static VALUE ifly_asr_iat(VALUE klass, VALUE fileName) {
char *audioFileName = StringValue(fileName);
int ret = 0, len, status = 2, ep_status = 0, rec_status = 0, rslt_status = 0, i = 0;
char buff[kBufferSize], text[kBufferSize];
FILE *fin;
ret = QISRInit("appid=50ebc3c9,vad_enable=0");
if (ret != 0) {
rb_raise(rb_eRuntimeError, "ISR Initialize Error");
}
const char *sess_id = QISRSessionBegin("", "ssm=1,sub=iat,auf=audio/L16;rate=8000,aue=amr;-1,ent=sms8k,rst=plain,rse=utf8", &ret);
if (ret != 0){
rb_raise(rb_eRuntimeError, "ISR Session Begin Error");
}
fin = fopen(audioFileName, "rb");
if (!fin) {
rb_raise(rb_eRuntimeError, "ISR Can not open audio input file");
}
while(!feof(fin)) {
len = fread(buff, 1, kBufferSize, fin);
ret = QISRAudioWrite(sess_id, buff, len, status, &ep_status, &rec_status);
if (ret != 0) break;
usleep(200000);
}
fclose(fin);
status = 4;
ret = QISRAudioWrite(sess_id, buff, 1, status, &ep_status, &rec_status);
if (ret != 0) {
rb_raise(rb_eRuntimeError, "ISR Audio Last Frame Write Error");
}
do {
const char *result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if (ret != 0) break;
if (rslt_status != 1 && result) strcat(text, result);
usleep(500000);
} while (rslt_status != 5 && ++i < 30);
QISRSessionEnd(sess_id, NULL);
QISRFini();
return rb_str_new(text, strlen(text));
}
int IflySTT::open()
{
// sub: 请求业务类型
// domain: 领域
// language: 语言
// accent: 方言
// sample_rate: 音频采样率
// result_type: 识别结果格式
// result_encoding: 结果编码格式
// 详细参数说明请参阅《iFlytek MSC Reference Manual》
int ret = MSP_SUCCESS;
int errcode = MSP_SUCCESS;
const char* login_params = "appid = 5743ed12, work_dir = ."; // 登录参数，appid与msc库绑定,请勿随意改动
const char* session_begin_params = "sub = iat, domain = iat, language = zh_ch, accent = mandarin, sample_rate = 1411000, result_type = plain, result_encoding = gb2312";
// 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，均传NULL即可，第三个参数是登录参数
if (MSP_SUCCESS != ret)
{
return ret;
}
session_id_ = QISRSessionBegin(NULL, session_begin_params, &errcode); // 听写不需要语法，第一个参数为NULL
if (MSP_SUCCESS != errcode)
{
MSPLogout();
return errcode;
}
run_ = true;
_beginthread(iflystt_get_text_thread, 0, this);
return 0;
}
void run_iat( const char* session_begin_params)
{
const char* session_id = NULL;
char rec_result[BUFFER_SIZE] = {NULL};
char hints[HINTS_SIZE] = {NULL}; //hints为结束本次会话的原因描述，由用户自定义
unsigned int total_len = 0;
int aud_stat = MSP_AUDIO_SAMPLE_CONTINUE ; //音频状态
int ep_stat = MSP_EP_LOOKING_FOR_SPEECH; //端点检测
int rec_stat = MSP_REC_STATUS_SUCCESS ; //识别状态
int errcode = MSP_SUCCESS ;
FILE* f_pcm = NULL;
char* p_pcm = NULL;
long pcm_count = 0;
long pcm_size = 0;
long read_size = 0;
#if 0
char *file_name = "wav/0003.pcm";
if (NULL == file_name)
goto iat_exit;
f_pcm = fopen(file_name, "rb");
if (NULL == f_pcm)
{
printf("\nopen [%s] failed! \n", file_name);
goto iat_exit;
}
fseek(f_pcm, 0, SEEK_END);
pcm_size = ftell(f_pcm); //获取音频文件大小
fseek(f_pcm, 0, SEEK_SET);
p_pcm = (char *)malloc(pcm_size);
if (NULL == p_pcm)
{
printf("\nout of memory! \n");
goto iat_exit;
}
read_size = fread((void *)p_pcm, 1, pcm_size, f_pcm); //读取音频文件内容
if (read_size != pcm_size)
{
printf("\nread [%s] error!\n", file_name);
goto iat_exit;
}
printf("\n开始语音听写 ...\n");
session_id = QISRSessionBegin(NULL, session_begin_params, &errcode); //听写不需要语法，第一个参数为NULL
if (MSP_SUCCESS != errcode)
{
printf("\nQISRSessionBegin failed! error code:%d\n", errcode);
goto iat_exit;
}
char *p_data = p_pcm;
#else
WSADATA wsaData;
WORD sockVersion = MAKEWORD(2,2);
if(WSAStartup(sockVersion, &wsaData) != 0)
