调用科大讯飞API实现文本转语音

最新推荐文章于 2024-10-15 11:53:10 发布

他过江

最新推荐文章于 2024-10-15 11:53:10 发布

阅读量1w

点赞数 1

分类专栏： FFmpeg 文章标签：语音合成 FFmpeg C

本文链接：https://blog.csdn.net/ET_Endeavoring/article/details/78694654

版权

FFmpeg 专栏收录该内容

19 篇文章 7 订阅

订阅专栏

text_to_speech

调用科大讯飞API实现文本转语音(wav)

采用科大讯飞提供的API接口，使用C语言实现语音合成功能。

语音合成功能：用户输入一段文字，会生成相应语音，可以选择发音的人物，音量，语速。

	/* 音频合成参数 */
	typedef struct SpeechSynsContext {
		const char *text;			/* 输入文本 */
		const char *out_path;		/* 音频输出路径 */
		const char *voice_name;		/* 发音人[xiaoqian,dalong,yufeng] */
		const char *volume;			/* 音量[0,100] */
		const char *speed;			/* 语速[0,100] */
		const char *pitch;			/* 音调[0,100] */
		const char *sample_rate;	/* 采样率[8000,16000] */
		const char *text_encoding;	/* 文本编码格式[GB2312；GBK；BIG5；UNICODE；GB18030；UTF8] */
		const char *rdn;			/* 音频数字发音方式[0：数值优先,1：完全数值,2：完全字符串，3：字符串优先。] */
		char * session_parm;		/* 会话参数 */
		void *callback_fun;			/* 回调函数 */
		void *current_ctx;			/* air当前的FREContext结构体 */
		int progress;				/* 回调进度（0-100） */
		void *options;				/* Structure TTS_Options */
		char *tmp_file;		/* wav临时文件 */
	}SpeechSynsContext;

/* wav音频头部格式 */
typedef struct _wave_pcm_hdr
{
	char            riff[4];                // = "RIFF"
	int				size_8;                 // = FileSize - 8
	char            wave[4];                // = "WAVE"
	char            fmt[4];                 // = "fmt "
	int				fmt_size;				/* = 下一个结构体的大小 : 16 */

	short int       format_tag;             // = PCM : 1
	short int       channels;               /* = 通道数 : 1*/
	int				samples_per_sec;        /* = 采样率 : 8000 | 6000 | 11025 | 16000 */
	int				avg_bytes_per_sec;      /* = 每秒字节数 : samples_per_sec * bits_per_sample / 8 */
	short int       block_align;            /* = 每采样点字节数 : wBitsPerSample / 8 */
	short int       bits_per_sample;        /* = 量化比特数: 8 | 16 */

	char            data[4];                // = "data";
	int				data_size;              /* = 纯数据长度 : FileSize - 44 */
} wave_pcm_hdr;

typedef struct TTS_Options {

	pthread_t tid;				/* 线程id */

}TTS_Options;

/*
*	fun:		回调函数
*	status:		结束状态
*	percent:	进度百分比
*	identifier:	结构体对象
*/
typedef void(*TTSCallbackFunction)(int, int, void*);

/* 默认wav音频头部数据 */
wave_pcm_hdr default_wav_hdr =
{
	{ 'R', 'I', 'F', 'F' },
	0,
	{ 'W', 'A', 'V', 'E' },
	{ 'f', 'm', 't', ' ' },
	16,
	1,
	1,
	16000,
	32000,
	2,
	16,
	{ 'd', 'a', 't', 'a' },
	0
};

static void tts_free(SpeechSynsContext *ssc);
static SpeechSynsContext *tts_alloc(const char *text, const char *out_path, const char *voice_name, const char *volume,
	const char *speed, const char *pitch, const char *sample_rate, const char *text_encoding,
	const char *rdn, void *callback_fun, void *fre_ctx);
static int tts_main(SpeechSynsContext *ssc);
/* 当前进度百分比 */
static int CurrentPercent(unsigned int wav_current_size, const char *src_text, const char *speed);
static void *tts_thread(void *SpeechSynsCtx);
static char* CreateTempFileName();

/*
*	fun:		回调函数
*	status:		结束状态
*	percent:	进度百分比
*	identifier:	结构体对象
*/
static void callback_function(TTSCallbackFunction fun, int status, int percent, void *identifier);

tts主函数：

void tts(const char *text, const char *out_path, const char *voice_name, const char *volume,
	const char *speed, const char *pitch, const char *sample_rate, const char *text_encoding,
	const char *rdn, void *callback_fun, void *fre_ctx)
{

	int ret;
	SpeechSynsContext *ssc = tts_alloc(text, out_path, voice_name, volume,
		speed, pitch, sample_rate, text_encoding,
		rdn, callback_fun, fre_ctx);
	if (ssc == NULL) {
		return;
	}
	TTS_Options *op = (TTS_Options *)ssc->options;
	if ((ret = pthread_create(&op->tid, NULL, tts_thread, ssc)) != 0) {
		printf("pthread_create failed\n");
		return;
	}
	/* Release all resource when thread destoryed */
	pthread_detach(op->tid);
	return;
}

tts_alloc:

static SpeechSynsContext *tts_alloc(const char *text, const char *out_path, const char *voice_name, const char *volume,
	const char *speed, const char *pitch, const char *sample_rate, const char *text_encoding,
	const char *rdn, void *callback_fun, void *fre_ctx)
{
	char *UTF8_Path;
	int ret;

