如何在安卓设备上解码M4A音频文件

Question

我正在尝试在Android上解码音频并获取原始数据以应用滤波器。

我正在使用MediaExtractor从文件中提取编码数据，这似乎有效。 然后我尝试混合来自 MediaExtractor文档和使用缓冲区进行同步处理的MediaCodec的代码，以分块方式提取数据并对其进行解码。

因此，我首先使用extractor.getTrackFormat(0);中获取的格式配置了解码器。

MediaExtractor extractor = new MediaExtractor();

String path = "...";
extractor.setDataSource(path);

MediaFormat format = extractor.getTrackFormat(0);
mAudioKeyMine = format.getString(MediaFormat.KEY_MIME);

extractor.selectTrack(0);

MediaCodec decoder;
decoder = MediaCodec.createDecoderByType(mAudioKeyMine);
decoder.configure(format, null, null, 0);

然后尝试获取数据：

public void getData(MediaExtractor extractor)
{
    int offset = 0;

    ByteBuffer inputBuffer = ByteBuffer.allocate(2048);

    MediaFormat outputFormat = decoder.getOutputFormat();
    Log.v(TAG, "outputFormat: " + outputFormat.toString());

    decoder.start();
    int index = decoder.dequeueInputBuffer(1000);

    boolean sawInputEOS = false;

    int sample = 0;
    while (sample >= 0)
    {

        int inputBufferId = decoder.dequeueInputBuffer(1000);
        if (inputBufferId >= 0)
        {
            inputBuffer = decoder.getInputBuffer(index);

            sample = extractor.readSampleData(inputBuffer, 0);

            long presentationTimeUs = 0;

            if (sample < 0)
            {
                sawInputEOS = true;
                sample = 0;
            }
            else
            {
                int trackIndex = extractor.getSampleTrackIndex();
                presentationTimeUs = extractor.getSampleTime();

                Log.v(TAG, "trackIndex: " + trackIndex + ", presentationTimeUs: " + presentationTimeUs);
                Log.v(TAG, "sample: " + sample + ", offset: " + offset);
                Log.v(TAG, "inputBuffer: " + inputBuffer.toString());
            }

            decoder.queueInputBuffer(inputBufferId, 0, sample, presentationTimeUs, sawInputEOS ? MediaCodec.BUFFER_FLAG_END_OF_STREAM : 0);

            if (!sawInputEOS)
            {
                extractor.advance();
            }

        }
        MediaCodec.BufferInfo info = new MediaCodec.BufferInfo();

        int outputBufferId = decoder.dequeueOutputBuffer(info, 1000);
        Log.v(TAG, "info: " + info.toString());

        if (outputBufferId >= 0)
        {
            ByteBuffer outputBuffer = decoder.getOutputBuffer(outputBufferId);
            MediaFormat bufferFormat = decoder.getOutputFormat(outputBufferId);

            Log.v(TAG, "option A");
            Log.v(TAG, "outputBufferId: " + outputBufferId);
            if (outputBuffer != null)
            {
                Log.v(TAG, "outputBuffer: " + outputBuffer.toString());
            }
            else
            {
                Log.v(TAG, "outputBuffer: null");
            }
            Log.v(TAG, "bufferFormat: " + bufferFormat.toString());

            if (outputBuffer != null)
            {
                int cont = 0;
                while (outputBuffer.hasRemaining())
                {
                    int pos = outputBuffer.position();
                    byte data = outputBuffer.get();

                    // do something with the data
                    if (cont < 10)
                    {
                        Log.v(TAG, "outputBuffer: " + pos + " -> " + data);
                    }
                    cont++;
                }
            }
            else
            {
                Log.v(TAG, "outputBuffer: null");
            }
            decoder.releaseOutputBuffer(outputBufferId, 0);
        }
        else if (outputBufferId == MediaCodec.INFO_OUTPUT_FORMAT_CHANGED)
        {
            Log.v(TAG, "option B");
            outputFormat = decoder.getOutputFormat(); 
            Log.v(TAG, "outputFormat: " + outputFormat.toString());
        }
        Log.v(TAG, "extractor.advance()");
        offset += sample;
    }
    Log.v(TAG, "end of track");
    extractor.release();
    extractor = null;
    decoder.stop();
    decoder.release();
}

但我在int outputBufferId = decoder.dequeueOutputBuffer(info, 1000);这一行遇到了一个IllegalStateException错误。

我搜索了这个错误以及如何正确解码m4a文件的方法，但大多数解决方案都已经过时了（仅适用于api 21），现在我被卡住了。

所以，有没有适用于api 26/28的音频解码示例，或者请有经验的人可以解释如何正确地进行解码？

整个项目托管在GitHub上。

- simone viozzi

---

看起来你正在出队输入缓冲区#1（index），然后出队输入缓冲区#2（inputBufferId），然后你将输入缓冲区#2入队。那么输入缓冲区#1发生了什么？你必须按照这些顺序使用它们！ - undefined

Answer 1

我使用回调函数以异步模式解决了这个问题。

基本的工作流程是：

• 使用MediaExtractor从文件中提取编码数据
• 将其传递给MediaCodec进行解码
• 将解码后的数据传递给AudioTrack进行播放（或者对数据进行其他处理）

首先，我们需要一些初始化。我把这些放在用于解码和播放文件的类的构造函数中：

// inizialize the mediaExtractor and set the source file
mediaExtractor = new MediaExtractor();
mediaExtractor.setDataSource(fileName);

// select the first audio track in the file and return it's format
mediaFormat = null;
int i;
int numTracks = mediaExtractor.getTrackCount();
for (i = 0; i < numTracks; i++)
{
    mediaFormat = mediaExtractor.getTrackFormat(i);
    if (mediaFormat.getString(MediaFormat.KEY_MIME).startsWith("audio/"))
    {
        mediaExtractor.selectTrack(i);
        break;
    }
}
// we get the parameter from the mediaFormat
channelCount = mediaFormat.getInteger(MediaFormat.KEY_CHANNEL_COUNT);
sampleRate = mediaFormat.getInteger(MediaFormat.KEY_SAMPLE_RATE);
duration = mediaFormat.getLong(MediaFormat.KEY_DURATION);
mimeType = mediaFormat.getString(MediaFormat.KEY_MIME);

// we can get the minimum buffer size from audioTrack passing the parameter of the audio
// to keep it safe it's good practice to create a buffer that is 8 times bigger
int minBuffSize = AudioTrack.getMinBufferSize(sampleRate,
                                              AudioFormat.CHANNEL_OUT_STEREO,
                                              AudioFormat.ENCODING_PCM_16BIT);

// to reproduce the data we need to initialize the audioTrack, by passing the audio parameter
// we use the MODE_STREAM so we can put more data dynamically with audioTrack.write()
audioTrack = new AudioTrack(AudioManager.STREAM_MUSIC,
                            sampleRate,
                            AudioFormat.CHANNEL_OUT_STEREO,
                            AudioFormat.ENCODING_PCM_16BIT,
                            minBuffSize * 8,
                            AudioTrack.MODE_STREAM);

对于开发者指南，我使用的初始化audioTrack的方法已被弃用，但它对我有效，而新方法不行，所以为了这个例子的目的，我保留了这种类型的初始化。

在初始化阶段后，我们需要创建解码器，为其设置回调，并启动解码器和audioTrack。媒体编解码器的回调如下：

• onInputBufferAvailable：当输入缓冲区可用时调用。
• onOutputBufferAvailable：当输出缓冲区可用时调用。
• onError：当媒体编解码器遇到错误时调用。
• onOutputFormatChanged：当输出格式更改时调用。

因此，我们需要：

• 使用提取器从文件中提取编码数据并将其用于填充我们从编解码器获取的输入缓冲区。
• 在编解码器完成解码后，我们可以从一个输出缓冲区获取解码数据并将其传递给audioTrack。

我的代码如下：

// we get the mediaCodec by creating it using the mime_type extracted form the track
MediaCodec decoder = MediaCodec.createDecoderByType(mimeType);

// to decode the file in asynchronous mode we set the callbacks
decoder.setCallback(new MediaCodec.Callback()
{
    private boolean mOutputEOS = false;
    private boolean mInputEOS = false;

    @Override
    public void onInputBufferAvailable (@NonNull MediaCodec codec,
                                        int index)
    {
        // if i reached the EOS i either the input or the output file i just skip
        if (mOutputEOS | mInputEOS) return;

        // i must use the index to get the right ByteBuffer from the codec
        ByteBuffer inputBuffer = codec.getInputBuffer(index);

        // if the codec is null i just skip and wait for another buffer
        if (inputBuffer == null) return;

        long sampleTime = 0;
        int result;

        // with this method i fill the inputBuffer with the data read from the mediaExtractor
        result = mediaExtractor.readSampleData(inputBuffer, 0);
        // the return parameter of readSampleData is the number of byte read from the file
        // and if it's -1 it means that i reached EOS
        if (result >= 0)
        {
            // if i read some bytes i can pass the index of the buffer, the number of bytes
            // that are in the buffer and the sampleTime to the codec, so that it can decode
            // that data
            sampleTime = mediaExtractor.getSampleTime();
            codec.queueInputBuffer(index, 0, result, sampleTime, 0);
            mediaExtractor.advance();
        }
        else
        {
            // if i reached EOS i need to tell the codec
            codec.queueInputBuffer(index, 0, 0, -1, MediaCodec.BUFFER_FLAG_END_OF_STREAM);
            mInputEOS = true;
        }
    }

    @Override
    public void onOutputBufferAvailable (@NonNull MediaCodec codec,
                                         int index,
                                         @NonNull MediaCodec.BufferInfo info)
    {
        // i can get the outputBuffer from the codec using the relative index
        ByteBuffer outputBuffer = codec.getOutputBuffer(index);

        // if i got a non null buffer
        if (outputBuffer != null)
        {
            outputBuffer.rewind();
            outputBuffer.order(ByteOrder.LITTLE_ENDIAN);

            // i just need to write the outputBuffer into the audioTrack passing the number of
            // bytes it contain and using the WRITE_BLOCKING so that this call will block
            // until it doesn't finish to write the data
            int ret = audioTrack.write(outputBuffer,
                                       outputBuffer.remaining(),
                                       AudioTrack.WRITE_BLOCKING);
        }

        // if the flags in the MediaCodec.BufferInfo contains the BUFFER_FLAG_END_OF_STREAM
        // it mean that i reached EOS so i set mOutputEOS to true, and to assure
        // that it remain true even if this callback is called again i use the logical or
        mOutputEOS |= ((info.flags & MediaCodec.BUFFER_FLAG_END_OF_STREAM) != 0);

        // i always need to release the buffer i use so the system can recycle them and use
        // it again
        codec.releaseOutputBuffer(index, false);

        // if i reached the end of the output stream i need to stop and release the codec
        // and the extractor
        if (mOutputEOS)
        {
            codec.stop();
            codec.release();
            mediaExtractor.release();
            audioTrack.release();
        }
    }

    @Override
    public void onError (@NonNull MediaCodec codec,
                         @NonNull MediaCodec.CodecException e)
    {
        Timber.e(e, "mediacodec collback onError: %s", e.getMessage());
    }

    @Override
    public void onOutputFormatChanged (@NonNull MediaCodec codec,
                                       @NonNull MediaFormat format)
    {
        Timber.d("onOutputFormatChanged: %s", format.toString());
    }

});
// now we can configure the codec by passing the mediaFormat and start it
decoder.configure(mediaFormat, null, null, 0);
decoder.start();
// also we need to start the audioTrack.
audioTrack.play();

- simone viozzi