return;
SOCKET serSocket = socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
if(serSocket == INVALID_SOCKET)
{
printf("socket failed \n");
return ;
}
sockaddr_in serAddr;
serAddr.sin_family = AF_INET;
serAddr.sin_port = htons(port);
serAddr.sin_addr.S_un.S_addr = INADDR_ANY;
if(bind(serSocket, (LPSOCKADDR)&serAddr, sizeof(serAddr)) == SOCKET_ERROR)
{
printf("bind error!\n");
return ;
}
printf("服务端开启\n");
//开始监听
if(listen(serSocket, 5) == SOCKET_ERROR)
{
printf("listen error !");
return ;
}
SOCKET sClient;
sockaddr_in remoteAddr;
int nAddrLen = sizeof(remoteAddr);
sClient = accept(serSocket, (SOCKADDR *)&remoteAddr, &nAddrLen);
if(sClient == INVALID_SOCKET)
{
printf("accept error !");
}
printf("接受到一个连接：%s \r\n", inet_ntoa(remoteAddr.sin_addr));
unsigned int len = 10 * FRAME_LEN; // 每次写入200ms音频(16k，16bit)：1帧音频20ms，10帧=200ms。16k采样率的16位音频，一帧的大小为640Byte
char *space_buffer = (char*)malloc(SPACE_SIZE * len);
#endif
char *p_data = space_buffer;
char *p_data2 = space_buffer;
int rec_number = 0;
printf("\n开始语音听写 ...\n");
session_id = QISRSessionBegin(NULL, session_begin_params, &errcode); //听写不需要语法，第一个参数为NULL
if (MSP_SUCCESS != errcode)
{
printf("\nQISRSessionBegin failed! error code:%d\n", errcode);
goto iat_exit;
}
while (1)
{
int ret = 0;
#if 0
if (pcm_size < 2 * len)
len = pcm_size;
#endif
if (len <= 0)
break;
aud_stat = MSP_AUDIO_SAMPLE_CONTINUE;
if (0 == pcm_count)
aud_stat = MSP_AUDIO_SAMPLE_FIRST;
printf(">");
int total_recv = 0;
int recvlen = 0;
while (total_recv < len)
{
recvlen = recv(sClient, p_data, len - total_recv, 0);
total_recv += recvlen;
p_data += recvlen;
}
//printf("recv size %d \n", total_recv);
ret = QISRAudioWrite(session_id, (const void *)p_data2, len, aud_stat, &ep_stat, &rec_stat);
if (MSP_SUCCESS != ret)
{
printf("\nQISRAudioWrite failed! error code:%d\n", ret);
goto iat_exit;
}
if (MSP_REC_STATUS_SUCCESS == rec_stat) //已经有部分听写结果
{
printf("识别出结果了 %d\n", rec_number);
const char *rslt = QISRGetResult(session_id, &rec_stat, 0, &errcode);
if (MSP_SUCCESS != errcode)
{
printf("\nQISRGetResult failed! error code: %d\n", errcode);
goto iat_exit;
}
if (NULL != rslt)
{
unsigned int rslt_len = strlen(rslt);
total_len += rslt_len;
if (total_len >= BUFFER_SIZE)
{
printf("\nno enough buffer for rec_result !\n");
goto iat_exit;
}
strncat(rec_result, rslt, rslt_len);
printf("%s\n",rec_result);
}
}
else
{
printf("还没有听写结果 %d\n", rec_number);
}
pcm_count += (long)len;
//pcm_size -= (long)len;
p_data2 += (long)len;
if (MSP_EP_AFTER_SPEECH == ep_stat)
{
printf("loop = %d, 检测到后端点\n", rec_number);
break;
}
rec_number++;
Sleep(200); //模拟人说话时间间隙。200ms对应10帧的音频
}
printf("\n语音听写结束\n");
printf("=============================================================\n");
printf("%s\n",rec_result);
printf("=============================================================\n");
iat_exit:
if (NULL != f_pcm)
{
fclose(f_pcm);
f_pcm = NULL;
}
if (NULL != p_pcm)
{ free(p_pcm);
p_pcm = NULL;
}
QISRSessionEnd(session_id, hints);
}
int run_asr(const char* asrfile , const char* param , const char* grammar)
{
char rec_result[1024*4] = {0};
const char *sessionID;
FILE *f_pcm = NULL;
char *pPCM = NULL;
int lastAudio = 0 ;
int audStat = 2 ;
int epStatus = 0;
int recStatus = 0 ;
long pcmCount = 0;
long pcmSize = 0;
int ret = 0 ;
sessionID = QISRSessionBegin(NULL, param, &ret); //asr
if(ret !=0)
{
printf("QISRSessionBegin Failed,ret=%d\n",ret);
}
ret = QISRGrammarActivate(sessionID, grammar, NULL, 0);//可以选择在QISRSessionBegin第一个参数传入grammar，亦可通过QISRGrammarActivate激活语法，可以多次调用QISRGrammarActivate，激活多个语法。
