| Paper: | SP-L5.6 | ||
| Session: | Pitch and Tone Based Speech Analysis | ||
| Time: | Thursday, May 20, 11:10 - 11:30 | ||
| Presentation: | Lecture | ||
| Topic: | Speech Processing: Speech Analysis | ||
| Title: | A NOVEL METHOD FOR COMPUTATION OF PERIODICITY, APERIODICITY AND PITCH OF SPEECH SIGNALS | ||
| Authors: | Om Deshmukh; University of Maryland, College Park | ||
| Jawahar Singh; University of Maryland, College Park | |||
| Carol Espy-Wilson; University of Maryland, College Park | |||
| Abstract: | This paper presents improvements to our previously proposed algorithm tocompute the proportion of periodic and aperiodic energies in speechsignals and to estimate the pitch period. Although previously the periodicand aperiodic energies were estimated independent of each other at eachframe, a binary decision was made at each of the non-silent channels. Inthis paper, we present an extension that replaces the binary decision witha measure of degree of periodicity and aperiodicity in each channel.Evaluation on synthetic speech-like data shows a better agreement in theestimated SNR and the actual SNR using with this improvement. Moreover, inthe task of estimating the SNRs, this method significantly outperforms amethod based on cepstral coefficients. When evaluated on a speechdatabase, the periodicity and aperiodicity accuracy increasedsignificantly. The previous pitch detector was prone to commiting pitchdoubling and pitch halving errors and was unable to reliably detect pitchin weakly periodic regions. Significant changes have reduced the errorrate by 28.7%. The pitch detector is also able to accurately detect thepitch of the synthetic speech-like signals and to capture the jitterpresent in the signals. | ||
| Back | |||
Home -||-
Organizing Committee -||-
Technical Committee -||-
Technical Program -||-
Plenaries
Paper Submission -||-
Special Sessions -||-
ITT -||-
Paper Review -||-
Exhibits -||-
Tutorials
Information -||-
Registration -||-
Travel Insurance -||-
Housing -||-
Workshops