	/* 保证所有参数不为NULL */
	if (!(text&&out_path&&voice_name&&volume&&speed&&pitch&&sample_rate&&text_encoding&&rdn/*&&callback_fun*/)) {
		LOG_PRINT("%s....Parameter invalid", __FUNCTION__);
		return NULL;
	}
	/* 没有判断指针为空 */
	SpeechSynsContext *ssc = (SpeechSynsContext*)malloc(sizeof(SpeechSynsContext));
	ssc->session_parm = (char *)malloc(300);
	ssc->options = malloc(sizeof(TTS_Options));

	const char *vn = "voice_name = ";
	const char *vol = "volume = ";
	const char *sp = "speed = ";
	const char *pit = "pitch = ";
	const char *sr = "sample_rate = ";
	const char *te = "text_encoding = ";
	const char *rd = "rdn = ";
	const char *pnct = ", ";

	if (!strcmp(text, "")) {
		tts_free(ssc);
		return NULL;
	}
	else {
		ssc->text = tts_parm_copy(text);
		LOG_PRINT("%s....text is %s", __FUNCTION__, ssc->text);
	}
	if (!strcmp(out_path, "")) {
		return NULL;
	}
	else {
		ssc->out_path = tts_parm_copy(out_path);
		LOG_PRINT("%s....out_path2 is %s", __FUNCTION__, ssc->out_path);
	}
	if (!strcmp(voice_name, "")) {
		ssc->voice_name = tts_parm_copy("xiaoyan");
	}
	else {
		ssc->voice_name = tts_parm_copy(voice_name);
	}
	if (!strcmp(volume, "")) {
		ssc->volume = tts_parm_copy("50");
	}
	else {
		ssc->volume = tts_parm_copy(volume);
	}
	if (!strcmp(speed, "")) {
		ssc->speed = tts_parm_copy("50");
	}
	else {
		ssc->speed = tts_parm_copy(speed);
	}
	if (!strcmp(pitch, "")) {
		ssc->pitch = tts_parm_copy("50");
	}
	else {
		ssc->pitch = tts_parm_copy(pitch);
	}
	if (!strcmp(sample_rate, "")) {
		ssc->sample_rate = tts_parm_copy("16000");
	}
	else {
		ssc->sample_rate = tts_parm_copy(sample_rate);
	}
	if (!strcmp(text_encoding, "")) {
		ssc->text_encoding = tts_parm_copy("utf8");		/*"gb2312"*/
	}
	else {
		ssc->text_encoding = tts_parm_copy(text_encoding);
	}
	if (!strcmp(rdn, "")) {
		ssc->rdn = tts_parm_copy("2");
	}
	else {
		ssc->rdn = tts_parm_copy(rdn);
	}

	ssc->callback_fun = callback_fun;
	ssc->current_ctx = fre_ctx;
	ssc->progress = 0;

	TTS_Options *op = (TTS_Options *)ssc->options;
	op->tid.p = NULL;
	op->tid.x = 0;
	callback_function((TTSCallbackFunction)ssc->callback_fun, 0, 0, ssc);

	/* UTF8 */
	UTF8_Path = CreateTempFileName();
	if (UTF8_Path == NULL) {
		LOG_PRINT("%s....Create temp file UTF8_Path failed", __FUNCTION__);
		return NULL;
	}

	ssc->tmp_file = (char*)malloc(1024);
	memset(ssc->tmp_file, 0, 1024);

	/* UTF8转ANSI */
	UTF8toANSI(UTF8_Path, ssc->tmp_file);

	free(UTF8_Path);

	if (ssc->tmp_file == NULL) {
		LOG_PRINT("%s....UTF8 to ANSI failed", __FUNCTION__);
		return NULL;
	}

	sprintf_s(ssc->session_parm, 300, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
		sp, ssc->speed, pnct,
		te, ssc->text_encoding, pnct,
		vn, ssc->voice_name, pnct,
		sr, ssc->sample_rate, pnct,
		vol, ssc->volume, pnct,
		pit, ssc->pitch, pnct,
		rd, ssc->rdn);

	return ssc;
}

tts_thread:

static void *tts_thread(void *SpeechSynsCtx)
{
	__try {
		int ret = -1;
		SpeechSynsContext *ssc = (SpeechSynsContext*)SpeechSynsCtx;
		ret = tts_main(ssc);
		if (ret) {
			LOG_PRINT("%s....tts_main failed", __FUNCTION__);
			callback_function((TTSCallbackFunction)ssc->callback_fun, -1, ssc->progress, ssc);
			//return 0;
			return NULL;
		}

		ret = transcode_audio(ssc->tmp_file, ssc);
		if (ret) {
			callback_function((TTSCallbackFunction)ssc->callback_fun, -1, ssc->progress, ssc);
			LOG_PRINT("%s....transcode_audio failed", __FUNCTION__);
			goto cleanup;
		}

		/* 回调100% */
		callback_function((TTSCallbackFunction)ssc->callback_fun, 1, 100, ssc);

		/* free */
	cleanup:
		tts_free(ssc);
	}
	__except (EXCEPTION_EXECUTE_HANDLER) {
		LOG_PRINT("%s error\n", __FUNCTION__);
	}
	return NULL;
}

tts_main:

static int tts_main(SpeechSynsContext *ssc)
{
	int         ret = MSP_SUCCESS;
	/*
	*--Insufficient authorization, and there are 500 daily restrictions on the number of online calls
	*/
	const char* login_params = "appid = 596f2c50, work_dir = .";//登录参数,appid与msc库绑定,请勿随意改动

	/* 用户登录 */
	ret = MSPLogin(NULL, NULL, login_params); //第一个参数是用户名，第二个参数是密码，第三个参数是登录参数，用户名和密码可在http://www.xfyun.cn注册获取
	if (MSP_SUCCESS != ret)
	{
		printf("MSPLogin failed, error code: %d.\n", ret);
		MSPLogout();
		tts_free(ssc);
		return -1;
	}

	/* 文本合成 */
	printf("开始合成 ...\n");
	ret = text_to_speech(ssc->text, ssc->tmp_file, ssc->session_parm, ssc);
	if (MSP_SUCCESS != ret)
	{
		printf("text_to_speech failed, error code: %d.\n", ret);
		MSPLogout();
		tts_free(ssc);
		return -1;
	}

	printf("合成完毕\n");
	MSPLogout();	//退出登录
	return 0;
}

text_to_speech:

/* 文本合成 */
static int text_to_speech(const char* src_text, const char* des_path, const char* params, SpeechSynsContext *ssc)
{
	int          ret = -1;
	FILE*        fp = NULL;
	const char*  sessionID = NULL;
	unsigned int audio_len = 0;
	wave_pcm_hdr wav_hdr = default_wav_hdr;
	int          synth_status = MSP_TTS_FLAG_STILL_HAVE_DATA;

	if (NULL == src_text || NULL == des_path)
	{
		printf("params is error!\n");
		return ret;
	}

	fp = fopen(des_path, "wb");
	if (NULL == fp)
	{
		printf("fopen %s error.\n", des_path);
		return ret;
	}

	/* 开始合成 */
	sessionID = QTTSSessionBegin(params, &ret);
	if (MSP_SUCCESS != ret)
	{
		printf("QTTSSessionBegin failed, error code: %d.\n", ret);
		fclose(fp);
		return ret;
	}
	ret = QTTSTextPut(sessionID, src_text, (unsigned int)strlen(src_text), NULL);
	if (MSP_SUCCESS != ret)
	{
		printf("QTTSTextPut failed, error code: %d.\n", ret);
		QTTSSessionEnd(sessionID, "TextPutError");
		fclose(fp);
		return ret;
	}
	printf("正在合成 ...\n");
	fwrite(&wav_hdr, sizeof(wav_hdr), 1, fp); //添加wav音频头，使用采样率为16000
	while (1)
	{
		/* 获取合成音频 */
		const void *data = QTTSAudioGet(sessionID, &audio_len, &synth_status, &ret);
		if (MSP_SUCCESS != ret)
			break;
		if (NULL != data)
		{

			wav_hdr.data_size += audio_len; //计算data_size大小
			ssc->progress = CurrentPercent(wav_hdr.data_size, src_text, ssc->speed);
			callback_function((TTSCallbackFunction)ssc->callback_fun, 0, ssc->progress, ssc);
			fwrite(data, audio_len, 1, fp);
		}
		if (MSP_TTS_FLAG_DATA_END == synth_status)
			break;
#ifdef DEBUG
		printf(">");
#endif
		Sleep(150); //防止频繁占用CPU
	}

#ifdef DEBUG
	printf("\n");
#endif
	if (MSP_SUCCESS != ret)
	{
		printf("QTTSAudioGet failed, error code: %d.\n", ret);
		QTTSSessionEnd(sessionID, "AudioGetError");
		fclose(fp);
		return ret;
	}
#if DEBUG
	printf("Total length is %d\n", wav_hdr.data_size);
#endif // DEBUG

	/* 修正wav文件头数据的大小 */
	wav_hdr.size_8 += wav_hdr.data_size + (sizeof(wav_hdr) - 8);

	/* 将修正过的数据写回文件头部,音频文件为wav格式 */
	fseek(fp, 4, 0);
	fwrite(&wav_hdr.size_8, sizeof(wav_hdr.size_8), 1, fp); //写入size_8的值
	fseek(fp, 40, 0); //将文件指针偏移到存储data_size值的位置
	fwrite(&wav_hdr.data_size, sizeof(wav_hdr.data_size), 1, fp); //写入data_size的值
	fclose(fp);
	fp = NULL;

	/* 合成完毕 */
	ret = QTTSSessionEnd(sessionID, "Normal");
	if (MSP_SUCCESS != ret)
	{
		printf("QTTSSessionEnd failed, error code: %d.\n", ret);
	}
	return ret;
}

tts_parm_copy:

/*
 *Warning*
 *malloc strlen(src) + 1
 *memset strlen(src) + 1
 *memcpy strlen(src)
*/
char* tts_parm_copy(const char *src)
{
	char *dst;
	int len = strlen(src) + 1;
	dst = (char*)malloc(len);
	if (dst == NULL) {
		return NULL;
	}
	memset(dst, 0, len);
	memcpy(dst, src, strlen(src));
	return dst;
}

tts_free:

static void tts_free(SpeechSynsContext *ssc)
{
	/* ssc->text(and so on) is const char*, so change it to void*, so can free the memory allocated by malloc */
	if (ssc->text) {
		free((void*)ssc->text);
	}
	if (ssc->out_path) {
		free((void*)ssc->out_path);
	}
	if (ssc->voice_name) {
		free((void*)ssc->voice_name);
	}
	if (ssc->volume) {
		free((void*)ssc->volume);
	}
	if (ssc->speed) {
		free((void*)ssc->speed);
	}
	if (ssc->pitch) {
		free((void*)ssc->pitch);
	}
	if (ssc->sample_rate) {
		free((void*)ssc->sample_rate);
	}
	if (ssc->text_encoding) {
		free((void*)ssc->text_encoding);
	}
	if (ssc->rdn) {
		free((void*)ssc->rdn);
	}
	if (ssc->session_parm) {
		free(ssc->session_parm);
	}
	if (ssc->options) {
		free(ssc->options);
	}
	if (ssc->tmp_file) {
		if (remove(ssc->tmp_file))
			LOG_PRINT("%s....Remove Temp file failed", __FUNCTION__);
		free(ssc->tmp_file);
	}
	if (ssc) {
		free(ssc);
	}
}

将wav转为mp3

transcode_audio.h

#ifndef TRANSCODE_AUDIO_H
#define	TRANSCODE_AUDIO_H

typedef void(*TranscodeCallbackFcn)(int, int, void*);
extern int transcode_audio(const char *inAudio, SpeechSynsContext *ssc);

transcode_audio.cpp

#include "stdafx.h"


#ifdef __cplusplus
extern "C" {
#endif

#include "libavutil/avassert.h"

#include "libavutil/audio_fifo.h"
#include "libavutil/avstring.h"
#include "libavutil/frame.h"
#include "libavutil/opt.h"
#include "libavutil/avassert.h"
#include "libswresample/swresample.h"

#include"logger.h"

#include "transcode_audio.h"