if(ret !=0)
{
printf("QISRGrammarActivate Failed,ret=%d\n",ret);
}
f_pcm = fopen(asrfile, "rb");
if (NULL != f_pcm) {
fseek(f_pcm, 0, SEEK_END);
pcmSize = ftell(f_pcm);
fseek(f_pcm, 0, SEEK_SET);
pPCM = (char *)malloc(pcmSize);
fread((void *)pPCM, pcmSize, 1, f_pcm);
fclose(f_pcm);
f_pcm = NULL;
}
while (1) {
unsigned int len = 6400;
if (pcmSize < 12800) {
len = pcmSize;
lastAudio = 1;
}
audStat = 2;
if (pcmCount == 0)
audStat = 1;
if (0) {
if (audStat == 1)
audStat = 5;
else
audStat = 4;
}
if (len<=0)
{
break;
}
printf("csid=%s,count=%d,aus=%d,",sessionID,pcmCount/len,audStat);
ret = QISRAudioWrite(sessionID, (const void *)&pPCM[pcmCount], len, audStat, &epStatus, &recStatus);
printf("eps=%d,rss=%d,ret=%d\n",epStatus,recStatus,ret);
if (ret != 0)
break;
pcmCount += (long)len;
pcmSize -= (long)len;
if (recStatus == 0) {
const char *rslt = QISRGetResult(sessionID, &recStatus, 0, &ret);
if(ret !=0)
{
printf("QISRGetResult Failed,ret=%d\n",ret);
break;
}
if (NULL != rslt)
printf("%s\n", rslt);
}
if (epStatus == MSP_EP_AFTER_SPEECH)
break;
usleep(150000);
}
ret=QISRAudioWrite(sessionID, (const void *)NULL, 0, 4, &epStatus, &recStatus);
if (ret !=0)
{
printf("QISRAudioWrite Failed,ret=%d\n",ret);
}
free(pPCM);
pPCM = NULL;
while (recStatus != 5 && ret == 0) {
const char *rslt = QISRGetResult(sessionID, &recStatus, 0, &ret);
if (NULL != rslt)
{
strcat(rec_result,rslt);
}
usleep(150000);
}
ret=QISRSessionEnd(sessionID, NULL);
if(ret !=MSP_SUCCESS)
{
printf("QISRSessionEnd Failed, ret=%d\n",ret);
} printf("=============================================================\n");
printf("The result is: %s\n",rec_result);
printf("=============================================================\n");
usleep(100000);
}
int run_asr(const char* grammar , const char* asrfile)
{
int ret = MSP_SUCCESS;
int i = 0;
FILE* fp = NULL;
char buff[BUFFER_NUM];
unsigned int len;
int status = MSP_AUDIO_SAMPLE_CONTINUE, ep_status = -1, rec_status = -1, rslt_status = -1;
const char* param = "rst=plain,sub=asr,ssm=1,aue=speex,auf=audio/L16;rate=16000,grt=abnf";//注意sub=asr,grt=abnf
const char* sess_id = QISRSessionBegin(grammar, param, &ret);//将语法传入QISRSessionBegin
if ( MSP_SUCCESS != ret )
{
printf("QISRSessionBegin err %d\n", ret);
return ret;
}
fp = fopen( asrfile , "rb");//我们提供了几个音频文件，测试时根据需要在这里更换
if ( NULL == fp )
{
printf("failed to open file,please check the file.\n");
QISRSessionEnd(sess_id, "normal");
return -1;
}
printf("writing audio...\n");
while ( !feof(fp) )
{
len = (unsigned int)fread(buff, 1, BUFFER_NUM, fp);
feof(fp) ? status = MSP_AUDIO_SAMPLE_LAST : status = MSP_AUDIO_SAMPLE_CONTINUE;//最后一块音频要使用last
ret = QISRAudioWrite(sess_id, buff, len, status, &ep_status, &rec_status);
if ( ret != MSP_SUCCESS )
{
printf("\nQISRAudioWrite err %d\n", ret);
break;
}
if ( rec_status == MSP_REC_STATUS_SUCCESS )
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if (ret != MSP_SUCCESS )
{
printf("error code: %d\n", ret);
break;
}
else if( rslt_status == MSP_REC_STATUS_NO_MATCH )
printf("get result nomatch\n");
else
{
if ( result != NULL )
printf("get result[%d/%d]:len:%d\n %s\n", ret, rslt_status,strlen(result), result);
}
}
printf(".");
Sleep(200);//因为是模拟录音，为了避免数据发送太快造成缓冲区溢出，所以这里暂停200ms，如果是实时录音，不必暂停
}
printf("\n");
if (ret == MSP_SUCCESS)
{
printf("get reuslt\n");
char asr_result[1024] = "";
unsigned int pos_of_result = 0;
int loop_count = 0;
do
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if ( ret != 0 )
{
printf("QISRGetResult err %d\n", ret);
break;
}
