使用MediaCodec+OpenSL编写简单的音频播放器
Posted blueberry_mu
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了使用MediaCodec+OpenSL编写简单的音频播放器相关的知识，希望对你有一定的参考价值。
使用MediaCodec+OpenSL编写简单的音频播放器
前言
通过MediaCodec Native API 和OpenSL编写一个简单的音频播放器。可以解码并播放一个mp3文件.
流程
初始化
- 使用AMediaExtractor解析Mp3文件，它可以得到音频文件的格式、以及帧（未解码）。
- 根据得到的音频信息（channelCount,channelMask,sampleRate)等可以创建出OpenSL AudioPlayer来播放解码后的音频。
- 创建根据解析得到的音频mime,创建合适的MediaCodec编码器
解码放入消费队列
- 使用
AMediaCodec_start开始解码。 - 使用
AMediaCodec_dequeueInputBuffer获取到一个空的buffer。 - 使用
AMediaExtractor_readSampleData获取到未解码的数据放入到buffer,读完之后需要使用AMediaExtractor_advance将指针前移，这样下一次再读的时候就是下一个帧数据了。 - 使用
AMediaCodec_queueInputBuffer获取一个输入buffer的index,使用index通过AMediaCodec_getInputBuffer获取输入buffer。 - 使用
AMediaCodec_dequeueOutputBuffer获取一个输出buffer的index,使用index通过AMediaCodec_getOutputBuffer获取对应的输出buffer。 - 将输出buffer的内容放入到一个消费队列，带openSL播放器消费。
读取消费队列传给openSL
- 将openSL的AudioPlayer状态设置为开始，并传入一个空的buffer.opensl播放器会进行回调来获取数据。
- 在回掉中读取消费队列的数据。
代码片段
初始化
JNIEXPORT jlong JNICALL
Java_com_blueberry_videoplayer_SLMediaCodecAudio_initialize(JNIEnv *env, jobject thiz,
jobject assetManager,
jstring res_path)
auto engine = new AudioPlayerEngine();
env->GetJavaVM(&engine->javaVm);
// 创建消费队列
engine->buffer_queue_ = new SafeQueue<MyBuffer>();
const char *fileName = env->GetStringUTFChars(res_path, NULL);
auto cAssetManager = AAssetManager_fromJava(env, assetManager);
engine->asset_ = AAssetManager_open(cAssetManager, fileName, AASSET_MODE_RANDOM);
off_t start = 0, len = 0;
auto fd = AAsset_openFileDescriptor(engine->asset_, &start, &len);
engine->mediaExtractor_ = AMediaExtractor_new();
AMediaExtractor_setDataSourceFd(engine->mediaExtractor_, fd, start, len);
AMediaExtractor_getTrackCount(engine->mediaExtractor_);
auto format = AMediaExtractor_getTrackFormat(engine->mediaExtractor_, 0);
const char *mime = nullptr;
AMediaFormat_getString(format, AMEDIAFORMAT_KEY_MIME, reinterpret_cast<const char **>(&mime));
AMediaFormat_getInt64(format, AMEDIAFORMAT_KEY_DURATION, &engine->metaData_.duration_);
AMediaFormat_getInt32(format, AMEDIAFORMAT_KEY_CHANNEL_COUNT, &engine->metaData_.channelCount_);
AMediaFormat_getInt32(format, AMEDIAFORMAT_KEY_SAMPLE_RATE, &engine->metaData_.sampleRate_);
AMediaFormat_getInt32(format, AMEDIAFORMAT_KEY_BIT_RATE, &engine->metaData_.bitrate_);
const char *formatStr = AMediaFormat_toString(format);
LOGD("the file format is %s", formatStr);
AMediaExtractor_selectTrack(engine->mediaExtractor_, 0);
// 创建解码器，并配置
engine->codec_ = AMediaCodec_createDecoderByType(mime);
AMediaCodec_configure(engine->codec_, format, nullptr, nullptr, 0);
// 创建openSL引擎
SLresult result = slCreateEngine(&engine->slEngineObj_, 0, nullptr, 0, nullptr, nullptr);
result = (*engine->slEngineObj_)->Realize(engine->slEngineObj_, false);
result = (*engine->slEngineObj_)->GetInterface(engine->slEngineObj_, SL_IID_ENGINE,
&engine->slEngineItf_);
// 根据音频格式信息，构建音频播放器的输入格式
SLandroidDataFormat_PCM_EX_ sourceFormat
.formatType = SL_DATAFORMAT_PCM,
.numChannels = static_cast<SLuint32>(engine->metaData_.channelCount_),
// millherzt
.sampleRate = static_cast<SLuint32>(engine->metaData_.sampleRate_) * 1000,
.bitsPerSample =SL_PCMSAMPLEFORMAT_FIXED_16,
.containerSize = SL_PCMSAMPLEFORMAT_FIXED_16,
.channelMask = SL_SPEAKER_FRONT_LEFT | SL_SPEAKER_FRONT_RIGHT,
.endianness = SL_BYTEORDER_LITTLEENDIAN,
.representation = SL_ANDROID_PCM_REPRESENTATION_SIGNED_INT
;
engine->silent_buf_ = new uint8_t[2 * 2 * 80]; // 80 frame per buf.
engine->silent_buf_size_ = 2 * 2 * 80;
// 输入数据使用内存队列
SLDataLocator_AndroidSimpleBufferQueue androidSimpleBufferQueueLocator
.locatorType = SL_DATALOCATOR_ANDROIDSIMPLEBUFFERQUEUE,
.numBuffers = 4
;
SLDataSource slDataSource
.pLocator = &androidSimpleBufferQueueLocator,
.pFormat = &sourceFormat,
;
SLObjectItf outputMixObj;
result = (*engine->slEngineItf_)->CreateOutputMix(engine->slEngineItf_,
&outputMixObj,
0,
nullptr,
nullptr);
result = (*outputMixObj)->Realize(outputMixObj, false);
SLDataLocator_OutputMix outputMixLocator
.locatorType = SL_DATALOCATOR_OUTPUTMIX,
.outputMix = outputMixObj,
;
SLDataSink slDataSink
.pLocator = &outputMixLocator,
.pFormat = nullptr,
;
SLInterfaceID interfaces[2]
SL_IID_VOLUME,
SL_IID_BUFFERQUEUE
;
SLboolean isRequired[2]
SL_BOOLEAN_TRUE,
SL_BOOLEAN_TRUE
;
(*engine->slEngineItf_)->CreateAudioPlayer(engine->slEngineItf_,
&engine->slPlayerObj_,
&slDataSource, &slDataSink,
sizeof(interfaces) /
sizeof(interfaces[0]),
interfaces, isRequired);
(*engine->slPlayerObj_)->Realize(engine->slPlayerObj_, SL_BOOLEAN_FALSE);
(*engine->slPlayerObj_)->GetInterface(engine->slPlayerObj_, SL_IID_PLAY,
&engine->slPlayItf_);
(*engine->slPlayerObj_)->GetInterface(engine->slPlayerObj_,
SL_IID_ANDROIDSIMPLEBUFFERQUEUE,
&engine->androidSimpleBufferQueueItf_);
result = (*engine->androidSimpleBufferQueueItf_)->RegisterCallback(
engine->androidSimpleBufferQueueItf_, bufferQueueCallback, engine);
assert(result == SL_RESULT_SUCCESS);
result = (*engine->slPlayItf_)->SetPlayState(engine->slPlayItf_, SL_PLAYSTATE_STOPPED);
// 创建解码线程
pthread_create(&engine->thread_, nullptr, threadRun, engine);
env->ReleaseStringUTFChars(res_path, fileName);
return reinterpret_cast<long>(engine);
解码
解码的逻辑配上这张图会更好理解一些。以解码为例：我会通过调用AMediaCodec_dequeueInputBuffer向编码器获取到空的buffer.然后放入我们的数据，再重新塞给解码器。该过程累死一个传送带，它由上层api驱而转动。获取解码后的数据时，我们通过
AMediaCodec_dequeueOutputBuffer来获取一个装有解码数据的buffer.将数据copy出后，在调用AMediaCodec_releaseOutputBuffer将buffer释放。
if (engine->status_ == Status::STARTED)
// 开始解码
AMediaCodec_start(engine->codec_);
while (engine->status_ == STARTED)
if (!engine->inputEnd)
// 读区一个输入buffer
auto index = AMediaCodec_dequeueInputBuffer(engine->codec_, TIMEOUT_US);
if (index >= 0)
size_t output_size = 0;
auto buffer = AMediaCodec_getInputBuffer(engine->codec_, index,
&output_size);
// 将为解码输入放入buffer
auto sampleSize = AMediaExtractor_readSampleData(engine->mediaExtractor_,
buffer,
output_size);
if (sampleSize <= 0)
sampleSize = 0;
engine->inputEnd = true;
auto presentationTimeUs = AMediaExtractor_getSampleTime(
engine->mediaExtractor_);
// 将buffer放会解码器
auto status = AMediaCodec_queueInputBuffer(engine->codec_,
index,
0,
sampleSize,
presentationTimeUs,