#ifdef __cplusplus
};
#endif


/** The output bit rate in kbit/s */
#define OUTPUT_BIT_RATE 25000
/** The number of output channels */
#define OUTPUT_CHANNELS 1

static int encode_audio_frame_flush(AVFrame *frame,
	AVFormatContext *output_format_context,
	AVCodecContext *output_codec_context,
	int *data_present);

static int current_percent(AVPacket pkt, AVFormatContext *fmt_ctx);

static void transcode_callback(TranscodeCallbackFcn fun, int status, int percent, void *identifier);


/** Open an input file and the required decoder. */
static int open_input_file(const char *filename,
	AVFormatContext **input_format_context,
	AVCodecContext **input_codec_context)
{
	AVCodecContext *avctx;
	AVCodec *input_codec;
	int error;

	/** Open the input file to read from it. */
	if ((error = avformat_open_input(input_format_context, filename, NULL,
		NULL)) < 0) {
		fprintf(stderr, "Could not open input file '%s' (error '%s')\n",
			filename, av_err2str(error));
		*input_format_context = NULL;
		return error;
	}

	/** Get information on the input file (number of streams etc.). */
	if ((error = avformat_find_stream_info(*input_format_context, NULL)) < 0) {
		fprintf(stderr, "Could not open find stream info (error '%s')\n",
			av_err2str(error));
		avformat_close_input(input_format_context);
		return error;
	}
	//为什么确保只有一个流
	/** Make sure that there is only one stream in the input file. */
	if ((*input_format_context)->nb_streams != 1) {
		fprintf(stderr, "Expected one audio input stream, but found %d\n",
			(*input_format_context)->nb_streams);
		avformat_close_input(input_format_context);
		return AVERROR_EXIT;
	}

	/** Find a decoder for the audio stream. */
	if (!(input_codec = avcodec_find_decoder((*input_format_context)->streams[0]->codecpar->codec_id))) {
		fprintf(stderr, "Could not find input codec\n");
		avformat_close_input(input_format_context);
		return AVERROR_EXIT;
	}
	/** allocate a new decoding context */
	avctx = avcodec_alloc_context3(input_codec);
	if (!avctx) {
		fprintf(stderr, "Could not allocate a decoding context\n");
		avformat_close_input(input_format_context);
		return AVERROR(ENOMEM);
	}
	/*int64_t i = (*input_format_context)->streams[0]->nb_frames;
	int64_t i2 = avctx->frame_number;*/
	/** initialize the stream parameters with demuxer information */
	error = avcodec_parameters_to_context(avctx, (*input_format_context)->streams[0]->codecpar);
	if (error < 0) {
		avformat_close_input(input_format_context);
		avcodec_free_context(&avctx);
		return error;
	}

	/** Open the decoder for the audio stream to use it later. */
	if ((error = avcodec_open2(avctx, input_codec, NULL)) < 0) {
		fprintf(stderr, "Could not open input codec (error '%s')\n",
			av_err2str(error));
		avcodec_free_context(&avctx);
		avformat_close_input(input_format_context);
		return error;
	}

	/** Save the decoder context for easier access later. */
	*input_codec_context = avctx;

	return 0;
}

/**
* Open an output file and the required encoder.
* Also set some basic encoder parameters.
* Some of these parameters are based on the input file's parameters.
*/
static int open_output_file(const char *filename,
	AVCodecContext *input_codec_context,
	AVFormatContext **output_format_context,
	AVCodecContext **output_codec_context)
{
	AVCodecContext *avctx = NULL;
	AVIOContext *output_io_context = NULL;
	AVStream *stream = NULL;
	AVCodec *output_codec = NULL;
	int error;
	/** Open the output file to write to it. */
	if ((error = avio_open(&output_io_context, filename,
		AVIO_FLAG_WRITE)) < 0) {
		fprintf(stderr, "Could not open output file '%s' (error '%s')\n",
			filename, av_err2str(error));
		return error;
	}

	/** Create a new format context for the output container format. */
	if (!(*output_format_context = avformat_alloc_context())) {
		fprintf(stderr, "Could not allocate output format context\n");
		return AVERROR(ENOMEM);
	}

	/** Associate the output file (pointer) with the container format context. */
	(*output_format_context)->pb = output_io_context;

	/** Guess the desired container format based on the file extension. */
	if (!((*output_format_context)->oformat = av_guess_format(NULL, filename,
		NULL))) {
		fprintf(stderr, "Could not find output file format\n");
		goto cleanup;
	}

	av_strlcpy((*output_format_context)->filename, filename,
		sizeof((*output_format_context)->filename));

	/** Find the encoder to be used by its name. */
	if (!(output_codec = avcodec_find_encoder(AV_CODEC_ID_MP3))) {
		fprintf(stderr, "Could not find an AAC encoder.\n");
		goto cleanup;
	}

	/** Create a new audio stream in the output file container. */
	if (!(stream = avformat_new_stream(*output_format_context, NULL))) {
		fprintf(stderr, "Could not create new stream\n");
		error = AVERROR(ENOMEM);
		goto cleanup;
	}

	avctx = avcodec_alloc_context3(output_codec);
	if (!avctx) {
		fprintf(stderr, "Could not allocate an encoding context\n");
		error = AVERROR(ENOMEM);
		goto cleanup;
	}

	/**
	* Set the basic encoder parameters.
	* The input file's sample rate is used to avoid a sample rate conversion.
	*/
	avctx->channels = OUTPUT_CHANNELS;
	avctx->channel_layout = av_get_default_channel_layout(OUTPUT_CHANNELS);
	avctx->sample_rate = input_codec_context->sample_rate;
	avctx->sample_fmt = output_codec->sample_fmts[0];
	avctx->bit_rate = OUTPUT_BIT_RATE;

	/** Allow the use of the experimental AAC encoder */
	avctx->strict_std_compliance = FF_COMPLIANCE_EXPERIMENTAL;

	/** Set the sample rate for the container. */
	stream->time_base.den = input_codec_context->sample_rate;
	stream->time_base.num = 1;