if( rslt_status == MSP_REC_STATUS_NO_MATCH )
{
printf("get result nomatch\n");
}
else if ( result != NULL )
{
printf("[%d]:get result[%d/%d]: %s\n", (loop_count), ret, rslt_status, result);
strcpy(asr_result+pos_of_result,result);
pos_of_result += (unsigned int)strlen(result);
}
else
{
printf("[%d]:get result[%d/%d]\n",(loop_count), ret, rslt_status);
}
Sleep(500);
} while (rslt_status != MSP_REC_STATUS_COMPLETE && loop_count++ < 30);
if (strcmp(asr_result,"")==0)
{
printf("no result\n");
}
}
QISRSessionEnd(sess_id, NULL);
printf("QISRSessionEnd.\n");
fclose(fp);
return 0;
}
int run_asr(UserData *udata)
{
char asr_params[MAX_PARAMS_LEN] = {NULL};
//const char *rec_rslt = NULL;
const char *session_id = NULL;
//const char *asr_audiof = NULL;
FILE *f_pcm = NULL;
char *pcm_data = NULL;
long pcm_count = 0;
long pcm_size = 0;
int last_audio = 0;
int aud_stat = MSP_AUDIO_SAMPLE_CONTINUE;
int ep_status = MSP_EP_LOOKING_FOR_SPEECH;
int rec_status = MSP_REC_STATUS_INCOMPLETE;
int rss_status = MSP_REC_STATUS_INCOMPLETE;
int errcode = -1;
if (record())
printf("Finish recording!!\n");
else
{
printf("Fail to record!\n");
return 1;
}
// asr_audiof = filename;
f_pcm = fopen(filename, "rb");
if (NULL == f_pcm) {
printf("打开\"%s\"失败！[%s]\n", f_pcm, strerror(errno));
goto run_error;
}
fseek(f_pcm, 0, SEEK_END);
pcm_size = ftell(f_pcm);
fseek(f_pcm, 0, SEEK_SET);
pcm_data = (char *)malloc(pcm_size);
if (NULL == pcm_data)
goto run_error;
fread((void *)pcm_data, pcm_size, 1, f_pcm);
fclose(f_pcm);
f_pcm = NULL;
//离线语法识别参数设置
snprintf(asr_params, MAX_PARAMS_LEN - 1,
"engine_type = local, \
asr_res_path = %s, sample_rate = %d, \
grm_build_path = %s, local_grammar = %s, \
result_type = xml, result_encoding = utf-8, ",
ASR_RES_PATH,
SAMPLE_RATE_16K,
GRM_BUILD_PATH,
udata->grammar_id
);
session_id = QISRSessionBegin(NULL, asr_params, &errcode);
if (NULL == session_id)
goto run_error;
printf("开始识别...\n");
while (1) {
unsigned int len = 6400;
if (pcm_size < 12800) {
len = pcm_size;
last_audio = 1;
}
aud_stat = MSP_AUDIO_SAMPLE_CONTINUE;
if (0 == pcm_count)
aud_stat = MSP_AUDIO_SAMPLE_FIRST;
if (len <= 0)
break;
printf(">");
fflush(stdout);
errcode = QISRAudioWrite(session_id, (const void *)&pcm_data[pcm_count], len, aud_stat, &ep_status, &rec_status);
if (MSP_SUCCESS != errcode)
goto run_error;
pcm_count += (long)len;
pcm_size -= (long)len;
//检测到音频结束
if (MSP_EP_AFTER_SPEECH == ep_status)
break;
usleep(150 * 1000); //模拟人说话时间间隙
}
//主动点击音频结束
QISRAudioWrite(session_id, (const void *)NULL, 0, MSP_AUDIO_SAMPLE_LAST, &ep_status, &rec_status);
free(pcm_data);
pcm_data = NULL;
//获取识别结果
while (MSP_REC_STATUS_COMPLETE != rss_status && MSP_SUCCESS == errcode) {
rec_rslt = QISRGetResult(session_id, &rss_status, 0, &errcode);
usleep(150 * 1000);
}
printf("\n识别结束：\n");
printf("=============================================================\n");
if (NULL != rec_rslt)
{
printf("%s\n", rec_rslt);
if(strstr(rec_rslt,tmpt1))
flag1=true;
else if (strstr(rec_rslt,tmpt2))
flag2=true;
}
else
printf("没有识别结果！\n");
printf("=============================================================\n");
goto run_exit;
run_error:
if (NULL != pcm_data) {
free(pcm_data);
pcm_data = NULL;
}
if (NULL != f_pcm) {
fclose(f_pcm);
f_pcm = NULL;
}
run_exit:
QISRSessionEnd(session_id, NULL);
return errcode;
}
void run_asr(const char* audio_file, const char* params, char* grammar_id)
{
const char* session_id = NULL;
char rec_result[BUFFER_SIZE] = {'\0'};
char hints[HINTS_SIZE] = {'\0'}; //hints为结束本次会话的原因描述，由用户自定义
unsigned int total_len = 0;
