简体   繁体   English

使用portaudio和sndfile播放WAV文件

[英]Playing wav file with portaudio and sndfile

I have written a function to play a sound file using portaudio and sndfile. 我编写了一个使用portaudio和sndfile播放声音文件的功能。 Unfortunately the sound quality is terrible. 不幸的是,声音质量太差了。 The sound is more like a hiss. 声音更像嘶嘶声。 The following is the source code of the function I am using. 以下是我正在使用的函数的源代码。

#define _GLIBCXX_USE_C99_MATH 1
#include "PlaySound_config.h"
#include <boost/predef.h>

#if !defined(USE_PORTAUDIO) // {
#  define USE_PORTAUDIO 0
#  if (! BOOST_OS_CYGWIN && ! BOOST_OS_WINDOWS) // {
#    undef USE_PORTAUDIO
#    define USE_PORTAUDIO 1
#  endif // }
#endif // }

#if (PLAY_SOUND_HAVE_PORTAUDIO_H && PLAY_SOUND_HAVE_SNDFILE_H && PLAY_SOUND_HAVE_SNDFILE_HH && USE_PORTAUDIO) // {

#if (PLAY_SOUND_HAVE_UNISTD_H)
#  include <unistd.h>
#endif

#include <errno.h>
#include <stdio.h>
#include <stdlib.h>

#include <portaudio.h>
#include <sndfile.hh>

#include <cmath>
#include <fstream>
#include <iostream>
#include <vector>

#include <boost/filesystem/operations.hpp>
#include <boost/filesystem/path.hpp>
#include <boost/format.hpp>

#include "PlaySound.h"
#include "PlaySoundStrings.h"

void SoundWarning(const std::string& message)
{
    std::cerr << message << std::endl;
}

bool PlaySoundFile(const std::string& soundFile, unsigned long /* volume */)
{
    const int MAX_CHANNELS = 1;
    const double SAMPLE_RATE = 11025.0;
    const unsigned long FRAMES_PER_BUFFER = 1024;
    const size_t BUFFER_LEN = 1024;
    using boost::format;
    using boost::io::group;
    std::string message;
    if (soundFile.empty())
    {
        errno = EINVAL;
        message = playSoundStrings[error_invalid_argument];
        SoundWarning(message);
        return false;
    }
    boost::filesystem::path soundFilePath(soundFile);
    if (! boost::filesystem::exists(soundFilePath))
    {
        errno = EINVAL;
        message = str(format(playSoundStrings[error_file_does_not_exist]) % soundFile.c_str());
        SoundWarning(message);
        return false;
    }
    PaError paError = Pa_Initialize();
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_initialize_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        return false;
    }
    SNDFILE* sndFile;
    SF_INFO sfInfo;
    sndFile = sf_open(soundFile.c_str(), SFM_READ, &sfInfo);
    if (! sndFile)
    {
        message = str(format(playSoundStrings[error_sf_open_failed]) % soundFile.c_str() % sf_strerror(nullptr));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    if (sfInfo.channels > MAX_CHANNELS)
    {
        message = str(format(playSoundStrings[error_too_many_channels]) % sfInfo.channels % MAX_CHANNELS);
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    PaStream* stream = nullptr;
    PaStreamParameters paStreamParameters;
    paStreamParameters.device = Pa_GetDefaultOutputDevice();
    paStreamParameters.channelCount = sfInfo.channels;
    paStreamParameters.sampleFormat = paInt16;
    paStreamParameters.suggestedLatency = Pa_GetDeviceInfo(paStreamParameters.device)->defaultLowOutputLatency;
    paStreamParameters.hostApiSpecificStreamInfo = nullptr;
    paError = Pa_OpenStream(
        &stream, nullptr, &paStreamParameters,
        SAMPLE_RATE, FRAMES_PER_BUFFER, paClipOff,
        nullptr, nullptr);
    if (paError != paNoError || ! stream)
    {
        message = str(format(playSoundStrings[error_pa_open_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    paError = Pa_StartStream(stream);
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_start_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    sf_count_t readCount = 0;
    double data[BUFFER_LEN];
    while ((readCount = sf_read_double(sndFile, data, BUFFER_LEN)))
    {
        paError = Pa_WriteStream(stream, data, BUFFER_LEN);
        if (paError != paNoError)
        {
            message = str(format(playSoundStrings[error_pa_write_stream_failed]) % Pa_GetErrorText(paError));
            SoundWarning(message);
            break;
        }
    }
    paError = Pa_CloseStream(stream);
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_close_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    Pa_Terminate();
    return true;
}