	/**
	* Some container formats (like MP4) require global headers to be present
	* Mark the encoder so that it behaves accordingly.
	*/
	if ((*output_format_context)->oformat->flags & AVFMT_GLOBALHEADER)
		avctx->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;

	/** Open the encoder for the audio stream to use it later. */
	if ((error = avcodec_open2(avctx, output_codec, NULL)) < 0) {
		fprintf(stderr, "Could not open output codec (error '%s')\n",
			av_err2str(error));
		goto cleanup;
	}

	error = avcodec_parameters_from_context(stream->codecpar, avctx);
	if (error < 0) {
		fprintf(stderr, "Could not initialize stream parameters\n");
		goto cleanup;
	}
	/** Save the encoder context for easier access later. */
	*output_codec_context = avctx;

	return 0;

cleanup:
	avcodec_free_context(&avctx);
	avio_closep(&(*output_format_context)->pb);
	avformat_free_context(*output_format_context);
	*output_format_context = NULL;
	return error < 0 ? error : AVERROR_EXIT;
}

/** Initialize one data packet for reading or writing. */
static void init_packet(AVPacket *packet)
{
	av_init_packet(packet);
	/** Set the packet data and size so that it is recognized as being empty. */
	packet->data = NULL;
	packet->size = 0;
}

/** Initialize one audio frame for reading from the input file */
static int init_input_frame(AVFrame **frame)
{
	if (!(*frame = av_frame_alloc())) {
		fprintf(stderr, "Could not allocate input frame\n");
		return AVERROR(ENOMEM);
	}
	return 0;
}

/**
* Initialize the audio resampler based on the input and output codec settings.
* If the input and output sample formats differ, a conversion is required
* libswresample takes care of this, but requires initialization.
*/
static int init_resampler(AVCodecContext *input_codec_context,
	AVCodecContext *output_codec_context,
	SwrContext **resample_context)
{
	int error;

	/**
	* Create a resampler context for the conversion.
	* Set the conversion parameters.
	* Default channel layouts based on the number of channels
	* are assumed for simplicity (they are sometimes not detected
	* properly by the demuxer and/or decoder).
	*/
	*resample_context = swr_alloc_set_opts(NULL,
		av_get_default_channel_layout(output_codec_context->channels),
		output_codec_context->sample_fmt,
		output_codec_context->sample_rate,
		av_get_default_channel_layout(input_codec_context->channels),
		input_codec_context->sample_fmt,
		input_codec_context->sample_rate,
		0, NULL);
	if (!*resample_context) {
		fprintf(stderr, "Could not allocate resample context\n");
		return AVERROR(ENOMEM);
	}
	/**
	* Perform a sanity check so that the number of converted samples is
	* not greater than the number of samples to be converted.
	* If the sample rates differ, this case has to be handled differently
	*/
	av_assert0(output_codec_context->sample_rate == input_codec_context->sample_rate);

	/** Open the resampler with the specified parameters. */
	if ((error = swr_init(*resample_context)) < 0) {
		fprintf(stderr, "Could not open resample context\n");
		swr_free(resample_context);
		return error;
	}
	return 0;
}

/** Initialize a FIFO buffer for the audio samples to be encoded. */
static int init_fifo(AVAudioFifo **fifo, AVCodecContext *output_codec_context)
{
	/** Create the FIFO buffer based on the specified output sample format. */
	if (!(*fifo = av_audio_fifo_alloc(output_codec_context->sample_fmt,
		output_codec_context->channels, 1))) {
		fprintf(stderr, "Could not allocate FIFO\n");
		return AVERROR(ENOMEM);
	}
	return 0;
}

/** Write the header of the output file container. */
static int write_output_file_header(AVFormatContext *output_format_context)
{
	int error;
	if ((error = avformat_write_header(output_format_context, NULL)) < 0) {
		fprintf(stderr, "Could not write output file header (error '%s')\n",
			av_err2str(error));
		return error;
	}
	return 0;
}

/** Decode one audio frame from the input file. */
static int decode_audio_frame(AVFrame *frame,
	AVFormatContext *input_format_context,
	AVCodecContext *input_codec_context,
	int *data_present, int *finished,
	SpeechSynsContext *ssc)
{
	/** Packet used for temporary storage. */
	AVPacket input_packet;
	int error;
	init_packet(&input_packet);
	int recv_err, send_err;
	/* */
	*data_present = 1;
	/* Ensure function current_percent() performed normally */
	if (input_format_context == NULL) {
		LOG_PRINT("%s....input_format_context is NULL", __FUNCTION__);
		return -1;
	}
	/** Read one audio frame from the input file into a temporary packet. */
	if ((error = av_read_frame(input_format_context, &input_packet)) < 0) {
		/** If we are at the end of the file, flush the decoder below. */
		if (error == AVERROR_EOF)
			*finished = 1;
		else {
			fprintf(stderr, "Could not read frame (error '%s')\n",
				av_err2str(error));
			return error;
		}
	}

	ssc->progress = current_percent(input_packet, input_format_context);
	transcode_callback((TranscodeCallbackFcn)ssc->callback_fun, 0, ssc->progress, ssc);

	/**
	* Decode the audio frame stored in the temporary packet.
	* The input audio stream decoder is used to do this.
	* If we are at the end of the file, pass an empty packet to the decoder
	* to flush it.
	*/

	/* if ((error = avcodec_decode_audio4(input_codec_context, frame,
	data_present, &input_packet)) < 0) {
	fprintf(stderr, "Could not decode frame (error '%s')\n",
	av_err2str(error));
	av_packet_unref(&input_packet);
	return error;
	}*/
	if ((send_err = avcodec_send_packet(input_codec_context, &input_packet)) < 0) {
		fprintf(stderr, "avcodec_send_packet failed (error '%s')\n",
			av_err2str(send_err));
	}
	if ((recv_err = avcodec_receive_frame(input_codec_context, frame)) < 0) {
		LOG_PRINT("%s....Decode flush finished", __FUNCTION__);
		if (recv_err == AVERROR_EOF) {
			*data_present = 0;
		}
		else {
			printf("recv_err: %d\n", recv_err);
			fprintf(stderr, "avcodec_receive_frame failed (error '%s')\n");
			av_err2str(recv_err);
		}
	}
	/*if (recv_err == AVERROR_EOF) {
	*data_present = 0;
	}*/
	if (send_err || recv_err) {
		av_packet_unref(&input_packet);
		return 0;
	}
	/**
	* If the decoder has not been flushed completely, we are not finished,
	* so that this function has to be called again.
	*/
	if (*finished && *data_present)
		*finished = 0;
	av_packet_unref(&input_packet);
	return 0;
}