int aud_stat = MSP_AUDIO_SAMPLE_CONTINUE; //音频状态
int ep_stat = MSP_EP_LOOKING_FOR_SPEECH; //端点检测
int rec_stat = MSP_REC_STATUS_SUCCESS; //识别状态
int errcode = MSP_SUCCESS;
FILE* f_pcm = NULL;
char* p_pcm = NULL;
long pcm_count = 0;
long pcm_size = 0;
long read_size = 0;
if (NULL == audio_file)
goto asr_exit;
f_pcm = fopen(audio_file, "rb");
if (NULL == f_pcm)
{
printf("\nopen [%s] failed!\n", audio_file);
goto asr_exit;
}
fseek(f_pcm, 0, SEEK_END);
pcm_size = ftell(f_pcm); //获取音频文件大小
fseek(f_pcm, 0, SEEK_SET);
p_pcm = (char*)malloc(pcm_size);
if (NULL == p_pcm)
{
printf("\nout of memory!\n");
goto asr_exit;
}
read_size = fread((void *)p_pcm, 1, pcm_size, f_pcm); //读取音频文件内容
if (read_size != pcm_size)
{
printf("\nread [%s] failed!\n", audio_file);
goto asr_exit;
}
printf("\n开始语音识别 ...\n");
session_id = QISRSessionBegin(grammar_id, params, &errcode);
if (MSP_SUCCESS != errcode)
{
printf("\nQISRSessionBegin failed, error code:%d\n", errcode);
goto asr_exit;
}
while (1)
{
unsigned int len = 10 * FRAME_LEN; // 每次写入200ms音频(16k，16bit)：1帧音频20ms，10帧=200ms。16k采样率的16位音频，一帧的大小为640Byte
int ret = 0;
if (pcm_size < 2 * len)
len = pcm_size;
if (len <= 0)
break;
aud_stat = MSP_AUDIO_SAMPLE_CONTINUE;
if (0 == pcm_count)
aud_stat = MSP_AUDIO_SAMPLE_FIRST;
printf(">");
ret = QISRAudioWrite(session_id, (const void *)&p_pcm[pcm_count], len, aud_stat, &ep_stat, &rec_stat);
if (MSP_SUCCESS != ret)
{
printf("\nQISRAudioWrite failed, error code:%d\n",ret);
goto asr_exit;
}
pcm_count += (long)len;
pcm_size -= (long)len;
if (MSP_EP_AFTER_SPEECH == ep_stat)
break;
usleep(200*1000); //模拟人说话时间间隙，10帧的音频长度为200ms
}
errcode = QISRAudioWrite(session_id, NULL, 0, MSP_AUDIO_SAMPLE_LAST, &ep_stat, &rec_stat);
if (MSP_SUCCESS != errcode)
{
printf("\nQISRAudioWrite failed, error code:%d\n",errcode);
goto asr_exit;
}
while (MSP_REC_STATUS_COMPLETE != rec_stat)
{
const char *rslt = QISRGetResult(session_id, &rec_stat, 0, &errcode);
if (MSP_SUCCESS != errcode)
{
printf("\nQISRGetResult failed, error code: %d\n", errcode);
goto asr_exit;
}
if (NULL != rslt)
{
unsigned int rslt_len = strlen(rslt);
total_len += rslt_len;
if (total_len >= BUFFER_SIZE)
{
printf("\nno enough buffer for rec_result !\n");
goto asr_exit;
}
strncat(rec_result, rslt, rslt_len);
}
usleep(150*1000); //防止频繁占用CPU
}
printf("\n语音识别结束\n");
printf("=============================================================\n");
printf("%s",rec_result);
printf("=============================================================\n");
asr_exit:
if (NULL != f_pcm)
{
fclose(f_pcm);
f_pcm = NULL;
}
if (NULL != p_pcm)
{
free(p_pcm);
p_pcm = NULL;
}
QISRSessionEnd(session_id, hints);
}
void run_iat()
{
bool error = false;
int ret = MSP_SUCCESS;
int i = 0;
FILE* fp = NULL;
FILE* fout = NULL;
char buff[BUFFER_NUM];
int len;
int status = MSP_AUDIO_SAMPLE_CONTINUE, ep_status = -1, rec_status = -1, rslt_status = -1;
///引擎初始化，只需初始化一次
///APPID请勿随意改动
ret = QISRInit("appid=510f2d72");
///第二个参数为传递的参数，使用会话模式，使用speex编解码，使用16k16bit的音频数据
///第三个参数为返回码
const char* param = "sub=iat,ssm=1,auf=audio/L16;rate=16000,aue=speex,ent=sms16k,rst=plain";
const char* sess_id = QISRSessionBegin(NULL, param, &ret);
if ( MSP_SUCCESS != ret )
{
printf("QISRSessionBegin err %d\n", ret);
error = true;
}
///模拟录音，输入音频
if (error == false)
{
fp = fopen("iat_demo_test.wav", "rb");
if ( NULL == fp )
{
printf("failed to open file,please check the file.\n");
error = true;
}
}
///结果输出到文件
if (error == false)
{
fout = fopen("iat_result.txt", "ab");
if( NULL == fout )
{
printf("failed to open file,please check the file.\n");
