使用OpenAL和FFMPEG解码并播放音乐
使用OpenAL和FFMPEG解码并播放音乐
OpenAL是一个开源的音效库，然而这里只用它来播放音乐。
FFMPEG负责把音乐解码并转换成指定的编码和采样率，然后送到OpenAL中播放。
（已在windows和ios平台简单测试通过）
AudioDecoder.h
#ifndef __AUDIODECODER_H__
#define __AUDIODECODER_H__
extern "C"
{
#include m_pFormatCtx;
AVCodecContext* m_pAudioCodecCtx;
AVCodec* m_pAudioCodec;
int m_nAudioIndex;
std::string m_strPath;
std::queue AudioDecoder.cpp
#include "stdafx.h"
#include "AudioDecoder.h"
#define CPP_TIME_BASE_Q (AVRational{1, AV_TIME_BASE})
AudioDecoder::AudioDecoder(int outSamplerate, int outChannels)
{
m_pmtx = nullptr;
m_pAudioCodec = nullptr;
m_pFormatCtx = nullptr;
m_pAudioCodecCtx = nullptr;
static bool bFFMPEGInit = false;
if (!bFFMPEGInit)
{
av_register_all();
avcodec_register_all();
bFFMPEGInit = true;
}
m_outSampleRate = outSamplerate;
m_outChs = outChannels;
m_wholeDuration = -1;
m_ch = -1;
m_strPath = "";
m_bStop = true;
m_bSeeked = false;
}
AudioDecoder::~AudioDecoder()
{
}
int AudioDecoder::OpenFile(std::string path)
{
if (!m_bStop)
{
StopDecode();
}
m_strPath = path;
AVFormatContext* pFormatCtx = nullptr;
if (0 != avformat_open_input( &pFormatCtx, path.c_str(), NULL, NULL))
{
return -1;
}
m_pFormatCtx = std::shared_ptr(pFormatCtx, [](AVFormatContext* pf) { if (pf) avformat_close_input(&pf); });
if (avformat_find_stream_info(m_pFormatCtx.get(), NULL) < 0)
{
return -1;
}
for (int i = 0; i < m_pFormatCtx->nb_streams; i++)
{
if (m_pFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
{
m_nAudioIndex = i;
break;
}
}
if (m_nAudioIndex == -1)
{
return -1;
}
m_pAudioCodecCtx = m_pFormatCtx->streams[m_nAudioIndex]->codec;
m_pAudioCodec = avcodec_find_decoder(m_pAudioCodecCtx->codec_id); //
if (m_pAudioCodec < 0)
{
return -1;
}
if (avcodec_open2(m_pAudioCodecCtx, m_pAudioCodec, NULL) < 0)
{
return -1;
}
m_ch = m_pAudioCodecCtx->channels;
m_wholeDuration = av_rescale_q(m_pFormatCtx->streams[m_nAudioIndex]->duration,
m_pFormatCtx->streams[m_nAudioIndex]->time_base,
CPP_TIME_BASE_Q);
return 0;
}
int AudioDecoder::StartDecode()
{
m_bStop = false;
m_pmtx = std::move(std::make_shared<std::mutex>());
m_pcond = std::move(std::make_shared<std::condition_variable>());
m_pDecode = std::move(std::make_shared<std::thread>(&AudioDecoder::DecodeThread, this));
return 0;
}
int AudioDecoder::StopDecode()
{
m_bStop = true;
m_pcond->notify_all();
if(m_pDecode->joinable())
m_pDecode->join();
std::unique_lock<std::mutex> lck(*m_pmtx);
int queueSize = m_queueData.size();
for (int i = queueSize - 1; i >= 0; i--)
{
PTFRAME f = m_queueData.front();
m_queueData.pop();
if (f)
{
if (f->data)
av_free(f->data);
delete f;
}
}
return 0;
}
PTFRAME AudioDecoder::GetFrame()
{
PTFRAME frame = nullptr;
std::unique_lock<std::mutex> lck(*m_pmtx);
m_pcond->wait(lck,
[this]() {return m_bStop || m_queueData.size() > 0; });
if (!m_bStop)
{
frame = m_queueData.front();
m_queueData.pop();
}
lck.unlock();
m_pcond->notify_one();
return frame;
}
int AudioDecoder::DecodeThread()
{
if (m_strPath == "")
return -1;
m_bDecoding = true;
AVPacket *packet = (AVPacket *)av_malloc(sizeof(AVPacket));
av_init_packet(packet);
std::shared_ptr pAudioFrame(av_frame_alloc(), [](AVFrame* pFrame) {av_frame_free(&pFrame);});
int64_t in_channel_layout = av_get_default_channel_layout(m_pAudioCodecCtx->channels);
struct SwrContext* au_convert_ctx = swr_alloc();