/**
* Initialize a temporary storage for the specified number of audio samples.
* The conversion requires temporary storage due to the different format.
* The number of audio samples to be allocated is specified in frame_size.
*/
static int init_converted_samples(uint8_t ***converted_input_samples,
	AVCodecContext *output_codec_context,
	int frame_size)
{
	int error;

	/**
	* Allocate as many pointers as there are audio channels.
	* Each pointer will later point to the audio samples of the corresponding
	* channels (although it may be NULL for interleaved formats).
	*/
	if (!(*converted_input_samples = (uint8_t **)calloc(output_codec_context->channels,
		sizeof(**converted_input_samples)))) {
		fprintf(stderr, "Could not allocate converted input sample pointers\n");
		return AVERROR(ENOMEM);
	}

	/**
	* Allocate memory for the samples of all channels in one consecutive
	* block for convenience.
	*/
	if ((error = av_samples_alloc(*converted_input_samples, NULL,
		output_codec_context->channels,
		frame_size,
		output_codec_context->sample_fmt, 0)) < 0) {
		fprintf(stderr,
			"Could not allocate converted input samples (error '%s')\n",
			av_err2str(error));
		av_freep(&(*converted_input_samples)[0]);
		free(*converted_input_samples);
		return error;
	}
	return 0;
}

/**
* Convert the input audio samples into the output sample format.
* The conversion happens on a per-frame basis, the size of which is specified
* by frame_size.
*/
static int convert_samples(const uint8_t **input_data,
	uint8_t **converted_data, const int frame_size,
	SwrContext *resample_context)
{
	int error;

	/** Convert the samples using the resampler. */
	if ((error = swr_convert(resample_context,
		converted_data, frame_size,
		input_data, frame_size)) < 0) {
		fprintf(stderr, "Could not convert input samples (error '%s')\n",
			av_err2str(error));
		return error;
	}

	return 0;
}

/** Add converted input audio samples to the FIFO buffer for later processing. */
static int add_samples_to_fifo(AVAudioFifo *fifo,
	uint8_t **converted_input_samples,
	const int frame_size)
{
	int error;

	/**
	* Make the FIFO as large as it needs to be to hold both,
	* the old and the new samples.
	*/
	if ((error = av_audio_fifo_realloc(fifo, av_audio_fifo_size(fifo) + frame_size)) < 0) {
		fprintf(stderr, "Could not reallocate FIFO\n");
		return error;
	}

	/** Store the new samples in the FIFO buffer. */
	if (av_audio_fifo_write(fifo, (void **)converted_input_samples,
		frame_size) < frame_size) {
		fprintf(stderr, "Could not write data to FIFO\n");
		return AVERROR_EXIT;
	}
	return 0;
}

/**
* Read one audio frame from the input file, decodes, converts and stores
* it in the FIFO buffer.
*/
static int read_decode_convert_and_store(AVAudioFifo *fifo,
	AVFormatContext *input_format_context,
	AVCodecContext *input_codec_context,
	AVCodecContext *output_codec_context,
	SwrContext *resampler_context,
	int *finished,
	SpeechSynsContext *ssc)
{
	/** Temporary storage of the input samples of the frame read from the file. */
	AVFrame *input_frame = NULL;
	/** Temporary storage for the converted input samples. */
	uint8_t **converted_input_samples = NULL;
	int data_present;
	int ret = AVERROR_EXIT;

	/** Initialize temporary storage for one input frame. */
	if (init_input_frame(&input_frame)) {
		goto cleanup;
	}
	/** Decode one frame worth of audio samples. */
	if (decode_audio_frame(input_frame, input_format_context,
		input_codec_context, &data_present, finished,
		ssc))
		goto cleanup;
	/**
	* If we are at the end of the file and there are no more samples
	* in the decoder which are delayed, we are actually finished.
	* This must not be treated as an error.
	*/
	if (*finished && !data_present) {
		ret = 0;
		goto cleanup;
	}
	/** If there is decoded data, convert and store it */
	if (data_present) {
		/** Initialize the temporary storage for the converted input samples. */
		if (init_converted_samples(&converted_input_samples, output_codec_context,
			input_frame->nb_samples))
			goto cleanup;

		/**
		* Convert the input samples to the desired output sample format.
		* This requires a temporary storage provided by converted_input_samples.
		*/
		if (convert_samples((const uint8_t**)input_frame->extended_data, converted_input_samples,
			input_frame->nb_samples, resampler_context))
			goto cleanup;

		/** Add the converted input samples to the FIFO buffer for later processing. */
		if (add_samples_to_fifo(fifo, converted_input_samples,
			input_frame->nb_samples))
			goto cleanup;
		ret = 0;
	}
	ret = 0;

cleanup:
	if (converted_input_samples) {
		av_freep(&converted_input_samples[0]);
		free(converted_input_samples);
	}
	av_frame_free(&input_frame);

	return ret;
}

/**
* Initialize one input frame for writing to the output file.
* The frame will be exactly frame_size samples large.
*/
static int init_output_frame(AVFrame **frame,
	AVCodecContext *output_codec_context,
	int frame_size)
{
	int error;

	/** Create a new frame to store the audio samples. */
	if (!(*frame = av_frame_alloc())) {
		fprintf(stderr, "Could not allocate output frame\n");
		return AVERROR_EXIT;
	}