engine->inputEnd
? AMEDIACODEC_BUFFER_FLAG_END_OF_STREAM
: 0);
if (status != AMEDIA_OK)
LOGD("status:%d", status);
break;
// 前移，以便下次AMediaExtractor_readSampleData将获取到新的数据
AMediaExtractor_advance(engine->mediaExtractor_);
// decode
while (!engine->outputEnd)
AMediaCodecBufferInfo bufferInfo;
// 获取一个输出buffer index
auto index = AMediaCodec_dequeueOutputBuffer(engine->codec_, &bufferInfo,
TIMEOUT_US);
if (index >= 0)
LOGD("process output index >=0,%d", index);
if (bufferInfo.flags & AMEDIACODEC_BUFFER_FLAG_END_OF_STREAM)
engine->outputEnd = true;
size_t buffer_size = 0;
// 获取输出的buffer
auto buffer = AMediaCodec_getOutputBuffer(engine->codec_, index,
&buffer_size);
if (bufferInfo.size > 0)
auto buf = new uint8_t[bufferInfo.size];
memcpy(buf, &buffer[bufferInfo.offset], bufferInfo.size);
MyBuffer myBuffer;
myBuffer.len_ = bufferInfo.size;
myBuffer.buf_ = buf;
// 将解码后的数据放入到消费队列
engine->buffer_queue_->enqueue(myBuffer);
// 释放输出buffer
AMediaCodec_releaseOutputBuffer(engine->codec_, index, false);
LOGD("bufferQueueCallback:index >=0 ");
break;
else
switch (index)
case AMEDIACODEC_INFO_OUTPUT_FORMAT_CHANGED :
auto format = AMediaCodec_getOutputFormat(engine->codec_);
const char *strFormat = AMediaFormat_toString(format);
const char *pcmEncoding = nullptr;
AMediaFormat_delete(format);
break;
case AMEDIACODEC_INFO_OUTPUT_BUFFERS_CHANGED:
break;
case AMEDIACODEC_INFO_TRY_AGAIN_LATER:
default:
break;
openSL消费队列
void bufferQueueCallback(SLAndroidSimpleBufferQueueItf androidSimpleBufferQueueItf, void *context)
auto engine = reinterpret_cast<AudioPlayerEngine * >(context);
auto queue = engine->buffer_queue_;
auto myBuffer = queue->dequeue();
if (myBuffer.buf_ != nullptr)
// 获取到消费队列中解码后的数据，放入opensl的内存队列中
(*engine->androidSimpleBufferQueueItf_)->Enqueue(
engine->androidSimpleBufferQueueItf_,
myBuffer.buf_,
myBuffer.len_);
else
(*engine->androidSimpleBufferQueueItf_)->Enqueue(
engine->androidSimpleBufferQueueItf_,
engine->silent_buf_,
1);
return;
消费队列的定义
#include <queue>
#include <mutex>
#include <condition_variable>
template <class T>
class SafeQueue
private:
std::queue<T> q;
mutable std::mutex m;
std::condition_variable c;
public:
SafeQueue(/* args */) : m(), q(), c()
~SafeQueue()
void enqueue(T t)
std::lock_guard<std::mutex> lock(m);
q.push(t);
c.notify_all();
T dequeue(void)
std::unique_lock<std::mutex> lock(m);
c.wait(lock, [this]
return !q.empty(); );
T val = q.front();
q.pop();
return val;
;
参考
https://developer.android.com/ndk/reference/group/media
https://developer.android.com/reference/android/media/MediaCodec#queueInputBuffer(int,%20int,%20int,%20long,%20int)
以上是关于使用MediaCodec+OpenSL编写简单的音频播放器的主要内容，如果未能解决你的问题，请参考以下文章