error = true;
}
}
if (error == false)
{
printf("writing audio...\n");
char param_value[32] = "";//参数值的字符串形式
size_t value_len = 32; //字符串长度或buffer长度
int volume = 0;//音量数值
while ( !feof(fp) )
{
len = fread(buff, 1, BUFFER_NUM, fp);
printf(".");
///开始向服务器发送音频数据
ret = QISRAudioWrite(sess_id, buff, len, status, &ep_status, &rec_status);
if ( ret != MSP_SUCCESS )
{
printf("\nQISRAudioWrite err %d\n", ret);
error = true;
break;
}
/*********获取当前发送音频的音量信息**********/
value_len = 32;//value_len既是传入参数，又是传出参数，每次调用QTTSGetParam时要调整为buffer长度
ret = QISRGetParam(sess_id,"volume",param_value,&value_len);//获取音量信息，获取上行流量和下行流量的例子见ttsdemo
if ( ret != MSP_SUCCESS )
{
printf("QISRGetParam: qisr get param failed Error code %d.\n",ret);
char key = _getch();
break;
}
volume = atoi(param_value);//获取到的音量信息可以用于在界面上用不同的图片展示动态效果
//printf("volume== %d \n",volume);
for (int i=0;i<volume;i++)
{
printf(".");
}
printf("\n");
/*******获取当前发送音频的音量信息结束**********/
if (ep_status == MSP_EP_AFTER_SPEECH)//检测到音频后端点
{
printf("QISRAudioWrite: ep_status == MSP_EP_AFTER_SPEECH.\n");
break;
}
///服务器返回部分结果
if ( rec_status == MSP_REC_STATUS_SUCCESS )
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if( rslt_status == MSP_REC_STATUS_NO_MATCH )
printf("get result nomatch\n");
else
{
if ( result != NULL )
fwrite(result, 1, strlen(result), fout);
printf("get result[%d/%d]: %s\n", ret, rslt_status, result);
}
}
Sleep(200);
}
printf("\n");
fclose(fp);
}
///最后一块数据
if (error == false)
{
status = MSP_AUDIO_SAMPLE_LAST;
ret = QISRAudioWrite(sess_id, buff, 1, status, &ep_status, &rec_status);
if ( ret != MSP_SUCCESS )
{
printf("QISRAudioWrite write last audio err %d\n", ret);
error = true;
}
}
///最后一块数据发完之后，循环从服务器端获取结果
///考虑到网络环境不好的情况下，需要对循环次数作限定
if (error == false)
{
printf("get reuslt\n");
int loop_count = 0;
do
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if ( ret != MSP_SUCCESS )
{
printf("QISRGetResult err %d\n", ret);
error = true;
break;
}
if( rslt_status == MSP_REC_STATUS_NO_MATCH )
printf("get result nomatch\n");
else
{
if ( result != NULL )
fwrite(result, 1, strlen(result), fout);
printf("[%d]:get result[%d/%d]: %s\n", (loop_count), ret, rslt_status, result);
}
Sleep(500);
} while (rslt_status != MSP_REC_STATUS_COMPLETE && loop_count++ < 30);
}
if( NULL != fout )
{
fclose(fout);
}
ret = QISRSessionEnd(sess_id, NULL);
if ( ret != MSP_SUCCESS )
{
printf("QISRSessionEnd err %d\n", ret);
return;
}
printf("QISRSessionEnd.\n");
ret = QISRFini();
return;
}
void run_iat(const char* audio_file, const char* session_begin_params)
{
const char* session_id = NULL;
char rec_result[BUFFER_SIZE] = {NULL};
char hints[HINTS_SIZE] = {NULL}; //hints为结束本次会话的原因描述，由用户自定义
unsigned int total_len = 0;
int aud_stat = MSP_AUDIO_SAMPLE_CONTINUE ; //音频状态
int ep_stat = MSP_EP_LOOKING_FOR_SPEECH; //端点检测
int rec_stat = MSP_REC_STATUS_SUCCESS ; //识别状态
int errcode = MSP_SUCCESS ;
FILE* f_pcm = NULL;
char* p_pcm = NULL;
long pcm_count = 0;
long pcm_size = 0;
long read_size = 0;
if (NULL == audio_file)
goto iat_exit;
f_pcm = fopen(audio_file, "rb");
if (NULL == f_pcm)
{
printf("\nopen [%s] failed! \n", audio_file);
goto iat_exit;
}
fseek(f_pcm, 0, SEEK_END);
pcm_size = ftell(f_pcm); //获取音频文件大小
fseek(f_pcm, 0, SEEK_SET);
p_pcm = (char *)malloc(pcm_size);
if (NULL == p_pcm)
{
printf("\nout of memory! \n");
goto iat_exit;
}
read_size = fread((void *)p_pcm, 1, pcm_size, f_pcm); //读取音频文件内容
if (read_size != pcm_size)
{
printf("\nread [%s] error!\n", audio_file);
goto iat_exit;
}
printf("\n开始语音听写 ...\n");
