当前位置：首页 > news >正文

音视频学习（五十四）：基于ffmpeg实现音频重采样

news 2025/8/17 5:52:13

概述

音频重采样指的是将音频数据的采样率从一个值转换到另一个值。

ffmpeg重采样

关键概念

AVFrame: FFmpeg中用于存储解码后的原始（raw）音视频数据的数据结构。对于音频，它包含了音频样本数据、采样率、声道布局、样本格式等信息。
AVCodecContext: 编解码器的上下文，包含了编解码器所需的各种参数，如码率、分辨率、采样率等。在重采样中，虽然我们不直接使用编解码器，但它帮助我们理解音频流的参数。
AVSampleFormat: 音频样本的格式，例如 AV_SAMPLE_FMT_S16（16位有符号整数）、AV_SAMPLE_FMT_FLT（32位浮点数）等。FFmpeg支持多种格式，重采样通常需要在不同格式之间进行转换。
AVChannelLayout: 声道布局，例如 AV_CHANNEL_LAYOUT_STEREO（立体声）、AV_CHANNEL_LAYOUT_MONO（单声道）等。

示例

// 使用ffmpeg6以上版本
#include <iostream>
#include <vector>extern "C" {
#include <libavcodec/avcodec.h>
#include <libswresample/swresample.h>
#include <libavutil/avutil.h>
#include <libavutil/opt.h>
#include <libavutil/error.h>
#include <libavutil/channel_layout.h>
}// 错误处理宏
#define CHECK_RET(ret) do { \if ((ret) < 0) { \char errbuf[AV_ERROR_MAX_STRING_SIZE]; \av_strerror(ret, errbuf, AV_ERROR_MAX_STRING_SIZE); \std::cerr << "FFmpeg Error: " << errbuf << std::endl; \return 1; \} \
} while(0)// 打印音频参数
void print_audio_params(const char* label, int sample_rate, enum AVSampleFormat format, const AVChannelLayout* channel_layout) {char layout_str[256];av_channel_layout_describe(channel_layout, layout_str, sizeof(layout_str));std::cout << "--- " << label << " ---" << std::endl;std::cout << "Sample Rate: " << sample_rate << " Hz" << std::endl;std::cout << "Sample Format: " << av_get_sample_fmt_name(format) << std::endl;std::cout << "Channel Layout: " << layout_str << std::endl;std::cout << "Number of Channels: " << channel_layout->nb_channels << std::endl;
}int main() {// 源音频参数int src_sample_rate = 44100;enum AVSampleFormat src_sample_fmt = AV_SAMPLE_FMT_S16;AVChannelLayout src_channel_layout;av_channel_layout_default(&src_channel_layout, 2); // 2声道int src_nb_channels = src_channel_layout.nb_channels;int src_nb_samples = 1024;// 目标音频参数int dst_sample_rate = 48000;enum AVSampleFormat dst_sample_fmt = AV_SAMPLE_FMT_FLT;AVChannelLayout dst_channel_layout;av_channel_layout_default(&dst_channel_layout, 1); // 1声道int dst_nb_channels = dst_channel_layout.nb_channels;print_audio_params("Source Audio", src_sample_rate, src_sample_fmt, &src_channel_layout);print_audio_params("Destination Audio", dst_sample_rate, dst_sample_fmt, &dst_channel_layout);SwrContext *swr_ctx = swr_alloc();if (!swr_ctx) {std::cerr << "Could not allocate SwrContext" << std::endl;return 1;}av_opt_set_chlayout(swr_ctx, "in_chlayout", &src_channel_layout, 0);av_opt_set_int(swr_ctx, "in_sample_rate", src_sample_rate, 0);av_opt_set_sample_fmt(swr_ctx, "in_sample_fmt", src_sample_fmt, 0);av_opt_set_chlayout(swr_ctx, "out_chlayout", &dst_channel_layout, 0);av_opt_set_int(swr_ctx, "out_sample_rate", dst_sample_rate, 0);av_opt_set_sample_fmt(swr_ctx, "out_sample_fmt", dst_sample_fmt, 0);int ret = swr_init(swr_ctx);CHECK_RET(ret);uint8_t **src_data = nullptr;int src_linesize = 0;ret = av_samples_alloc_array_and_samples(&src_data, &src_linesize, src_nb_channels, src_nb_samples, src_sample_fmt, 0);CHECK_RET(ret);for (int i = 0; i < src_nb_samples * src_nb_channels; ++i) {if (src_sample_fmt == AV_SAMPLE_FMT_S16) {((int16_t*)src_data[0])[i] = i;}}int dst_nb_samples = av_rescale_rnd(swr_get_delay(swr_ctx, src_sample_rate) + src_nb_samples, dst_sample_rate, src_sample_rate, AV_ROUND_UP);uint8_t **dst_data = nullptr;int dst_linesize = 0;ret = av_samples_alloc_array_and_samples(&dst_data, &dst_linesize, dst_nb_channels, dst_nb_samples, dst_sample_fmt, 0);CHECK_RET(ret);ret = swr_convert(swr_ctx, dst_data, dst_nb_samples, (const uint8_t **)src_data, src_nb_samples);CHECK_RET(ret);std::cout << "Successfully resampled " << src_nb_samples << " samples to " << ret << " samples." << std::endl;while (true) {ret = swr_convert(swr_ctx, dst_data, dst_nb_samples, nullptr, 0);if (ret <= 0) break;std::cout << "Flushed " << ret << " remaining samples." << std::endl;}if (src_data) {av_freep(&src_data[0]);av_freep(&src_data);}if (dst_data) {av_freep(&dst_data[0]);av_freep(&dst_data);}swr_free(&swr_ctx);return 0;
}

编译运行：

g++ resample.cpp -o resample_example -I/usr/local/ffmpeg/include -L/usr/local/ffmpeg/lib -lavcodec -lswresample -lavutil

输出：

--- Source Audio ---
Sample Rate: 44100 Hz
Sample Format: s16
Channel Layout: stereo
Number of Channels: 2
--- Destination Audio ---
Sample Rate: 48000 Hz
Sample Format: flt
Channel Layout: mono
Number of Channels: 1
Successfully resampled 1024 samples to 1098 samples.
Flushed 17 remaining samples.

查看全文

http://www.dtcms.com/a/334156.html