无法通过Android MediaCodec API转码音频

Question

我正在尝试写一个基本的原始AAC数据到一个文件，希望我可以使用mp4parser封装它与一个视频轨道。为此，我需要将任何给定的音频文件编码为该格式。MediaCodec API从API16开始就很容易获得，所以我决定使用它来进行编解码操作。

我不确定为什么没有很多资源可以在线上关于这一点，可能是由于相关的复杂性。不过，我已经了解到基本的方法应该是：

通过MediaExtractor获取样本数据->编码器输入缓冲区入列->输出缓冲区出列并获取解码数据->编码器输入缓冲区入列->编码器输出缓冲区出列->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();

}

编辑：设法修复了一个异常，问题仍然存在。

编辑2：我已经通过将缓冲区大小设置为编码器格式设置中的比特率来防止缓冲区溢出。目前存在两个问题：1。在很短的间隔之后，它就会卡在这里，可能会无限期地等待。int encInputBufferIndex=dequeueInputBuffer(-1)；2。译码的时间和轨道一样长，为什么这会考虑样本的实际间隔呢？

编辑3：使用audiotrack.write（）进行测试，音频播放得很好，但这并不是有意的，它暗示解码是与被馈送的媒体文件同步进行的，这应该尽可能快地进行，以允许编码器快速地完成它的工作。更改Decoder.QueueInputBuffer（）中的presentationTimeUs没有执行任何操作。

Answer 1

你是在正确的方式，缺少的部分是复用编码帧到有效的MP4文件与MediaMuxer。在Bigflake上有一个很好的（也是唯一的）例子。这方面最相关的例子有

• EncodeAndMuxTest.java
• DecodeEditEncodeTest.java

您必须组合、简化/修改它们，以便使用音频而不是视频。您需要API 18来实现上述功能

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);