session_id = QISRSessionBegin(NULL, session_begin_params, &errcode); //听写不需要语法，第一个参数为NULL
if (MSP_SUCCESS != errcode)
{
printf("\nQISRSessionBegin failed! error code:%d\n", errcode);
goto iat_exit;
}
char *p_data = p_pcm;
int rec_number = 0;
while (1)
{
unsigned int len = 10 * FRAME_LEN; // 每次写入200ms音频(16k，16bit)：1帧音频20ms，10帧=200ms。16k采样率的16位音频，一帧的大小为640Byte
int ret = 0;
if (pcm_size < 2 * len)
len = pcm_size;
if (len <= 0)
break;
aud_stat = MSP_AUDIO_SAMPLE_CONTINUE;
if (0 == pcm_count)
aud_stat = MSP_AUDIO_SAMPLE_FIRST;
printf(">");
ret = QISRAudioWrite(session_id, (const void *)p_data, len, aud_stat, &ep_stat, &rec_stat);
if (MSP_SUCCESS != ret)
{
printf("\nQISRAudioWrite failed! error code:%d\n", ret);
goto iat_exit;
}
if (MSP_REC_STATUS_SUCCESS == rec_stat) //已经有部分听写结果
{
const char *rslt = QISRGetResult(session_id, &rec_stat, 0, &errcode);
if (MSP_SUCCESS != errcode)
{
printf("\nQISRGetResult failed! error code: %d\n", errcode);
goto iat_exit;
}
if (NULL != rslt)
{
unsigned int rslt_len = strlen(rslt);
total_len += rslt_len;
if (total_len >= BUFFER_SIZE)
{
printf("\nno enough buffer for rec_result !\n");
goto iat_exit;
}
strncat(rec_result, rslt, rslt_len);
}
}
else
{
//printf("还没有听写结果\n");
}
pcm_count += (long)len;
//pcm_size -= (long)len;
p_data += (long)len;
rec_number++;
if (MSP_EP_AFTER_SPEECH == ep_stat)
{
printf("loop = %d, 检测到后端点\n", rec_number);
break;
}
Sleep(200); //模拟人说话时间间隙。200ms对应10帧的音频
}
printf("\n语音听写结束\n");
printf("=============================================================\n");
printf("%s\n",rec_result);
printf("=============================================================\n");
iat_exit:
if (NULL != f_pcm)
{
fclose(f_pcm);
f_pcm = NULL;
}
if (NULL != p_pcm)
{ free(p_pcm);
p_pcm = NULL;
}
QISRSessionEnd(session_id, hints);
}
void run_iat(const char* src_wav_filename , const char* des_text_filename , const char* param)
{
bool error = false;
int ret = MSP_SUCCESS;
int i = 0;
FILE* fp = NULL;
FILE* fout = NULL;
char buff[BUFFER_NUM];
unsigned int len;
int status = MSP_AUDIO_SAMPLE_CONTINUE, ep_status = -1, rec_status = -1, rslt_status = -1;
///第二个参数为传递的参数，使用会话模式，使用speex编解码，使用16k16bit的音频数据
///第三个参数为返回码
const char* sess_id = QISRSessionBegin(NULL, param, &ret);
if ( MSP_SUCCESS != ret )
{
printf("QISRSessionBegin err %d\n", ret);
error = true;
}
///模拟录音，输入音频
if (error == false)
{
fp = fopen( src_wav_filename , "rb");
if ( NULL == fp )
{
printf("failed to open file,please check the file.\n");
error = true;
}
}
///结果输出到文件
if (error == false)
{
fout = fopen( des_text_filename , "ab");
if( NULL == fout )
{
printf("failed to open file,please check the file.\n");
error = true;
}
}
if (error == false)
{
printf("writing audio...\n");
char param_value[32] = "";//参数值的字符串形式
unsigned int value_len = 32; //字符串长度或buffer长度
int volume = 0;//音量数值
while ( !feof(fp) )
{
len = (unsigned int)fread(buff, 1, BUFFER_NUM, fp);
printf(".");
feof(fp) ? status = MSP_AUDIO_SAMPLE_LAST : status = MSP_AUDIO_SAMPLE_CONTINUE;//最后一块音频要使用last
///开始向服务器发送音频数据
ret = QISRAudioWrite(sess_id, buff, len, status, &ep_status, &rec_status);
if ( ret != MSP_SUCCESS )
{
printf("\nQISRAudioWrite err %d\n", ret);
error = true;
break;
}
/*********获取当前发送音频的音量信息**********/
value_len = 32;//value_len既是传入参数，又是传出参数，每次调用QTTSGetParam时要调整为buffer长度
ret = QISRGetParam(sess_id,"volume",param_value,&value_len);//获取音量信息，获取上行流量和下行流量的例子见ttsdemo
if ( ret != MSP_SUCCESS )
{
printf("QISRGetParam: qisr get param failed Error code %d.\n",ret);
char key = getchar();
break;
}
volume = atoi(param_value);//获取到的音量信息可以用于在界面上用不同的图片展示动态效果
