I'm trying to write a basic raw AAC data to a file, in hopes that I can use mp4parser to encapsulate it with a video track. For that, I need to encode any given audio file to that format. MediaCodec API is readily available since API 16, so I've decided to use that for the codec operation.
I'm not sure why not many resources are available online regarding this, possibly due to the complexity associated. Although, I've managed to learn that the fundamental approach should be:
Get sample data via MediaExtractor -> Enqueue decoder input buffer -> Dequeue output buffer and get the decoded data -> Enqueue encoder input buffer -> Dequeue encoder output buffer -> Write the encoded data to file.
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();
}
The output file is not valid for some reason. Why?
EDIT: Managed to fix an Exception, issue persists.
EDIT 2: I've prevented the buffer overflow by setting the buffer size to the bitrate in the encoder format settings. There are two issues currently:
1. After a very short interval, it gets stuck here, possibly waiting indefinitely.int encInputBufferIndex = dequeueInputBuffer(-1);
2. Decoding takes as long as the track is, why does this regard for the actual interval of samples?
EDIT 3: Testing with AudioTrack.write(), audio plays nice and fine, but this isn't intended and suggests that the decoding is taking place in sync to the media file being fed, this should take place as fast as possible to allow encoder to do its job quick. Changing the presentationTimeUs in decoder.queueInputBuffer() did nothing.