	/**
	* Set the frame's parameters, especially its size and format.
	* av_frame_get_buffer needs this to allocate memory for the
	* audio samples of the frame.
	* Default channel layouts based on the number of channels
	* are assumed for simplicity.
	*/
	(*frame)->nb_samples = frame_size;
	(*frame)->channel_layout = output_codec_context->channel_layout;
	(*frame)->format = output_codec_context->sample_fmt;
	(*frame)->sample_rate = output_codec_context->sample_rate;

	/**
	* Allocate the samples of the created frame. This call will make
	* sure that the audio frame can hold as many samples as specified.
	*/
	if ((error = av_frame_get_buffer(*frame, 0)) < 0) {
		fprintf(stderr, "Could not allocate output frame samples (error '%s')\n",
			av_err2str(error));
		av_frame_free(frame);
		return error;
	}

	return 0;
}

/** Global timestamp for the audio frames */
static int64_t pts = 0;

/** Encode one frame worth of audio to the output file. */
static int encode_audio_frame(AVFrame *frame,
	AVFormatContext *output_format_context,
	AVCodecContext *output_codec_context,
	int *data_present)
{
	/** Packet used for temporary storage. */
	AVPacket output_packet;
	int error;
	init_packet(&output_packet);

	int send_err, recv_err;

	/** Set a timestamp based on the sample rate for the container. */
	if (frame) {
		frame->pts = pts;
		pts += frame->nb_samples;
	}
	/* av_write_frame if 1, else jump out */
	*data_present = 1;
	/**
	* Encode the audio frame and store it in the temporary packet.
	* The output audio stream encoder is used to do this.
	*/
	//if ((error = avcodec_encode_audio2(output_codec_context, &output_packet,
	//                                   frame, data_present)) < 0) {
	//    fprintf(stderr, "Could not encode frame (error '%s')\n",
	//            av_err2str(error));
	//    av_packet_unref(&output_packet);
	//    return error;
	//}

	if ((send_err = avcodec_send_frame(output_codec_context, frame)) < 0) {
		fprintf(stderr, "avcodec_send_frame failed (error '%s')\n",
			av_err2str(send_err));
	}
	if ((recv_err = avcodec_receive_packet(output_codec_context, &output_packet)) < 0) {
		fprintf(stderr, "avcodec_receive_packet failed (error '%s')\n",
			av_err2str(recv_err));
	}
	if (send_err || recv_err) {
		if (send_err == AVERROR_EOF || recv_err == AVERROR_EOF) {
			*data_present = 0;
		}
		av_packet_unref(&output_packet);
		return 0;
	}
	/** Write one audio frame from the temporary packet to the output file. */
	if (*data_present) {
		if ((error = av_write_frame(output_format_context, &output_packet)) < 0) {
			fprintf(stderr, "Could not write frame (error '%s')\n",
				av_err2str(error));
			av_packet_unref(&output_packet);
			return error;
		}

		av_packet_unref(&output_packet);
	}

	return 0;
}

/**
* Load one audio frame from the FIFO buffer, encode and write it to the
* output file.
*/
static int load_encode_and_write(AVAudioFifo *fifo,
	AVFormatContext *output_format_context,
	AVCodecContext *output_codec_context)
{
	/** Temporary storage of the output samples of the frame written to the file. */
	AVFrame *output_frame;
	/**
	* Use the maximum number of possible samples per frame.
	* If there is less than the maximum possible frame size in the FIFO
	* buffer use this number. Otherwise, use the maximum possible frame size
	*/
	const int frame_size = FFMIN(av_audio_fifo_size(fifo),
		output_codec_context->frame_size);
	int data_written;

	/** Initialize temporary storage for one output frame. */
	if (init_output_frame(&output_frame, output_codec_context, frame_size))
		return AVERROR_EXIT;

	/**
	* Read as many samples from the FIFO buffer as required to fill the frame.
	* The samples are stored in the frame temporarily.
	*/
	if (av_audio_fifo_read(fifo, (void **)output_frame->data, frame_size) < frame_size) {
		fprintf(stderr, "Could not read data from FIFO\n");
		av_frame_free(&output_frame);
		return AVERROR_EXIT;
	}

	/** Encode one frame worth of audio samples. */
	if (encode_audio_frame(output_frame, output_format_context,
		output_codec_context, &data_written)) {
		av_frame_free(&output_frame);
		return AVERROR_EXIT;
	}
	av_frame_free(&output_frame);
	return 0;
}

/** Write the trailer of the output file container. */
static int write_output_file_trailer(AVFormatContext *output_format_context)
{
	int error;
	if ((error = av_write_trailer(output_format_context)) < 0) {
		fprintf(stderr, "Could not write output file trailer (error '%s')\n",
			av_err2str(error));
		return error;
	}
	return 0;
}

/** Convert an audio file to an AAC file in an MP4 container. */
int transcode_audio(const char *inAudio, SpeechSynsContext *ssc)
{
	AVFormatContext *input_format_context = NULL, *output_format_context = NULL;
	AVCodecContext *input_codec_context = NULL, *output_codec_context = NULL;
	SwrContext *resample_context = NULL;
	AVAudioFifo *fifo = NULL;
	int ret = AVERROR_EXIT;

	/* if (argc < 3) {
	fprintf(stderr, "Usage: %s <input file> <output file>\n", argv[0]);
	exit(1);
	}*/

	/** Register all codecs and formats so that they can be used. */
	av_register_all();
	/** Open the input file for reading. */
	if (open_input_file(inAudio, &input_format_context,
		&input_codec_context)) {
		LOG_PRINT("%s....open_input_file failed", __FUNCTION__);
		goto cleanup;
	}
	/** Open the output file for writing. */
	if (open_output_file(ssc->out_path, input_codec_context,
		&output_format_context, &output_codec_context)) {
		LOG_PRINT("%s....open_output_file failed", __FUNCTION__);
		goto cleanup;
	}
	/** Initialize the resampler to be able to convert audio sample formats. */
	if (init_resampler(input_codec_context, output_codec_context,
		&resample_context)) {
		LOG_PRINT("%s....init_resampler failed", __FUNCTION__);
		goto cleanup;
	}
	/** Initialize the FIFO buffer to store audio samples to be encoded. */
	if (init_fifo(&fifo, output_codec_context)) {
		LOG_PRINT("%s....init_fifo failed", __FUNCTION__);
		goto cleanup;
	}
	/** Write the header of the output file container. */
	if (write_output_file_header(output_format_context)) {
		LOG_PRINT("%s....write_output_file_header failed", __FUNCTION__);
		goto cleanup;
	}

