ffmpeg录音存储

有时候项目需要单独录音，这里使用ffmepg，利用dshow的虚拟音频设备采集语音，可以使用命令查看ffmpeg的音频采集设备：ffmpeg -list_devices true -f dshow -i dummy

上图红色框框里的设备名称就是接下来需要用到的。

    //使用ffmpeg dshow自带的音频采集设备，可以通过ffmpeg -list_devices true -f dshow -i dummy命令查看
	pAudioInputFmt = av_find_input_format("dshow");
	if (!pAudioInputFmt)
		return -1;
	char * psDevName = wchar_to_utf8(L"audio=virtual-audio-capturer");
	if (avformat_open_input(&pFmtInputCtx, psDevName, pAudioInputFmt, NULL) < 0)
		return -1;

通过avformat_open_input打开dshow音频采集器。
其它像音频输出、编码器这里略过…
当音频源和音频输出格式、采样率、声音通道数不同时，需要进行重新采样。
这里介绍AVFilter过滤器来进行：

#include "stdafx.h"
#include "AudioResample.h"

CAudioResample::CAudioResample()
{
}

CAudioResample::~CAudioResample()
{
	avfilter_graph_free(&filter_graph);
	filter_graph = nullptr;
	buffersrc_ctx = nullptr;
	buffersink_ctx = nullptr;
}

AVFilterContext* CAudioResample::GetBuffersinkPtr()
{
	return buffersink_ctx;
}

AVFilterContext* CAudioResample::GetBuffersrcoPtr()
{
	return buffersrc_ctx;
}

bool CAudioResample::InitFilter(AVCodecContext* dec_ctx, AVCodecContext* enc_ctx)
{
	if (!dec_ctx || !enc_ctx)
		return false;

	if (dec_ctx->codec_type != AVMEDIA_TYPE_AUDIO)
		return false;

	AVFilterInOut* inputs = avfilter_inout_alloc();
	AVFilterInOut* outputs = avfilter_inout_alloc();
	filter_graph = avfilter_graph_alloc();
	const AVFilter* buffersrc = nullptr;
	const AVFilter* buffersink = nullptr;
	buffersrc_ctx = nullptr;
	buffersink_ctx = nullptr;
	bool result = true;
	char args[512];
	char args2[512];

	if (!inputs || !outputs || !filter_graph)
	{
		result = false;
		goto End;
	}
	buffersrc = avfilter_get_by_name("abuffer");
	buffersink = avfilter_get_by_name("abuffersink");
	if (!buffersrc || !buffersink)
	{
		result = false;
		goto End;
	}
	_snprintf(args, sizeof(args),
		"time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%I64x",
		dec_ctx->time_base.num, dec_ctx->time_base.den, dec_ctx->sample_rate,
		av_get_sample_fmt_name(dec_ctx->sample_fmt),
		dec_ctx->channel_layout);
	if (avfilter_graph_create_filter(&buffersrc_ctx, buffersrc, "in", args, NULL, filter_graph) < 0)
	{
		result = false;
		goto End;
	}
	if (avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out", NULL, NULL, filter_graph)  < 0)
	{
		result = false;
		goto End;
	}

	if (av_opt_set_bin(buffersink_ctx, "sample_fmts", (uint8_t*)&enc_ctx->sample_fmt,
		sizeof(enc_ctx->sample_fmt), AV_OPT_SEARCH_CHILDREN) < 0)
	{
		result = false;
		goto End;
	}
	if (av_opt_set_bin(buffersink_ctx, "sample_rates", (uint8_t*)&enc_ctx->sample_rate,
		sizeof(enc_ctx->sample_rate), AV_OPT_SEARCH_CHILDREN) < 0)
	{
		result = false;
		goto End;
	}
	if (av_opt_set_bin(buffersink_ctx, "channel_layouts", (uint8_t*)&enc_ctx->channel_layout,
		sizeof(enc_ctx->channel_layout), AV_OPT_SEARCH_CHILDREN) < 0)
	{
		result = false;
		goto End;
	}
	inputs->name = av_strdup("out");
	inputs->filter_ctx = buffersink_ctx;
	inputs->pad_idx = 0;
	inputs->next = NULL;
	outputs->name = av_strdup("in");
	outputs->filter_ctx = buffersrc_ctx;
	outputs->pad_idx = 0;
	outputs->next = NULL;
	if (!inputs->name || !outputs->name)
	{
		result = false;
		goto End;
	}
	if (avfilter_graph_parse_ptr(filter_graph, "anull", &inputs, &outputs, NULL) < 0)
	{
		result = false;
		goto End;
	}
	if (avfilter_graph_config(filter_graph, NULL) < 0)
	{
		result = false;
		goto End;
	}
End:
	avfilter_inout_free(&inputs);
	avfilter_inout_free(&outputs);
    
	return result;
}

int CAudioResample::AudioResample_Frame(AVFrame* srcFrame, int flags, AVFrame* destFrame)
{
	if (!srcFrame || !destFrame)
		return -1;

	if (av_buffersrc_add_frame_flags(buffersrc_ctx, srcFrame, flags) < 0)
		return -1;

	if (av_buffersink_get_frame(buffersink_ctx, destFrame) < 0)
		return -1;

	return 0;
}

使用时调用audioresample->AudioResample_Frame(frame, 0, filter_frame);这里frame是音频源解码后的帧，filter_frame是我们需要的重新采样后的帧。
采样后，发现音频源的帧大小和音频输出的帧大小不同：frame_size，因此我们还需要再对采样后的帧加入fifo缓冲区，根据音频输出帧大小再一帧一帧的读取。

int CAudioFrameResize::WriteFrame(AVFrame* frame)
{
	if (!frame)
		return -1;

	return av_audio_fifo_write(mFifo, (void**)frame->data, frame->nb_samples);
}

int CAudioFrameResize::ReadFrame(AVFrame* frame)
{
	if (frame)
		av_frame_free(&frame);
	frame = av_frame_alloc();

	if (av_audio_fifo_size(mFifo) < (mOutAVcodec_ctx->frame_size > 0 ? mOutAVcodec_ctx->frame_size : 1024))
		return -1;

	frame->nb_samples = mOutAVcodec_ctx->frame_size > 0 ? mOutAVcodec_ctx->frame_size : 1024;
	frame->channel_layout = mOutAVcodec_ctx->channel_layout;
	frame->format = mOutAVcodec_ctx->sample_fmt;
	frame->sample_rate = mOutAVcodec_ctx->sample_rate;
	av_frame_get_buffer(frame, 0);

	int ret = av_audio_fifo_read(mFifo, (void **)frame->data, (mOutAVcodec_ctx->frame_size > 0 ? mOutAVcodec_ctx->frame_size : 1024));
	return ret;
}

读取fifo缓冲区读取的帧再进行编码

//循环读取数据，直到fifo缓冲区里数据采样数不够
		while (audioframeresize->ReadFrame(frameInput) >= 0)
		{
			ret = avcodec_send_frame(pOutputCodecCtx, frameInput);
			if (ret < 0)
				continue;

			av_init_packet(&pkt_out);
			ret = avcodec_receive_packet(pOutputCodecCtx, &pkt_out);
			if (ret != 0)
				continue;

			pkt_out.pts = frameIndex * pOutputCodecCtx->frame_size;
			pkt_out.dts = frameIndex * pOutputCodecCtx->frame_size;
			pkt_out.duration = pOutputCodecCtx->frame_size;
			av_write_frame(ofmt_ctx_a, &pkt_out);
			av_packet_unref(&pkt_out);
			frameIndex++;
		}

【github下载】https://github.com/luodh/ffmpeg-record-audio