int64_t outLayout;
if (m_outChs == 1)
{
outLayout = AV_CH_LAYOUT_MONO;
}
else
{
outLayout = AV_CH_LAYOUT_STEREO;
}
au_convert_ctx = swr_alloc_set_opts(au_convert_ctx,
outLayout, AV_SAMPLE_FMT_S16,
m_outSampleRate, in_channel_layout,
m_pAudioCodecCtx->sample_fmt, m_pAudioCodecCtx->sample_rate, 0,
NULL);
swr_init(au_convert_ctx);
while (true)
{
int ret = av_read_frame(m_pFormatCtx.get(), packet);
if (ret < 0)
break;
if (packet->stream_index == m_nAudioIndex)
{
int nAudioFinished = 0;
int nRet = avcodec_decode_audio4(m_pAudioCodecCtx, pAudioFrame.get(),
&nAudioFinished, packet);
if (nRet > 0 && nAudioFinished != 0)
{
PTFRAME frame = new TFRAME;
frame->chs = m_outChs;
frame->samplerate = m_outSampleRate;
frame->duration = av_rescale_q(packet->duration, m_pFormatCtx->streams[m_nAudioIndex]->time_base, CPP_TIME_BASE_Q);
frame->pts = av_rescale_q(packet->pts, m_pFormatCtx->streams[m_nAudioIndex]->time_base, CPP_TIME_BASE_Q);
//resample
int outSizeCandidate = m_outSampleRate * 8 *
double(frame->duration) / 1000000.0;
uint8_t* convertData = (uint8_t*)av_malloc(sizeof(uint8_t) * outSizeCandidate);
int out_samples = swr_convert(au_convert_ctx,
&convertData, outSizeCandidate,
(const uint8_t**)&pAudioFrame->data[0], pAudioFrame->nb_samples);
int Audiobuffer_size = av_samples_get_buffer_size(NULL,
m_outChs, out_samples,AV_SAMPLE_FMT_S16,1);
frame->data = convertData;
frame->size = Audiobuffer_size;
std::unique_lock<std::mutex> lck(*m_pmtx);
m_pcond->wait(lck,
[this]() {return m_bStop || m_queueData.size() < MAX_BUFF_SIZE; });
if (m_bStop)
{
av_free_packet(packet);
break;
}
if (m_bSeeked)
{
m_bSeeked = false;
av_free_packet(packet);
continue;
}
m_queueData.push(frame);
lck.unlock();
m_pcond->notify_one();
}
}
av_free_packet(packet);
}
swr_free(&au_convert_ctx);
return 0;
}
//for seek
int AudioDecoder::InternalAudioSeek(uint64_t start_time_in_us)
{
if (start_time_in_us < 0)
{
return -1;
}
int64_t seek_pos = start_time_in_us;
if (m_pFormatCtx->start_time != AV_NOPTS_VALUE)
seek_pos += m_pFormatCtx->start_time;
if (av_seek_frame(m_pFormatCtx.get(), -1, seek_pos, AVSEEK_FLAG_BACKWARD) < 0)
{
return -2;
}
avcodec_flush_buffers(m_pFormatCtx->streams[m_nAudioIndex]->codec);
return 0;
}
//设置seek
void AudioDecoder::SetSeekPosition(double playProcess)
{
uint64_t pos = (uint64_t)(playProcess * (double)m_wholeDuration);
std::unique_lock<std::mutex> lck(*m_pmtx);
m_bSeeked = true;
InternalAudioSeek(pos);
int queueSize = m_queueData.size();
for (int i = queueSize - 1; i >= 0; i--)
{
PTFRAME f = m_queueData.front();
m_queueData.pop();
if (f)
{
if (f->data)
av_free(f->data);
delete f;
}
}
m_pcond->notify_all();//如果是从wait里面拿的锁，就要手动去激活他们
return;
} ALEngine.h
#ifndef __ALENGINE_H__
#define __ALENGINE_H__
#define NUMBUFFERS (4)
#define SERVICE_UPDATE_PERIOD (20)
#include m_decoder;
std::unique_ptr m_ptPlaying;
std::atomic_bool m_bStop;
ALuint m_buffers[NUMBUFFERS];
ALuint m_bufferTemp;
private:
int SoundPlayingThread();
int SoundCallback(ALuint& bufferID);
int InitEngine();
int DestroyEngine();
};
#endif ALEngine.cpp
#include "stdafx.h"
#include "ALEngine.h"
ALEngine::ALEngine()
{
InitEngine();
m_bStop = true;
}
ALEngine::~ALEngine()
{
DestroyEngine();
}
int ALEngine::InitEngine()
{
ALCdevice* pDevice;
ALCcontext* pContext;
