Analysis/Synthesis Speech Model Based on the Pitch-Tracking Periodic-Aperiodic Decomposition
This paper presents a speech analysis/synthesis model based on periodic-aperiodic decomposition. In presented approach, decomposition is performed in whole speech band without making identification of voiced/unvoiced regions. Other important feature is pitch-tracking ability of decomposition algorithm. For this purpose a new pitch-tracking transformation called Time-Varying Discrete Fourier Transform (TVDFT) is employed. Periodic component is modelled as a sum of pitch harmonics with amplitudes and phases estimated with TVDFT. Aperiodic component is defined as a difference between original speech signal and synthesised periodic component. TVDFT needs accurate fundamental pitch estimation. This paper also presents a robust pitch estimation.. Experimental results showing advantages of suggested model are also given.
Keywordsspeech decomposition Time-Varying DFT
Unable to display preview. Download preview PDF.
- Kondoz A.M., “Digital speech: coding for low bit rate communication systems”, John Wiley & Sons, Inc., New York, 1996.Google Scholar
- Almeida L.B., Tribolet J.M., “Harmonic Coding: A Low Bit-Rate, Good Quality, Speech Coding Technique”, Proc. IEEE Int. Conf. on Accoust., Speech and Signal Processing, pp. 1664–1667, 1982.Google Scholar
- McAulay R.J., Quatieri T.F., „Sinusoidal Coding“ in “Speech Coding and Synthesis” (W. Klein and K. Palival, eds.), Elsevier Science Publishers, Amsterdam, 1995.Google Scholar
- Sercov V., Petrovsky A., „An Improved Speech Model with Allowance for Time-Varying Pitch Harmonic Amplitudes and Frequencies in Low Bit-Rate MBE Coders”, Proc. of the 6ht European Conf. on Speech Communication and Technology EUROSPEECH'99, pp. 1479–1482 Budapest, Hungary, 1999.Google Scholar
- Petrovsky A., Sercov V., “Low Bit-Rate AbS Spectral Coding Based on the Harmonic Analysis of Speech Agreed Upon with Time-Varying Pitch Frequency and Psychoacoustical Optimization”, Proc. of Nordic Signal Processing Symposium NORSIG2000, pp. 45–48, 2000.Google Scholar
- Petrovsky A., Zubrycki P., Sawicki A., Tonal and noise components separation based on a pitch synchronous DFT analyzer as a speech coding method // Proc. of European Conference on Circuit Theory and Devices ECCTD2003, Vol. III, pp. 169–172, 2003.Google Scholar
- Eric W. M. Yu, Cheung-Fat Chan, A harmonic+noise coder with improved transient speech performance // Proc. of European Signal Processing Conference EUSIPCO'99, Special Session “Speech Coding”, 1999.Google Scholar