现，配置上没启用alsa啊，

但板子上带的是alsa架构的linux音频驱动。

总体上操作是就这么几步，但是，你会发现，如果照这个步骤来，仍是有错。

具体细节。，根据编译提示的错误，基本都能定为到。比如，找不到上几步编译出的库，就去改makefile吧，添加进去路径

或者仍拷贝到 --prifix指定的目录中。

如果编译遇到以下错误：

![20200313153652197.png][]

只需更新alsa api,libmad里的api太旧了，新的alsa-lib不支持了

解决方案参考：[https://gitlab.savoirfairelinux.com/netdsa/buildroot/commit/97b392b98468859ab2366995b6bee4140b38e388][https_gitlab.savoirfairelinux.com_netdsa_buildroot_commit_97b392b98468859ab2366995b6bee4140b38e388]

最后再说一点儿，编译网上这种开源库，最好设置下交叉工具链的环境变量为全局的，且用root权限。否则，坑真的好多。

附截图：

![Center][]

![Center 1][]

如果不用这个现成的播放器madplay,只测试下libmad是否成功，

可以编译测试下 libmad提供的一个简单demo,这个demo 不是播放mp3的，而是把mp3解码成 pcm文件 。

测试如下：

./testmad.out <demo1.mp3 >out1.pcm

显示出了信息，且在当前路径下产生了out1.pcm文件。

**mad功能：**

madpay功能很强求，可以实现多个节目循环播放，快进拖动播放，暂停，恢复，支持wav等格式输出等。  
目前madplay还不支持网络播放.不过mp123支持网络播放。  
madplay使用方法： 传入打印既可以使用短传入(-n )，也可以使用长传入（--abc)，二者等效，下面只讲短传入  
\-v  获取播放时间  
\-q  不存在任何打印,但现实警告  
\-Q  不存在任何打印  
\--downsample  只采用一半数据  
\-i  忽略CRC校验错误  
\-o PATH/xx.wav 可以用来转码，将mp3转为wav  
  也可以是其他格式，见说明;.raw  表示是元素pcm，.hex等  
\-a 开启衰减音量 ，增加音量，衰减系统为-175~+18    （当然通过键盘+ —也可以调整音量）  
\-A 同-a  
\-1 -2 -m -S 分别指左声道，右声道，双声道，立体声  
\-s  用于seek播放 如：0:1:20:11 ，seek到1小时，2分钟，11秒时开始播放  
\-t  用于播放时间现在  0:1:20:11 ，播放到1小时，2分钟，11秒时就停止  
\-z  用于随机播放列表  
\-r, --repeat\[=MAX\]   循环播放无限次或Max次   
\--tty-control            enable keyboard controls  默认是使能热键  
\--no-tty-control         disable keyboard controls  
热键使用  
下一首歌曲; f，或ctrl+n 或者>  
上一首歌曲  b ,或ctrl+p 或者<  
退出:q, Q,获取ctrl+c  
获取播放状态信息: i  ？  
调节音量: - + \_ =  
暂停恢复：p  
停止:s  
Usage: ./madplay \[OPTIONS\] FILE \[...\]  
Decode and play MPEG audio FILE(s).

Verbosity:  
  -v, --verbose                show status while decoding  
  -q, --quiet                  be quiet but show warnings  
  -Q, --very-quiet             be quiet and do not show warnings  
      --display-time=MODE      use default verbose time display MODE  
                                 (remaining, current, overall)

Decoding:  
      --downsample             reduce sample rate 2:1  
  -i, --ignore-crc             ignore CRC errors  
      --ancillary-output=PATH  write ancillary data to PATH

Audio output:  
  -o, --output=\[TYPE:\]PATH     write output to PATH with format TYPE (below)  
  -b, --bit-depth=DEPTH        request DEPTH bits per sample  
  -R, --sample-rate=HERTZ      request HERTZ samples per second  
  -d, --no-dither              do not dither output PCM samples  
      --fade-in\[=DURATION\]     fade-in songs over DURATION (default 0:05)  
  -a, --attenuate=DECIBELS     attenuate signal by DECIBELS (-)  
  -a, --amplify=DECIBELS       amplify signal by DECIBELS (+)  
  -A, --adjust-volume=DECIBELS override per-file volume adjustments  
  -G, --replay-gain\[=PROFILE\]  enable Replay Gain volume adjustments using  
                                 PROFILE (radio, audiophile)

