10.07.2015 Views

第22回 ロボット聴覚特集 - 奥乃研究室 - 京都大学

第22回 ロボット聴覚特集 - 奥乃研究室 - 京都大学

第22回 ロボット聴覚特集 - 奥乃研究室 - 京都大学

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

semitone 3.2.1 Vocal Fry (creaky) 2 F0 Vocal Fry (creaky) [7][8]Vocal Fry F0 “very short-term”5 msF0avg2a2.5 ms F0tgt2bF0move F0tgt2b –F0avg2a F0 F0 Vocal[7]Fry 4 3 (PPw: Peak Power)IFP:Intra-Frame Periodicity20 dB IPS:Inter-Pulse Similarityduration[8]3 F0move vs. duration PPw > 7 dB, IFP < 0.8, IPS > 0.6 SpeechPPw IFP IPSVery short-termpowerPickpeaksCheckperiodicityChecksimilarityBand-pass filterShort-term periodicity:Merge the inter-100 ~ 1500 Hzsub-harmonic constraintpulse intervalsVocal frysignal Frame-synchronized Glottal pulse-synchronizedsegmentsFig. 4 Simplified block diagram of the vocal fry detection.3.2.2 (aperiodicity; double- periodicity)Fig. 3 Distributions of the prosodic parameters for each Vocal fry Harsh perceived speech act group.[9] Creaky (Vocal Fry) Harsh 3.2.1 Vocal Fry Harsh WindowedLPC-basedNormalizedPeakNACRspeech signal inverse filteringautocorrelationrelationsTLRFig. 5 Simplified block diagram of the parameters foraperiodicity/double-periodicity detection.3.2 5 3.2.1 3.2.3 NACR(Normalized Auto-Correlation Ratio)2 3.2.4 74

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!