first
This commit is contained in:
Misaki
@@ -0,0 +1,306 @@
|
||||
//
|
||||
// Created by Administrator on 2025/1/17.
|
||||
//
|
||||
|
||||
#include "AudioInput.h"
|
||||
#include <QDebug>
|
||||
#include <QtMath>
|
||||
#include <QtEndian> // 用于处理字节序
|
||||
|
||||
AudioInput *AudioInput::instance = nullptr;
|
||||
AudioInput *AudioInput::getInstance()
|
||||
{
|
||||
// 懒汉式 依旧单线程无需加锁
|
||||
if (instance == nullptr) {
|
||||
instance = new AudioInput();
|
||||
}
|
||||
return instance;
|
||||
}
|
||||
|
||||
AudioInput::AudioInput(QObject *parent) : QObject(parent)
|
||||
{
|
||||
// new一些必要的对象
|
||||
// 初始化定时器
|
||||
m_timer = new QTimer(this);
|
||||
m_silenceTimer = new QTimer(this);
|
||||
m_thresholdTimer = new QTimer(this);
|
||||
m_thresholdTimer->setSingleShot(true);
|
||||
|
||||
// 连接定时器信号
|
||||
connect(m_timer, &QTimer::timeout, this, &AudioInput::onTimeout); // 录音超时槽函数
|
||||
connect(m_silenceTimer, &QTimer::timeout, this, &AudioInput::stopAudio); // 录音超时槽函数
|
||||
connect(m_thresholdTimer, &QTimer::timeout, this, &AudioInput::thresholdTimeout); // 阈值检测超时槽函数
|
||||
|
||||
// 初始化默认设备和格式
|
||||
m_currentDevice = QMediaDevices::defaultAudioInput();
|
||||
setAudioSettings(); // 使用默认参数
|
||||
}
|
||||
|
||||
AudioInput::~AudioInput()
|
||||
{
|
||||
stopAudio(); // 停止录音
|
||||
if (m_audioSource) {
|
||||
delete m_audioSource;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
void AudioInput::setAudioSettings(const int rate, const int channels)
|
||||
{
|
||||
m_format.setSampleRate(rate);
|
||||
m_format.setChannelCount(channels);
|
||||
// 重要：Qt6 默认可能是 float，为了生成标准 WAV 且方便计算 RMS，强制设为 Int16
|
||||
m_format.setSampleFormat(QAudioFormat::Int16);
|
||||
|
||||
// 检查设备是否支持该格式，不支持则使用最接近的
|
||||
if (!m_currentDevice.isFormatSupported(m_format)) {
|
||||
qWarning() << "Requested format not supported, using preferred format.";
|
||||
m_format = m_currentDevice.preferredFormat();
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
void AudioInput::setAudioPath(const QString &path, const QString &fileName)
|
||||
{
|
||||
this->m_outputFilePath = path + fileName;
|
||||
}
|
||||
|
||||
|
||||
void AudioInput::startAudio()
|
||||
{
|
||||
// 每次开始前重新创建 QAudioSource，确保状态重置
|
||||
if (m_audioSource) {
|
||||
delete m_audioSource;
|
||||
m_audioSource = nullptr;
|
||||
}
|
||||
|
||||
m_audioSource = new QAudioSource(m_currentDevice, m_format, this);
|
||||
|
||||
// 调大缓冲区以避免溢出
|
||||
m_audioSource->setBufferSize(128000);
|
||||
|
||||
// start() 返回一个 QIODevice，我们可以从中读取数据