Channel selection:  
  -1, --left                   output first (left) channel only  
  -2, --right                  output second (right) channel only  
  -m, --mono                   mix left and right channels for monaural output  
  -S, --stereo                 force stereo output

Playback:  
  -s, --start=TIME             skip to begin at TIME (HH:MM:SS.DDD)  
  -t, --time=DURATION          play only for DURATION (HH:MM:SS.DDD)  
  -z, --shuffle                randomize file list  
  -r, --repeat\[=MAX\]           play files MAX times, or indefinitely  
      --tty-control            enable keyboard controls  
      --no-tty-control         disable keyboard controls

Miscellaneous:  
  -T, --show-tags-only         show ID3/encoder tags only (do not decode)  
  -V, --version                display version number and exit  
      --license                show copyright/license message and exit  
  -h, --help                   display this help and exit

Supported output formats:  
  cdda    CD audio, 16-bit big-endian 44100 Hz stereo PCM (\*.cdr, \*.cda)  
  aiff    Audio IFF, \[16-bit\] PCM (\*.aif, \*.aiff)  
  wave    Microsoft RIFF/WAVE, \[16-bit\] PCM (\*.wav)  
  snd     Sun/NeXT audio, 8-bit ISDN mu-law (\*.au, \*.snd)  
  raw     binary \[16-bit\] host-endian linear PCM  
  hex     ASCII hexadecimal \[24-bit\] linear PCM  
  null    no output (decode only)  
————————————————

9522 frames decoded (0:04:08.7), +1.7 dB peak amplitude, 4202 clipped samples

/*
     * libmad - MPEG audio decoder library
     * Copyright (C) 2000-2004 Underbit Technologies, Inc.
     *
     * This program is free software; you can redistribute it and/or modify
     * it under the terms of the GNU General Public License as published by
     * the Free Software Foundation; either version 2 of the License, or
     * (at your option) any later version.
     *
     * This program is distributed in the hope that it will be useful,
     * but WITHOUT ANY WARRANTY; without even the implied warranty of
     * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
     * GNU General Public License for more details.
     *
     * You should have received a copy of the GNU General Public License
     * along with this program; if not, write to the Free Software
     * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
     *
     * $Id: minimad.c,v 1.4 2004/01/23 09:41:32 rob Exp $
     */
     
    # include <stdio.h>
    # include <unistd.h>
    # include <sys/stat.h>
    # include <sys/mman.h>
     
    # include "mad.h"
     
    /*
     * This is perhaps the simplest example use of the MAD high-level API.
     * Standard input is mapped into memory via mmap(), then the high-level API
     * is invoked with three callbacks: input, output, and error. The output
     * callback converts MAD's high-resolution PCM samples to 16 bits, then
     * writes them to standard output in little-endian, stereo-interleaved
     * format.
     */
     
    static int decode(unsigned char const *, unsigned long);
     
    int main(int argc, char *argv[])
    {
      struct stat stat;
      void *fdm;
     
      if (argc != 1)
        return 1;
     
      if (fstat(STDIN_FILENO, &stat) == -1 ||
          stat.st_size == 0)
        return 2;
     
      fdm = mmap(0, stat.st_size, PROT_READ, MAP_SHARED, STDIN_FILENO, 0);
      if (fdm == MAP_FAILED)
        return 3;
     
      decode(fdm, stat.st_size);
     
      if (munmap(fdm, stat.st_size) == -1)
        return 4;
     
      return 0;
    }
     
    /*
     * This is a private message structure. A generic pointer to this structure
     * is passed to each of the callback functions. Put here any data you need
     * to access from within the callbacks.
     */
     
    struct buffer {
      unsigned char const *start;
      unsigned long length;
    };
     
    /*
     * This is the input callback. The purpose of this callback is to (re)fill
     * the stream buffer which is to be decoded. In this example, an entire file
     * has been mapped into memory, so we just call mad_stream_buffer() with the
     * address and length of the mapping. When this callback is called a second
     * time, we are finished decoding.
     */
     
    static
    enum mad_flow input(void *data,
                struct mad_stream *stream)
    {
      struct buffer *buffer = data;
     
      if (!buffer->length)
        return MAD_FLOW_STOP;
     
      mad_stream_buffer(stream, buffer->start, buffer->length);
     
      buffer->length = 0;
     
      return MAD_FLOW_CONTINUE;
    }
     
    /*
     * The following utility routine performs simple rounding, clipping, and
     * scaling of MAD's high-resolution samples down to 16 bits. It does not
     * perform any dithering or noise shaping, which would be recommended to
     * obtain any exceptional audio quality. It is therefore not recommended to
     * use this routine if high-quality output is desired.
     */
     
    static inline
    signed int scale(mad_fixed_t sample)
    {
      /* round */
      sample += (1L << (MAD_F_FRACBITS - 16));
     
      /* clip */
      if (sample >= MAD_F_ONE)
        sample = MAD_F_ONE - 1;
      else if (sample < -MAD_F_ONE)
        sample = -MAD_F_ONE;
     
      /* quantize */
      return sample >> (MAD_F_FRACBITS + 1 - 16);
    }
     
    /*
     * This is the output callback function. It is called after each frame of
     * MPEG audio data has been completely decoded. The purpose of this callback
     * is to output (or play) the decoded PCM audio.
     */
     
    static
    enum mad_flow output(void *data,
                 struct mad_header const *header,
                 struct mad_pcm *pcm)
    {
      unsigned int nchannels, nsamples;
      mad_fixed_t const *left_ch, *right_ch;
     
      /* pcm->samplerate contains the sampling frequency */
     
      nchannels = pcm->channels;
      nsamples  = pcm->length;
      left_ch   = pcm->samples[0];
      right_ch  = pcm->samples[1];
     
      while (nsamples--) {
        signed int sample;
     
        /* output sample(s) in 16-bit signed little-endian PCM */
     
        sample = scale(*left_ch++);
        putchar((sample >> 0) & 0xff);
        putchar((sample >> 8) & 0xff);
     
        if (nchannels == 2) {
          sample = scale(*right_ch++);
          putchar((sample >> 0) & 0xff);
          putchar((sample >> 8) & 0xff);
        }
      }
     
      return MAD_FLOW_CONTINUE;
    }
     
    /*
     * This is the error callback function. It is called whenever a decoding
     * error occurs. The error is indicated by stream->error; the list of
     * possible MAD_ERROR_* errors can be found in the mad.h (or stream.h)
     * header file.
     */
     
    static
    enum mad_flow error(void *data,
                struct mad_stream *stream,
                struct mad_frame *frame)
    {
      struct buffer *buffer = data;
     
      fprintf(stderr, "decoding error 0x%04x (%s) at byte offset %u\n",
          stream->error, mad_stream_errorstr(stream),
          stream->this_frame - buffer->start);
     
      /* return MAD_FLOW_BREAK here to stop decoding (and propagate an error) */
     
      return MAD_FLOW_CONTINUE;
    }
     
    /*
     * This is the function called by main() above to perform all the decoding.
     * It instantiates a decoder object and configures it with the input,
     * output, and error callback functions above. A single call to
     * mad_decoder_run() continues until a callback function returns
     * MAD_FLOW_STOP (to stop decoding) or MAD_FLOW_BREAK (to stop decoding and
     * signal an error).
     */
     
    static
    int decode(unsigned char const *start, unsigned long length)
    {
      struct buffer buffer;
      struct mad_decoder decoder;
      int result;
     
      /* initialize our private message structure */
     
      buffer.start  = start;
      buffer.length = length;
     
      /* configure input, output, and error functions */
     
      mad_decoder_init(&decoder, &buffer,
               input, 0 /* header */, 0 /* filter */, output,
               error, 0 /* message */);
     
      /* start decoding */
     
      result = mad_decoder_run(&decoder, MAD_DECODER_MODE_SYNC);
     
      /* release the decoder */
     
      mad_decoder_finish(&decoder);
     
      return result;
    }

[https_www.underbit.com_products_mad]: https://www.underbit.com/products/mad/
[20200313153652197.png]: /images/20230529/92a19dd79324415e8455e4c2afe0ae18.png
[https_gitlab.savoirfairelinux.com_netdsa_buildroot_commit_97b392b98468859ab2366995b6bee4140b38e388]: https://gitlab.savoirfairelinux.com/netdsa/buildroot/commit/97b392b98468859ab2366995b6bee4140b38e388
[Center]: /images/20230529/b89c5567df904d2c8acffd3b3f9a2bbc.png
[Center 1]: /images/20230529/e9338e84ba66442ba4a60887a3d03e6f.png