pDevice = alcOpenDevice(NULL);
pContext = alcCreateContext(pDevice, NULL);
alcMakeContextCurrent(pContext);
if (alcGetError(pDevice) != ALC_NO_ERROR)
return AL_FALSE;
return 0;
}
int ALEngine::DestroyEngine()
{
ALCcontext* pCurContext;
ALCdevice* pCurDevice;
pCurContext = alcGetCurrentContext();
pCurDevice = alcGetContextsDevice(pCurContext);
alcMakeContextCurrent(NULL);
alcDestroyContext(pCurContext);
alcCloseDevice(pCurDevice);
return 0;
}
int ALEngine::OpenFile(std::string path)
{
if (!m_bStop)
{
Stop();
}
m_bStop = false;
alGenSources(1, &m_source);
if (alGetError() != AL_NO_ERROR)
{
printf("Error generating audio source.");
return -1;
}
ALfloat SourcePos[] = { 0.0, 0.0, 0.0 };
ALfloat SourceVel[] = { 0.0, 0.0, 0.0 };
ALfloat ListenerPos[] = { 0.0, 0, 0 };
ALfloat ListenerVel[] = { 0.0, 0.0, 0.0 };
// first 3 elements are "at", second 3 are "up"
ALfloat ListenerOri[] = { 0.0, 0.0, -1.0, 0.0, 1.0, 0.0 };
alSourcef(m_source, AL_PITCH, 1.0);
alSourcef(m_source, AL_GAIN, 1.0);
alSourcefv(m_source, AL_POSITION, SourcePos);
alSourcefv(m_source, AL_VELOCITY, SourceVel);
alSourcef(m_source, AL_REFERENCE_DISTANCE, 50.0f);
alSourcei(m_source, AL_LOOPING, AL_FALSE);
alDistanceModel(AL_LINEAR_DISTANCE_CLAMPED);
alListener3f(AL_POSITION, 0, 0, 0);
m_decoder = std::move(std::make_unique());
//m_decoder.reset(new AudioDecoder()); //for ios
m_decoder->OpenFile(path);
m_decoder->StartDecode();
alGenBuffers(NUMBUFFERS, m_buffers);
//m_ptPlaying = std::move(std::make_unique(&ALEngine::SoundPlayingThread, this));
m_ptPlaying.reset(new std::thread(&ALEngine::SoundPlayingThread, this)); //for ios
return 0;
}
int ALEngine::Play()
{
int state;
alGetSourcei(m_source, AL_SOURCE_STATE, &state);
if (state == AL_STOPPED || state == AL_INITIAL)
{
alSourcePlay(m_source);
}
return 0;
}
int ALEngine::PausePlay()
{
int state;
alGetSourcei(m_source, AL_SOURCE_STATE, &state);
if (state == AL_PAUSED)
{
alSourcePlay(m_source);
}
return 0;
}
int ALEngine::Pause()
{
alSourcePause(m_source);
return 0;
}
int ALEngine::Stop()
{
if(m_bStop)
return 0;
m_bStop = true;
alSourceStop(m_source);
alSourcei(m_source, AL_BUFFER, 0);
m_decoder->StopDecode();//要先把decoder stop,否则可能hang住 播放线程
if(m_ptPlaying->joinable())
m_ptPlaying->join();
//
alDeleteBuffers(NUMBUFFERS, m_buffers);
alDeleteSources(1, &m_source);
return 0;
}
int ALEngine::SoundPlayingThread()
{
//get frame
for (int i = 0; i < NUMBUFFERS; i++)
{
SoundCallback(m_buffers[i]);
}
Play();
//
while (true)
{
if (m_bStop)
break;
std::this_thread::sleep_for(std::chrono::milliseconds(SERVICE_UPDATE_PERIOD));
ALint processed = 0;
alGetSourcei(m_source, AL_BUFFERS_PROCESSED, &processed);
//printf("the processed is:%d\n", processed);
while (processed > 0)
{
ALuint bufferID = 0;
alSourceUnqueueBuffers(m_source, 1, &bufferID);
SoundCallback(bufferID);
processed--;
}
Play();
}
return 0;
}
int ALEngine::SoundCallback(ALuint& bufferID)
{
PTFRAME frame = m_decoder->GetFrame();
if (frame == nullptr)
return -1;
ALenum fmt;
if (frame->chs == 1)
{
fmt = AL_FORMAT_MONO16;
}
else
{
fmt = AL_FORMAT_STEREO16;
}
alBufferData(bufferID, fmt, frame->data, frame->size, frame->samplerate);
alSourceQueueBuffers(m_source, 1, &bufferID);
if (frame)
{
av_free(frame->data);
delete frame;
}
return 0;
} Test main.cpp
#include "ALEngine.h"
#include