|
||||
m_ioDevice = m_audioSource->start();
|
||||
|
||||
if (m_ioDevice) {
|
||||
connect(m_ioDevice, &QIODevice::readyRead, this, &AudioInput::onReadyRead);
|
||||
qDebug() << "Started recording with device:" << m_currentDevice.description();
|
||||
} else {
|
||||
qCritical() << "Failed to start audio recording.";
|
||||
}
|
||||
}
|
||||
|
||||
void AudioInput::stopAudio()
|
||||
{
|
||||
if (m_audioSource) {
|
||||
m_audioSource->stop();
|
||||
// 注意：不要立即 delete m_audioSource，某些情况下可能导致 crash，停止即可
|
||||
}
|
||||
|
||||
// 停止所有定时器
|
||||
m_timer->stop();
|
||||
m_silenceTimer->stop();
|
||||
m_thresholdTimer->stop();
|
||||
|
||||
// 生成 WAV 数据
|
||||
QByteArray wavData;
|
||||
if (!m_rawPCMData.isEmpty()) {
|
||||
wavData = generateWavHeader(m_rawPCMData.size());
|
||||
wavData.append(m_rawPCMData);
|
||||
|
||||
// 如果需要保存文件
|
||||
if (!m_outputFilePath.isEmpty()) {
|
||||
QFile file(m_outputFilePath);
|
||||
if (file.open(QIODevice::WriteOnly)) {
|
||||
file.write(wavData);
|
||||
file.close();
|
||||
qDebug() << "Saved WAV to:" << m_outputFilePath;
|
||||
}
|
||||
}
|
||||
|
||||
m_rawPCMData.clear();
|
||||
}
|
||||
|
||||
isAutoRecording = false;
|
||||
isAutoThreshold = false;
|
||||
|
||||
emit recordingFinished();
|
||||
emit recordingFinished_Byte(wavData);
|
||||
qDebug() << "Recording stopped.";
|
||||
}
|
||||
|
||||
// 阈值检测超时槽函数
|
||||
void AudioInput::onReadyRead()
|
||||
{
|
||||
if (!m_ioDevice) return;
|
||||
|
||||
// 读取当前所有可用的音频数据
|
||||
QByteArray data = m_ioDevice->readAll();
|
||||
if (data.isEmpty()) return;
|
||||
|
||||
// 1. 保存原始 PCM 数据
|
||||
m_rawPCMData.append(data);
|
||||
|
||||
// 2. 计算 RMS (仅用于分析，取最后一小段或者整体计算，这里计算当前块的RMS)
|
||||
m_rmsValue = calculateRMS(data);
|
||||
|
||||
// 3. 自动停止逻辑 (VAD)
|
||||
if (isAutoRecording) {
|
||||
// 输出 RMS 用于调试
|
||||
// qDebug() << "RMS:" << m_rmsValue;
|
||||
|
||||
if (m_rmsValue < m_silenceThreshold) {
|
||||
// 静音状态
|
||||
if (!m_silenceTimer->isActive()) {
|
||||
m_silenceTimer->start(m_silenceDuration);
|
||||
}
|
||||
} else {
|
||||
// 有声音，重置定时器
|
||||
m_silenceTimer->stop();
|
||||
m_silenceTimer->start(m_silenceDuration);
|
||||
}
|
||||
}
|
||||
|
||||
// 4. 自动阈值计算逻辑
|
||||
if (isAutoThreshold) {
|
||||
m_rmsValues.push_back(m_rmsValue);
|
||||
emit rmsRealValue(m_rmsValue);
|
||||
}
|
||||
}
|
||||
|
||||
qreal AudioInput::calculateRMS(const QByteArray& buffer)
|
||||
{
|
||||
if (buffer.isEmpty()) return 0;
|
||||
|
||||
// 假设是 Int16 格式 (16位深)
|
||||
// 如果是 Stereo，数据排列是 L R L R...