//printf("volume== %d \n",volume);
for (int i=0;i<volume;i++)
{
printf(".");
}
printf("\n");
/*******获取当前发送音频的音量信息结束**********/
if (ep_status == MSP_EP_AFTER_SPEECH)//检测到音频后端点
{
printf("QISRAudioWrite: ep_status == MSP_EP_AFTER_SPEECH.\n");
break;
}
///服务器返回部分结果
if ( rec_status == MSP_REC_STATUS_SUCCESS )
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if( rslt_status == MSP_REC_STATUS_NO_MATCH )
printf("get result nomatch\n");
else
{
if ( result != NULL )
fwrite(result, 1, strlen(result), fout);
printf("get result[%d/%d]: %s\n", ret, rslt_status, result);
}
}
usleep(200000);//因为是模拟录音，为了避免数据发送太快造成缓冲区溢出，所以这里暂停200ms，如果是实时录音，不必暂停
}
printf("\n");
fclose(fp);
}
///最后一块数据发完之后，循环从服务器端获取结果
///考虑到网络环境不好的情况下，可以对循环次数作限定
if (error == false)
{
printf("get reuslt\n");
int loop_count = 0;
do
{
const char* result = QISRGetResult(sess_id, &rslt_status, 0, &ret);
if ( ret != MSP_SUCCESS )
{
printf("QISRGetResult err %d\n", ret);
error = true;
break;
}
if( rslt_status == MSP_REC_STATUS_NO_MATCH )
printf("get result nomatch\n");
else
{
if ( result != NULL )
fwrite(result, 1, strlen(result), fout);
printf("[%d]:get result[%d/%d]: %s\n", (loop_count), ret, rslt_status, result);
}
usleep(500000);
} while (rslt_status != MSP_REC_STATUS_COMPLETE && loop_count++ < 30);
}
if( NULL != fout )
{
const char* result = "\r\n";
fwrite(result, 1, strlen(result), fout);
fclose(fout);
}
ret = QISRSessionEnd(sess_id, NULL);
if ( ret != MSP_SUCCESS )
{
printf("QISRSessionEnd err %d\n", ret);
return;
}
printf("QISRSessionEnd.\n");
return;
}
/*! function to open the asr interface */
static switch_status_t pocketsphinx_asr_open(switch_asr_handle_t *ah, const char *codec, int rate, const char *dest, switch_asr_flag_t *flags)
{
pocketsphinx_t *ps;
const char* session_begin_params = "sub = iat, domain = iat, language = zh_ch, accent = mandarin, sample_rate = 16000, result_type = plain, result_encoding = utf8";
int errcode;
/* chen */
switch_log_printf(SWITCH_CHANNEL_LOG, SWITCH_LOG_ERROR, ">>>>>>>>pocketsphinx_asr_open<<<<<<<<<\n");
if (!(ps = (pocketsphinx_t *) switch_core_alloc(ah->memory_pool, sizeof(*ps)))) {
return SWITCH_STATUS_MEMERR;
}
switch_mutex_init(&ps->flag_mutex, SWITCH_MUTEX_NESTED, ah->memory_pool);
ah->private_info = ps;
if (rate == 8000) {
ah->rate = 8000;
} else if (rate == 16000) {
ah->rate = 16000;
} else {
switch_log_printf(SWITCH_CHANNEL_LOG, SWITCH_LOG_ERROR, "Invalid rate %d. Only 8000 and 16000 are supported.\n", rate);
}
codec = "L16";
ah->codec = switch_core_strdup(ah->memory_pool, codec);
ps->thresh = globals.thresh;
ps->silence_hits = globals.silence_hits;
ps->listen_hits = globals.listen_hits;
ps->org_silence_hits = ps->silence_hits;
ps->start_input_timers = globals.start_input_timers;
ps->no_input_timeout = globals.no_input_timeout;
ps->speech_timeout = globals.speech_timeout;
ps->confidence_threshold = globals.confidence_threshold;
/* ify login*/
ps->ifly_session_id = NULL;
//ps->ifly_rec_result = {NULL};
//ps->ifly_hints = {NULL}; //hints为结束本次会话的原因描述，由用户自定义
ps->ifly_total_len = 0;
ps->ifly_aud_stat = MSP_AUDIO_SAMPLE_CONTINUE ; //音频状态
ps->ifly_ep_stat = MSP_EP_LOOKING_FOR_SPEECH; //端点检测
ps->ifly_rec_stat = MSP_REC_STATUS_SUCCESS ; //识别状态
ps->ifly_wait_result = SWITCH_FALSE;
/* ifly open session */
ps->ifly_session_id = QISRSessionBegin(NULL, session_begin_params, &errcode);
if (MSP_SUCCESS != errcode) {
switch_log_printf(SWITCH_CHANNEL_LOG, SWITCH_LOG_ERROR, ">>>>>>>>QISRSessionBegin fail error !!!<<<<<<<<<\n");
}
return SWITCH_STATUS_SUCCESS;
}