I saw some sample code in the article What is a lightweight cross platform WAV playing library? 我在文章“ 什么是轻量级跨平台WAV播放库”中看到了一些示例代码 but the sample is incomplete. 但样本不完整。 It appears that it will only play the first five seconds of the file. 看来它只会播放文件的前五秒。 I want to play the entire file. 我想播放整个文件。

Any idea what I am doing wrong? 知道我在做什么错吗?

This code is part of my PlaySound project. 此代码是我的PlaySound项目的一部分。

I made several mistakes in the original version of my code. 我在代码的原始版本中犯了几个错误。 The first was in the line in which I initialized the sampleFormat member of the PaStreamParameters structure. 第一个是我初始化PaStreamParameters结构的sampleFormat成员的行。

In my original code I initialized this member as follows. 在我的原始代码中,我按如下方式初始化了该成员。

paStreamParameters.sampleFormat = paInt16;

I should have initialized it as follows. 我应该将其初始化如下。

paStreamParameters.sampleFormat = paInt32;

My next mistake was in the call to the Pa_OpenStream function. 我的下一个错误是在调用Pa_OpenStream函数。 I set the sampleRate parameter to a hard coded constant, in this case 11025.0. 我将sampleRate参数设置为硬编码常量,在这种情况下为11025.0。 I should have set it to the value of the samplerate member of the SF_INFO structure. 我应该将其设置为SF_INFO结构的samplerate成员的值。

My third mistake was to use the sf_read_double function to read from the sound file. 我的第三个错误是使用sf_read_double函数从声音文件中读取。 In several working samples I eventually found, including the sndfile-play application, the sf_read_float function is used instead. 我最终发现了几个工作示例,包括sndfile-play应用程序,而是使用sf_read_float函数。

My forth mistake is that I did not scale the data read from the sound file before passing it to the Pa_WriteStream function. 我的第四个错误是,在传递给Pa_WriteStream函数之前,我没有缩放从声音文件读取的数据。 I found the code to scale the data in the source code of the sndfile-play application. 我在sndfile-play应用程序的源代码中找到了用于缩放数据的代码。

For anyone who is interested, the final version of my source code is as follows. 对于任何有兴趣的人,我的源代码的最终版本如下。

#define _GLIBCXX_USE_C99_MATH 1
#include "PlaySound_config.h"
#include <boost/predef.h>

#if !defined(USE_PORTAUDIO) // {
#  define USE_PORTAUDIO 0
#  if (! BOOST_OS_CYGWIN && ! BOOST_OS_WINDOWS) // {
#    undef USE_PORTAUDIO
#    define USE_PORTAUDIO 1
#  endif // }
#endif // }

#if (PLAY_SOUND_HAVE_PORTAUDIO_H && PLAY_SOUND_HAVE_SNDFILE_H && PLAY_SOUND_HAVE_SNDFILE_HH && USE_PORTAUDIO) // {

#if (PLAY_SOUND_HAVE_UNISTD_H)
#  include <unistd.h>
#endif

#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include <portaudio.h>
#include <sndfile.hh>

#include <cmath>
#include <fstream>
#include <iostream>
#include <vector>

#include <boost/filesystem/operations.hpp>
#include <boost/filesystem/path.hpp>
#include <boost/format.hpp>

