java - 无法通过 Android MediaCodec API 转码音频

Question

我正在尝试将基本的原始 AAC 数据写入文件，希望我可以使用 mp4parser 将其封装为视频轨道。为此，我需要将任何给定的音频文件编码为该格式。 MediaCodec API 从 API 16 开始就可用，因此我决定将其用于编解码器操作。

我不确定为什么网上没有很多关于此的资源，可能是因为相关的复杂性。虽然，我已经设法了解到基本方法应该是:

通过MediaExtractor获取样本数据->入队解码器输入缓冲区->出队输出缓冲区并获取解码数据->入队编码器输入缓冲区->出队编码器输出缓冲区->将编码数据写入文件。

private void transcodeFile(File source, File destination) throws IOException {
    FileInputStream inputStream = new FileInputStream(source);
    FileOutputStream outputStream = new FileOutputStream(destination);

    log("Transcoding file: " + source.getName());

    MediaExtractor extractor;
    MediaCodec encoder;
    MediaCodec decoder;

    ByteBuffer[] encoderInputBuffers;
    ByteBuffer[] encoderOutputBuffers;
    ByteBuffer[] decoderInputBuffers;
    ByteBuffer[] decoderOutputBuffers;

    int noOutputCounter = 0;
    int noOutputCounterLimit = 10;

    extractor = new MediaExtractor();
    extractor.setDataSource(inputStream.getFD());
    extractor.selectTrack(0);

    log(String.format("TRACKS #: %d", extractor.getTrackCount()));
    MediaFormat format = extractor.getTrackFormat(0);
    String mime = format.getString(MediaFormat.KEY_MIME);
    log(String.format("MIME TYPE: %s", mime));


    final String outputType = MediaFormat.MIMETYPE_AUDIO_AAC;
    encoder = MediaCodec.createEncoderByType(outputType);
    MediaFormat encFormat = MediaFormat.createAudioFormat(outputType, 44100, 2);
    encFormat.setInteger(MediaFormat.KEY_BIT_RATE, 64000);
    encoder.configure(encFormat, null, null, MediaCodec.CONFIGURE_FLAG_ENCODE);

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

    encoder.start();
    decoder.start();

    encoderInputBuffers = encoder.getInputBuffers();
    encoderOutputBuffers = encoder.getOutputBuffers();

    decoderInputBuffers = decoder.getInputBuffers();
    decoderOutputBuffers = decoder.getOutputBuffers();

    int timeOutUs = 1000;
    long presentationTimeUs = 0;

    MediaCodec.BufferInfo info = new MediaCodec.BufferInfo();
    boolean inputEOS = false;
    boolean outputEOS = false;

    while(!outputEOS && noOutputCounter < noOutputCounterLimit) {
        noOutputCounter++;

        if(!inputEOS) {
            int decInputBufferIndex = decoder.dequeueInputBuffer(timeOutUs);
            log("decInputBufferIndex: " + decInputBufferIndex);
            if (decInputBufferIndex >= 0) {
                ByteBuffer dstBuffer = decoderInputBuffers[decInputBufferIndex];

                //Getting sample with MediaExtractor
                int sampleSize = extractor.readSampleData(dstBuffer, 0);
                if (sampleSize < 0) {
                    inputEOS = true;
                    log("Input EOS");
                    sampleSize = 0;
                } else {
                    presentationTimeUs = extractor.getSampleTime();
                }

                log("Input sample size: " + sampleSize);

                //Enqueue decoder input buffer
                decoder.queueInputBuffer(decInputBufferIndex, 0, sampleSize, presentationTimeUs, inputEOS ? MediaCodec.BUFFER_FLAG_END_OF_STREAM : 0);
                if (!inputEOS) extractor.advance();

            } else {
                log("decInputBufferIndex: " + decInputBufferIndex);
            }
        }

        //Dequeue decoder output buffer
        int res = decoder.dequeueOutputBuffer(info, timeOutUs);
        if(res >= 0) {
            if(info.size > 0) noOutputCounter = 0;

            int decOutputBufferIndex = res;
            log("decOutputBufferIndex: " + decOutputBufferIndex);

            ByteBuffer buffer = decoderOutputBuffers[decOutputBufferIndex];
            buffer.position(info.offset);
            buffer.limit(info.offset + info.size);

            final int size = buffer.limit();
            if(size > 0) {
                //audioTrack.write(buffer, buffer.limit(), AudioTrack.MODE_STATIC);

                int encInputBufferIndex = encoder.dequeueInputBuffer(-1);
                log("encInputBufferIndex: " + encInputBufferIndex);
                //fill the input buffer with the decoded data
                if(encInputBufferIndex >= 0) {
                    ByteBuffer dstBuffer = encoderInputBuffers[encInputBufferIndex];
                    dstBuffer.clear();
                    dstBuffer.put(buffer);

                    encoder.queueInputBuffer(encInputBufferIndex, 0, info.size, info.presentationTimeUs, 0);
                    int encOutputBufferIndex = encoder.dequeueOutputBuffer(info, timeOutUs);
                    if(encOutputBufferIndex >= 0) {
                        log("encOutputBufferIndex: " + encOutputBufferIndex);
                        ByteBuffer outBuffer = encoderOutputBuffers[encOutputBufferIndex];
                        byte[] out = new byte[outBuffer.remaining()];
                        outBuffer.get(out);
                        //write data to file
                        outputStream.write(out);
                    }
                }
            }
            decoder.releaseOutputBuffer(decOutputBufferIndex, false);
            if((info.flags & MediaCodec.BUFFER_FLAG_END_OF_STREAM) != 0) {
                outputEOS = true;
                log("Output EOS");
            }
        } else if (res == MediaCodec.INFO_OUTPUT_BUFFERS_CHANGED) {
            decoderOutputBuffers = decoder.getOutputBuffers();
            log("Output buffers changed.");
        } else if (res == MediaCodec.INFO_OUTPUT_FORMAT_CHANGED) {
            log("Output format changed.");
        } else {
            log("Dequeued output buffer returned: " + res);
        }
    }

    log("Stopping..");
    releaseCodec(decoder);
    releaseCodec(encoder);
    inputStream.close();
    outputStream.close();

}

输出文件由于某种原因无效。为什么？

编辑:设法修复异常，问题仍然存在。

编辑 2:我通过在编码器格式设置中将缓冲区大小设置为比特率来防止缓冲区溢出。目前有两个问题: 1. 在很短的时间间隔后，它会卡在这里，可能会无限期地等待。int encInputBufferIndex = dequeueInputBuffer(-1); 2. 轨道有多长解码就有多长，为什么要考虑采样的实际间隔？

编辑 3:使用 AudioTrack.write() 进行测试，音频播放效果很好，但这不是故意的，并且表明解码与正在馈送的媒体文件同步进行，这应该尽可能快地进行可以让编码器快速完成它的工作。在 decoder.queueInputBuffer() 中更改 presentationTimeUs 没有任何作用。

Answer 1

您的方法是正确的，缺少的部分是使用 MediaMuxer 将编码帧混合到有效的 MP4 文件中.在 bigflake 上有一个很好的(也是唯一的)例子.这件事最相关的例子是

• EncodeAndMuxTest.java
• DecodeEditEncodeTest.java

您必须合并并简化/修改它们才能使用音频而不是视频。上述你需要 API 18

编辑:我如何将解码器缓冲区转发到编码器(或多或少)。到目前为止，我还没有遇到缓冲区溢出，只是希望理智的实现将具有相同容量的编码器和解码器缓冲区:

int decoderStatus = audioDecoder.dequeueOutputBuffer(info, TIMEOUT_USEC);
  if (decoderStatus >= 0) {
      // no output available yet
      if (VERBOSE) Log.d(TAG, "no output from audio decoder available");
...
   } else if (decoderStatus == MediaCodec.INFO_OUTPUT_BUFFERS_CHANGED) {
            audioDecoderOutputBuffers = audioDecoder.getOutputBuffers();
            if (VERBOSE) Log.d(TAG, "decoder output buffers changed (we don't care)");
    } else if (decoderStatus == MediaCodec.INFO_OUTPUT_FORMAT_CHANGED) {
            // expected before first buffer of data
            if (VERBOSE) {
                    MediaFormat newFormat = audioDecoder.getOutputFormat();
                    Log.d(TAG, "decoder output format changed: " + newFormat);
                }
    } else if (decoderStatus < 0) {
            Log.e(TAG, "unexpected result from decoder.dequeueOutputBuffer: "+decoderStatus);
            throw new RuntimeException("Issue with dencoding audio");
    } else { // decoderStatus >= 0
            if (VERBOSE) Log.d(TAG, "audio decoder produced buffer "
                                + decoderStatus + " (size=" + info.size + ")");

            if (info.size! = 0) {                           
                // Forward decoder buffer to encoder
                ByteBuffer decodedData = audioDecoderOutputBuffers[decoderStatus];
                decodedData.position(info.offset);
                decodedData.limit(info.offset + info.size);

                 // Possibly edit buffer data

                // Send it to the audio encoder.
                int encoderStatus = audioEncoder.dequeueInputBuffer(-1);
                if (encoderStatus < 0) {
                    throw new RuntimeException("Could not get input buffer for audio encoder!!!");
                }
            audioEncoderInputBuffers[encoderStatus].clear();
            audioEncoderInputBuffers[encoderStatus].put(decodedData);
         }
audioEncoder.queueInputBuffer(encoderStatus, 0, info.size, mAudioMediaTime, 0);
     if (VERBOSE) Log.d(TAG, "Submitted to AUDIO encoder frame, size=" + info.size + " time=" + mAudioMediaTime);
    }
 audioDecoder.releaseOutputBuffer(decoderStatus, false);