voice activity detection,VAD是啥意思

Posted

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了voice activity detection,VAD是啥意思相关的知识,希望对你有一定的参考价值。

voice activity detection,
话音激活检测
VAD = Voice Activity Detection
例句
Voice Activity Detection ( VAD ) algorithm based on multiresolution spectrum is presented.
针对静音检测提出基于小波变分辨率频谱特征的检测算法.
参考技术A VAD 是前几个单词的缩写。是声音活动情况探测的意思。 参考技术B 你好,很高兴为你解答

voice activity detection,
话音激活检测
VAD = Voice Activity Detection
例句
Voice Activity Detection ( VAD ) algorithm based on multiresolution spectrum is presented.
针对静音检测提出基于小波变分辨率频谱特征的检测算法.本回答被提问者和网友采纳

unimrcp-voice-activity语音检测

研究 unimrcp有一段时间了,其中unimrcp voice acitve的算法,是遭到大家频繁吐槽。今天我们简单的介绍一下unimrcp voice activity 的这个简单粗暴的算法:

 

unimrcp 语音活动检测是通过能量来控制的,设定几个常量:

struct mpf_activity_detector_t 
/* 静音检测阀值 */
apr_size_t level_threshold;

/* 转换成active状态的超时时长*/
apr_size_t speech_timeout;
/* 转换成inactive状态超时时长 */
apr_size_t silence_timeout;
/* 没有输入的超时时长 */
apr_size_t noinput_timeout;

/* 当前的状态 */
mpf_detector_state_e state;
/* 当前状态的保持的时长 */
apr_size_t duration;
;

 

来看一下这几个参数的初始化的值,根据实际的测试,我们后期做过改动:

/** Create activity detector */
MPF_DECLARE(mpf_activity_detector_t*) mpf_activity_detector_create(apr_pool_t *pool)

    mpf_activity_detector_t *detector = apr_palloc(pool,sizeof(mpf_activity_detector_t));
    detector->level_threshold = 50; /* 0 .. 255 */
    detector->speech_timeout = 300; /* 0.3 s  = 300*/
    detector->silence_timeout = 1000; /* 0.3 s  =300 */
    detector->noinput_timeout = 5000; /* 5 s =5000*/
    detector->duration = 0;
    detector->state = DETECTOR_STATE_INACTIVITY;
    return detector;

 

看一下重要的函数,能量的计算:

根据frame的信息,对能量进行粗暴的累加,所以对于噪音,这个算法完全不可用。后面将会介绍如何采用webrtc的voice active来取代这个算法。

static apr_size_t mpf_activity_detector_level_calculate(const mpf_frame_t *frame)

    apr_size_t sum = 0;
//计算多少个 apr_size_t count
= frame->codec_frame.size/2;
//初始值
const apr_int16_t *cur = frame->codec_frame.buffer;
//最后一个值
const apr_int16_t *end = cur + count; for(; cur < end; cur++) if(*cur < 0) sum -= *cur; else sum += *cur; //取平均值,简单粗暴,被吐槽的原因 return sum / count;

 

最后看一下,状态切换的过程,下面mpf_activity_detector_process函数,主要是通过计算frame的平均值,来完成状态切换的逻辑:

处理过程共有四个状态:

      ACTIVITY状态

      INACTIVITY状态

     TRANS_ACTIVITY状态

     TRANS_INACTIVITY状态

     其中TRANS状态是中间状态,再切换为ACTIVITY状态和INACTIVITY状态的时,需要经过这个状态来累计设定时长,如果满足了,才会切换,否则不予切换。

/** Process current frame */
MPF_DECLARE(mpf_detector_event_e) mpf_activity_detector_process(mpf_activity_detector_t *detector, const mpf_frame_t *frame)

    mpf_detector_event_e det_event = MPF_DETECTOR_EVENT_NONE;
    apr_size_t level = 0;
    if((frame->type & MEDIA_FRAME_TYPE_AUDIO) == MEDIA_FRAME_TYPE_AUDIO) 
        /* first, calculate current activity level of processed frame */
//此处计算得到level的值
level = mpf_activity_detector_level_calculate(frame); #if 0 apt_log(APT_LOG_MARK,APT_PRIO_INFO,"Activity Detector --------------------- [%"APR_SIZE_T_FMT"]",level); #endif /*如果当前状态处于INACTIVITY状态,并且level 大于我们设定的阀值,开始向活动状态切换,但是并没有变成活动状态*/ if(detector->state == DETECTOR_STATE_INACTIVITY) if(level >= detector->level_threshold) /* start to detect activity */ mpf_activity_detector_state_change(detector,DETECTOR_STATE_ACTIVITY_TRANSITION); else detector->duration += CODEC_FRAME_TIME_BASE; if(detector->duration >= detector->noinput_timeout) /* detected noinput */ det_event = MPF_DETECTOR_EVENT_NOINPUT; else if(detector->state == DETECTOR_STATE_ACTIVITY_TRANSITION)
//处于向活动状态转换的过程。
if(level >= detector->level_threshold)
//如果level 大于阀值 detector
->duration += CODEC_FRAME_TIME_BASE;
//并且超过了设定了向活动状态转换的超时时长
if(detector->duration >= detector->speech_timeout) /* finally detected activity */
//切换为活动状态
det_event = MPF_DETECTOR_EVENT_ACTIVITY; mpf_activity_detector_state_change(detector,DETECTOR_STATE_ACTIVITY); else /* fallback to inactivity */
//降级为非活动状态
mpf_activity_detector_state_change(detector,DETECTOR_STATE_INACTIVITY); else if(detector->state == DETECTOR_STATE_ACTIVITY)
//处于活动状态
if(level >= detector->level_threshold)
//如果level大于阀值,增加duration detector
->duration += CODEC_FRAME_TIME_BASE; else /* start to detect inactivity */
//准备转换成inactivity状态
mpf_activity_detector_state_change(detector,DETECTOR_STATE_INACTIVITY_TRANSITION); else if(detector->state == DETECTOR_STATE_INACTIVITY_TRANSITION)
//处于inactivity transtion状态
if(level >= detector->level_threshold) /* fallback to activity */
//如果大于阀值了,则回归到activity状态
mpf_activity_detector_state_change(detector,DETECTOR_STATE_ACTIVITY); else
//如果检测仍然小于阀值,增加判断时长,如果大于设定的时长了,则进入inactivity状态。 detector
->duration += CODEC_FRAME_TIME_BASE; if(detector->duration >= detector->silence_timeout) /* detected inactivity */ det_event = MPF_DETECTOR_EVENT_INACTIVITY; mpf_activity_detector_state_change(detector,DETECTOR_STATE_INACTIVITY); return det_event;

 

 

 

 

 

 

 

以上是关于voice activity detection,VAD是啥意思的主要内容,如果未能解决你的问题,请参考以下文章

[React] Detect user activity with a custom useIdle React Hook

使用Matlab训练神经网络时出现“此类型的变量不支持使用点进行索引”,是为什么?求大神指教

CSS问题:voice

Google voice的用法

我获得了google voice的邀请码,怎么用啊

voice号用啥软件来打电话