	/**
	* Loop as long as we have input samples to read or output samples
	* to write; abort as soon as we have neither.
	*/
	while (1) {
		/** Use the encoder's desired frame size for processing. */
		const int output_frame_size = output_codec_context->frame_size;
		int finished = 0;

		/**
		* Make sure that there is one frame worth of samples in the FIFO
		* buffer so that the encoder can do its work.
		* Since the decoder's and the encoder's frame size may differ, we
		* need to FIFO buffer to store as many frames worth of input samples
		* that they make up at least one frame worth of output samples.
		*/
		while (av_audio_fifo_size(fifo) < output_frame_size) {
			/**
			* Decode one frame worth of audio samples, convert it to the
			* output sample format and put it into the FIFO buffer.
			*/
			if (read_decode_convert_and_store(fifo, input_format_context,
				input_codec_context,
				output_codec_context,
				resample_context, &finished,
				ssc)) {
				LOG_PRINT("%s....read_decode_convert_and_store failed", __FUNCTION__);
				goto cleanup;
			}

			/**
			* If we are at the end of the input file, we continue
			* encoding the remaining audio samples to the output file.
			*/
			if (finished)
				break;
		}

		/**
		* If we have enough samples for the encoder, we encode them.
		* At the end of the file, we pass the remaining samples to
		* the encoder.
		*/
		int i_tmp;
		while ((i_tmp = av_audio_fifo_size(fifo)) >= output_frame_size ||
			(finished && av_audio_fifo_size(fifo) > 0))
			/**
			* Take one frame worth of audio samples from the FIFO buffer,
			* encode it and write it to the output file.
			*/
			if (load_encode_and_write(fifo, output_format_context,
				output_codec_context)) {
				LOG_PRINT("%s....load_encode_and_write failed", __FUNCTION__);
				goto cleanup;
			}

		/**
		* If we are at the end of the input file and have encoded
		* all remaining samples, we can exit this loop and finish.
		*/
		if (finished) {
			int flush_not_done;
			/** Flush the encoder as it may have delayed frames. */
			//avcodec_send_frame(output_codec_context, NULL);
			do {
				if (encode_audio_frame_flush(NULL, output_format_context,
					output_codec_context, &flush_not_done)) {
					LOG_PRINT("%s....encode_audio_frame_flush failed", __FUNCTION__);
					goto cleanup;
				}
			} while (flush_not_done);
			LOG_PRINT("%s....Encode flush finished", __FUNCTION__);
			break;
		}
	}

	/** Write the trailer of the output file container. */
	if (write_output_file_trailer(output_format_context)) {
		LOG_PRINT("%s....write_output_file_trailer failed", __FUNCTION__);
		goto cleanup;
	}
	ret = 0;

	/* Set static variable s_cumulative_time to 0 because of EXIT */
	AVPacket pkt_in = { 0 };
	current_percent(pkt_in, NULL);

cleanup:
	if (fifo)
		av_audio_fifo_free(fifo);
	swr_free(&resample_context);
	if (output_codec_context)
		avcodec_free_context(&output_codec_context);
	if (output_format_context) {
		avio_closep(&output_format_context->pb);
		avformat_free_context(output_format_context);
	}
	if (input_codec_context)
		avcodec_free_context(&input_codec_context);
	if (input_format_context)
		avformat_close_input(&input_format_context);

	return ret;
}

static int encode_audio_frame_flush(AVFrame *frame,
	AVFormatContext *output_format_context,
	AVCodecContext *output_codec_context,
	int *flush_not_done)
{
	AVPacket output_packet;
	init_packet(&output_packet);
	//int flush_frame_cnt = 1;
	int ret;
	avcodec_send_frame(output_codec_context, NULL);
	Sleep(100);
	ret = avcodec_receive_packet(output_codec_context, &output_packet);
	if (ret == AVERROR_EOF) {
		*flush_not_done = 0;	//flush finished，exit
	}
	else if (ret == 0) {
		//vp->pkt->stream_index = vp->stream->index;
		ret = av_write_frame(output_format_context, &output_packet);
		if (ret == 0) {
#ifdef DEBUG
			printf("flush frame succeed %3d times\n", flush_frame_cnt++);
#endif
		}
		else if (ret < 0) {
			LOG_PRINT("%s....av_write_frame failed", __FUNCTION__);
			av_packet_unref(&output_packet);
			return -1;
		}
		else {
			//LOG_PRINT("%s....flushed and there is no more data to flush", __FUNCTION__);
		}
	}
	av_packet_unref(&output_packet);
	return 0;
}

static int current_percent(AVPacket pkt, AVFormatContext *fmt_ctx)
{
	static int64_t s_cumulative_time = 0;
	/* Set s_cumulative_time to 0 and exit when transcode finished */
	if (fmt_ctx == NULL) {
		s_cumulative_time = 0;
		return 0;
	}
	if (!fmt_ctx->streams[0]->duration) {
		LOG_PRINT("%s....Duration is 0", __FUNCTION__);
		return -1;
	}
	s_cumulative_time += pkt.duration;
	int cur_percent = (int)((float)s_cumulative_time / (float)fmt_ctx->streams[0]->duration * 20 + 80);
	return cur_percent;
}

static void transcode_callback(TranscodeCallbackFcn fun, int status, int percent, void *identifier)
{
	if (fun != NULL) {
		fun(status, percent, identifier);
	}
}