Spotting Chief Speaker from Press Conference Recordings Based on Silence Detection
This paper presents a method for spotting chief speaker from press conference recordings where durations of silence segment in one utterance of chief speaker and other speakers are obviously different.In the proposed method, speech endpoint detection is first performed on the audio conference recordings for obtaining the durations of silence segment (i.e.Si sequence).Then, Si sequence is converted into 1-0 sequence where outliers are revised Finally, speech segments limited by the continuous 1 sequences are extracted as the chief speakers voices.The experiments are conducted on two data sets with different durations of silence segment in chief speakers utterances for comparing the proposed method with the conventional approach (based on speaker segmentation using B1C and spectrum clustering).The experimental results show that the proposed method achieves higher F measures (harmonic mean of precision rate and recall rate) with faster speed in comparison with the conventional approach for spotting chief speaker from press conference recordings.
press conferences recordings chief speakers speech endpoint detection speaker segmentation speaker clustering
Wu Wei Li Yanxiong Wang Zili Chen Zhuyun
School of Electronic and Information Engineering,South China University of Technology,Guangzhou,China
国际会议
哈尔滨
英文
147-150
2013-08-16（万方平台首次上网日期，不代表论文的发表时间）