vector<SpeechSegment> CVoiceDetection::FindSpeechSegment( const float* buffer, int sampleRate )
{
assert( buffer != NULL);
assert( sampleRate != 0 );
//我们通过amdfSize大小的区域来检查人声频率
//我们确保amdfSize内至少含有人声两个周期
int amdfSize = sampleRate / MIN_VOICE_FREQUENCY * 2;
for( auto it = m_startEndMap.begin(); it != m_startEndMap.end(); ++it )
{
//获取人声区域的中间点，确保这个点是人声
int middleFrameIndex = (it->first + it->second)/2;
int middleSampleIndex = middleFrameIndex * m_hop;
vector<float> dataForAMDF( buffer+middleSampleIndex, buffer+middleSampleIndex+amdfSize );
//计算平均幅度差
auto amdfResult = AMDFCalc( dataForAMDF );
//计算人声频率
auto voiceFrequence = VoiceFrequenceCalc( amdfResult, sampleRate );
if( 0 != voiceFrequence )
{
float beat = float((it->second - it->first)*m_hop)/sampleRate;
SpeechSegment smg(voiceFrequence, it->first, it->second, beat);
m_speechSegment.push_back( smg );
}
}
return m_speechSegment;
}
void see(t_env *e, int fd, char **arg)
{
int levelplayer;
t_players *player;
(void) arg;
if ((player = get_player_from_id(e->players, fd)) == NULL)
smg(fd, "Player not found ? wtf");
else
{
levelplayer = player->level;
make_see(levelplayer, player, fd, e);
}
}