|
||||
// 简单的 RMS 计算可以将所有通道数据视为一个长序列
|
||||
|
||||
const qint16 *data = reinterpret_cast<const qint16*>(buffer.constData());
|
||||
const int sampleCount = buffer.size() / sizeof(qint16); // 样本数量
|
||||
|
||||
if (sampleCount == 0) return 0;
|
||||
|
||||
qreal sumSquared = 0;
|
||||
for (int i = 0; i < sampleCount; ++i) {
|
||||
const qreal sample = static_cast<qreal>(data[i]);
|
||||
sumSquared += sample * sample;
|
||||
}
|
||||
|
||||
return qSqrt(sumSquared / sampleCount);
|
||||
}
|
||||
|
||||
// 启动带时长的录音
|
||||
void AudioInput::startAudioWithDuration(int duration)
|
||||
{
|
||||
startAudio();
|
||||
m_timer->start(duration * 1000);
|
||||
}
|
||||
|
||||
void AudioInput::onTimeout()
|
||||
{
|
||||
stopAudio();
|
||||
qDebug() << "Recording stopped by duration timeout.";
|
||||
}
|
||||
|
||||
// 获取所有音频输入设备
|
||||
QList<QString> AudioInput::getAvailableAudioInputDevices()
|
||||
{
|
||||
QList<QString> list;
|
||||
const auto devices = QMediaDevices::audioInputs();
|
||||
for (const auto &device : devices) {
|
||||
list.append(device.description());
|
||||
}
|
||||
return list;
|
||||
}
|
||||
|
||||
// 设置当前录音设备
|
||||
void AudioInput::setAudioInputDevice(const QString &deviceName)
|
||||
{
|
||||
const auto devices = QMediaDevices::audioInputs();
|
||||
for (const auto &device : devices) {
|
||||
if (device.description() == deviceName) {
|
||||
m_currentDevice = device;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 启动自动录音 (VAD)
|
||||
void AudioInput::startAutoStopAudio(const qreal silenceThreshold, const int silenceDuration)
|
||||
{
|
||||
isAutoRecording = true;
|
||||
m_silenceThreshold = silenceThreshold;
|
||||
m_silenceDuration = silenceDuration;
|
||||
|
||||
startAudio();
|
||||
|
||||
// 延迟启动静音检测，给一点缓冲时间
|
||||
QTimer::singleShot(200, this, [this](){
|
||||
m_silenceTimer->start(m_silenceDuration);
|
||||
});
|
||||
}
|
||||
|
||||
// 启动阈值计算
|
||||
void AudioInput::startAutoThresholdClu(int Duration)
|
||||
{
|
||||
isAutoThreshold = true;
|
||||
m_rmsValues.clear();
|
||||
startAudio();
|
||||
m_thresholdTimer->start(Duration);
|
||||
}
|
||||
|
||||
void AudioInput::thresholdTimeout()
|
||||
{
|
||||
isAutoThreshold = false;
|
||||
stopAudio(); // 内部会处理 stop
|
||||
|
||||
if (!m_rmsValues.empty()) {
|
||||
const double sum = std::accumulate(m_rmsValues.begin(), m_rmsValues.end(), 0.0);
|
||||
const double avg = sum / m_rmsValues.size();
|
||||
m_silenceThreshold = avg + 500.0; // 这里的 500 是经验值，可以根据需要调整
|
||||
emit thresholdCalculated(m_silenceThreshold);
|
||||
} else {
|
||||
emit thresholdCalculated(0);
|
||||
}
|
||||
}
|
||||
|
||||
QByteArray AudioInput::generateWavHeader(const quint32 dataSize) const {
|
||||
// WAV头结构定义
|
||||
struct WavHeader {
|
||||
char riff[4] = {'R','I','F','F'};
|
||||
quint32 chunkSize;
|
||||
char wave[4] = {'W','A','V','E'};
|
||||
char fmt[4] = {'f','m','t',' '};
|
||||
quint32 fmtSize = 16;
|
||||
quint16 audioFormat = 1; // PCM
|
||||
quint16 numChannels;
|
||||
quint32 sampleRate;
|
||||
quint32 byteRate;
|
||||
quint16 blockAlign;
|
||||
quint16 bitsPerSample;
|
||||
char data[4] = {'d','a','t','a'};
|
||||
quint32 dataSize;
|
||||
} header;
|
||||
|
||||
header.numChannels = static_cast<quint16>(m_format.channelCount());
|
||||
header.sampleRate = static_cast<quint32>(m_format.sampleRate());
|
||||
header.bitsPerSample = 16; // 我们强制使用了 Int16
|
||||
|
||||
header.byteRate = header.sampleRate * header.numChannels * (header.bitsPerSample / 8);
|
||||
header.blockAlign = header.numChannels * (header.bitsPerSample / 8);
|
||||
header.dataSize = dataSize;
|
||||
header.chunkSize = 36 + dataSize;
|
||||
|
||||
return QByteArray(reinterpret_cast<const char*>(&header), sizeof(WavHeader));
|
||||
}
|
||||
|
||||
void AudioInput::setSilenceThreshold(const qreal silenceThreshold)
|
||||
{
|
||||
this->m_silenceThreshold = silenceThreshold;
|
||||
}
|
||||
|
||||
qreal AudioInput::getSilenceThreshold() const
|
||||
{
|
||||
return this->m_silenceThreshold;
|
||||
}
|
||||
Reference in New Issue
Block a user