#include "PlaySound.h"
#include "PlaySoundStrings.h"

void SoundWarning(const std::string& message)
{
    std::cerr << message << std::endl;
}

bool PlaySoundFile(const std::string& soundFile, unsigned long /* volume */)
{
    const int MAX_CHANNELS = 1;
    const size_t BUFFER_LEN = 1024;
    using boost::format;
    using boost::io::group;
    std::string message;
    if (soundFile.empty())
    {
        errno = EINVAL;
        message = playSoundStrings[error_invalid_argument];
        SoundWarning(message);
        return false;
    }
    boost::filesystem::path soundFilePath(soundFile);
    if (! boost::filesystem::exists(soundFilePath))
    {
        errno = EINVAL;
        message = str(format(playSoundStrings[error_file_does_not_exist]) % soundFile.c_str());
        SoundWarning(message);
        return false;
    }
    PaError paError = Pa_Initialize();
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_initialize_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        return false;
    }
    SNDFILE* sndFile = nullptr;
    SF_INFO sfInfo;
    ::memset(&sfInfo, 0, sizeof(sfInfo));
    sndFile = sf_open(soundFile.c_str(), SFM_READ, &sfInfo);
    if (! sndFile)
    {
        message = str(format(playSoundStrings[error_sf_open_failed]) % soundFile.c_str() % sf_strerror(nullptr));
        SoundWarning(message);
           Pa_Terminate();
        return false;
    }
    if (sfInfo.channels > MAX_CHANNELS)
    {
        message = str(format(playSoundStrings[error_too_many_channels]) % sfInfo.channels % MAX_CHANNELS);
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    PaStream* stream = nullptr;
    PaStreamParameters paStreamParameters;
    paStreamParameters.device = Pa_GetDefaultOutputDevice();
    paStreamParameters.channelCount = sfInfo.channels;
    paStreamParameters.sampleFormat = paInt32;
    paStreamParameters.suggestedLatency = Pa_GetDeviceInfo(paStreamParameters.device)->defaultLowOutputLatency;
    paStreamParameters.hostApiSpecificStreamInfo = nullptr;
    paError = Pa_OpenStream(
        &stream, nullptr, &paStreamParameters,
        sfInfo.samplerate, paFramesPerBufferUnspecified, paClipOff,
        nullptr, nullptr);
    if (paError != paNoError || ! stream)
    {
        message = str(format(playSoundStrings[error_pa_open_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    paError = Pa_StartStream(stream);
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_start_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    int subFormat = sfInfo.format & SF_FORMAT_SUBMASK;
    double scale = 1.0;
    if (subFormat == SF_FORMAT_FLOAT || subFormat == SF_FORMAT_DOUBLE)
    {
        sf_command(sndFile, SFC_CALC_SIGNAL_MAX, &scale, sizeof(scale));
        if (scale < 1e-10)
        {
            scale = 1.0;
        }
        else
        {
            scale = 32700.0 / scale;
        }
    }
    sf_count_t readCount = 0;
    float data[BUFFER_LEN];
    ::memset(data, 0, sizeof(data));
    while ((readCount = sf_read_float(sndFile, data, BUFFER_LEN)))
    {
        if (subFormat == SF_FORMAT_FLOAT || subFormat == SF_FORMAT_DOUBLE)
        {
            int m = 0;
            for (m = 0 ; m < readCount ; ++m)
            {
                data[m] *= scale;
            }
        }
        paError = Pa_WriteStream(stream, data, BUFFER_LEN);
        if (paError != paNoError)
        {
            message = str(format(playSoundStrings[error_pa_write_stream_failed]) % Pa_GetErrorText(paError));
            SoundWarning(message);
            break;
        }
        ::memset(data, 0, sizeof(data));
    }
    paError = Pa_CloseStream(stream);
    if (paError != paNoError)
    {
        message = str(format(playSoundStrings[error_pa_close_stream_failed]) % Pa_GetErrorText(paError));
        SoundWarning(message);
        Pa_Terminate();
        return false;
    }
    Pa_Terminate();
    return true;
}

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM