Click here to load reader

음성인식 2013.12.02

  • Upload
    blaze

  • View
    220

  • Download
    0

Embed Size (px)

DESCRIPTION

음성인식 2013.12.02. 김 성 Multimedia DSP LAB. Humor or . 목차. ☞ Endpoint Detection 1. STE 이론 2. ZCR 이론 3. STE-ZCR Endpoint Detection ☞ Pattern Recognition. 이상 음향 감지 / 분류 1 차 분류 : 사람의 음성 / 비음성 2 차 분류 사람의 음성 → ( 공포 , 평상 ) 감성 분류 비음성 신호 → 자동차 충돌소리 , 폭발소리 등. - PowerPoint PPT Presentation

Citation preview

Sound Source Localization

2013.12.02 Multimedia DSP LAB Humor or

Endpoint Detection 1. STE 2. ZCR 3. STE-ZCR Endpoint Detection Pattern Recognition /

1 : / 2 (, ) ,

EDNPOINT DETECTION STE?Short-Time-Energy

(Voiced Speech) (Unvoiced Speech) .

STE

EDNPOINT DETECTION ZCR?Zero-Crossing Rate ZCR ZCR

ZCR

*

25frame25frame Speech interval Endpoint Detection STE-ZCR : 100ms Threshold Maximum magnitude, Minimum magnitude Zero Crossing rate Short time average magnitude filtering average magnitude average magnitude ( ) 25frame zero-crossing rate zero-crossing rate ( )

DAQ.(DTW,HMM.ANN)

ReferencesL. R. RABINER and M. R. SAMBUR, "An Algorithm for Determining the Endpoints of Isolated Utterances", June 10, 1974

G. Saha, Sandipan and Suman Senapati,"A New Silence Removal and Endpoint Detection Algorithm for Speech and Speaker Recognition Applications", Department of Electronics and electrical Communication Engineering Indian Institute of Technology

WWW.NAVER.COM

WWW.GOOGLE.COM

Q&A~