video-pc/AvPlayer/AudioPlayer.cpp

#include "AudioPlayer.h"
#include <cstring>
#include <cmath>
extern "C" {
#include <libavutil/avutil.h>
#include <libavutil/samplefmt.h>
#include <libavutil/channel_layout.h>
#include <libavutil/frame.h>
#include <libavutil/mem.h>
#include <libswresample/swresample.h>
}
#include "sonic/sonic.h"
#include <QAudioDeviceInfo>
#include <QAudioFormat>
#include <QDebug>

AudioPlayer::AudioPlayer() {}
AudioPlayer::~AudioPlayer() { reset(); }

void AudioPlayer::reset() {
    if (m_sonicCtx) {
        sonicDestroyStream(m_sonicCtx);
        m_sonicCtx = nullptr;
    }
    if (m_swrCtx) {
        swr_free(&m_swrCtx);
        m_swrCtx = nullptr;
    }
    if (m_swrBuffer) {
        av_free(m_swrBuffer);
        m_swrBuffer = nullptr;
        m_swrBufferSize = 0;
    }
    freeBuffers();
    if (m_audioOutput) {
        m_audioOutput->stop();
        delete m_audioOutput;
        m_audioOutput = nullptr;
    }
    m_audioDevice = nullptr;
    m_sampleRate = 0;
    m_channels = 0;
    m_format = AV_SAMPLE_FMT_NONE;
    m_lastSpeed = 1.0f;
}

void AudioPlayer::freeBuffers() {
    if (m_abuf) {
        av_freep(&m_abuf);
        m_abuf = nullptr;
    }
    if (m_abufOut) {
        av_freep(&m_abufOut);
        m_abufOut = nullptr;
        m_abufOutSize = 0;
    }
    m_maxBufSize = -1;
}

void AudioPlayer::init(const AVFrame* frame, float speed) {
    reset();
    initAudioOutput(frame, speed);
}

void AudioPlayer::initAudioOutput(const AVFrame* frame, float speed) {
    m_sampleRate = frame->sample_rate;
    m_channels = frame->ch_layout.nb_channels;
    m_format = (AVSampleFormat)frame->format;
    m_lastSpeed = speed;
    if (m_sonicCtx) {
        sonicDestroyStream(m_sonicCtx);
        m_sonicCtx = nullptr;
    }
    m_sonicCtx = sonicCreateStream(m_sampleRate, m_channels);
    setSpeed(speed);
    QAudioFormat fmt;
    fmt.setSampleRate(m_sampleRate);
    fmt.setChannelCount(m_channels);
    fmt.setSampleSize(16); // 目标格式 S16
    fmt.setCodec("audio/pcm");
    fmt.setByteOrder(QAudioFormat::LittleEndian);
    fmt.setSampleType(QAudioFormat::SignedInt);
    QAudioDeviceInfo info(QAudioDeviceInfo::defaultOutputDevice());
    if (!info.isFormatSupported(fmt)) {
        qWarning() << "音频格式不支持，尝试最近格式";
        fmt = info.nearestFormat(fmt);
    }
    m_audioOutput = new QAudioOutput(fmt, nullptr);
    m_audioDevice = m_audioOutput->start();
}

void AudioPlayer::setSpeed(float speed) {
    if (m_sonicCtx) {
        sonicSetSpeed(m_sonicCtx, speed);
        sonicSetPitch(m_sonicCtx, 1.0f);
        sonicSetRate(m_sonicCtx, 1.0f);
    }
    m_lastSpeed = speed;
}

void AudioPlayer::setOutputDevice(QIODevice* device) {
    m_audioDevice = device;
}

void AudioPlayer::setAudioOutput(QAudioOutput* output) {
    m_audioOutput = output;
}

QIODevice* AudioPlayer::getOutputDevice() const {
    return m_audioDevice;
}

QAudioOutput* AudioPlayer::getAudioOutput() const {
    return m_audioOutput;
}

bool AudioPlayer::needReinit(const AVFrame* frame, float speed) const {
    if (!m_audioOutput) return true;
    if (m_sampleRate != frame->sample_rate) return true;
    if (m_channels != frame->ch_layout.nb_channels) return true;
    if (m_format != (AVSampleFormat)frame->format) return true;
    if (std::abs(m_lastSpeed - speed) > 1e-3) return true;
    return false;
}

void AudioPlayer::play(AVFrame* frame, float speed) {
    if (!m_audioDevice) return;
    m_sampleRate = frame->sample_rate;
    m_channels = frame->ch_layout.nb_channels;
    m_format = (AVSampleFormat)frame->format;
    // 1. 重采样到 S16
    const AVSampleFormat targetFmt = AV_SAMPLE_FMT_S16;
    uint8_t* inputData = nullptr;
    int inputSamples = frame->nb_samples;
    int inputBytes = av_samples_get_buffer_size(nullptr, m_channels, inputSamples, targetFmt, 1);
    if (m_format != targetFmt) {
        if (!m_swrCtx) {
            m_swrCtx = swr_alloc();
            swr_alloc_set_opts2(&m_swrCtx,
                                &frame->ch_layout, // out_ch_layout
                                targetFmt,
                                m_sampleRate,
                                &frame->ch_layout, // in_ch_layout
                                m_format,
                                m_sampleRate,
                                0,
                                nullptr);
            swr_init(m_swrCtx);
        }
        if (!m_swrBuffer || m_swrBufferSize < inputBytes) {
            if (m_swrBuffer) av_free(m_swrBuffer);
            m_swrBuffer = (uint8_t*)av_malloc(inputBytes);
            m_swrBufferSize = inputBytes;
        }
        int outSamples = swr_convert(
            m_swrCtx,
            &m_swrBuffer, inputSamples,
            (const uint8_t**)frame->extended_data, inputSamples
        );
        inputData = m_swrBuffer;
        inputSamples = outSamples;
        inputBytes = av_samples_get_buffer_size(nullptr, m_channels, outSamples, targetFmt, 1);
    } else {
        inputData = frame->data[0];
        inputBytes = av_samples_get_buffer_size(nullptr, m_channels, inputSamples, targetFmt, 1);
    }
    // 2. 送 sonic 变速
    if (!m_sonicCtx) {
        m_sonicCtx = sonicCreateStream(m_sampleRate, m_channels);
    }
    setSpeed(speed);
    int out_ret = sonicWriteShortToStream(m_sonicCtx, (int16_t*)inputData, inputBytes / (m_channels * 2));
    int num_samples = sonicSamplesAvailable(m_sonicCtx);
    int out_size = num_samples * 2 * m_channels;
    av_fast_malloc(&m_abufOut, &m_abufOutSize, out_size);
    int sonic_samples = 0;
    if (out_ret) {
        sonic_samples = sonicReadShortFromStream(m_sonicCtx, (int16_t*)m_abufOut, num_samples);
        out_size = sonic_samples * 2 * m_channels;
    }
    // 3. 输出到Qt
    int bytesWrite = 0;
    while (bytesWrite < out_size) {
        int written = m_audioDevice->write((const char*)m_abufOut + bytesWrite, out_size - bytesWrite);
        if (written <= 0) break;
        bytesWrite += written;
    }
}