A novel recurrent network based pitch detection technique for quasi-periodic/pitch-varying signals

The accuracy of pitch detection algorithms affects the performance of many speech and audio applications such as speech compression, computer music analysis/synthesis and information retrieval of audio signals. In many applications, it is also desired that the algorithms should be robust to background noise. A recurrent network based method is proposed in this paper. Though the proposed method requires more computation compared to some existing methods, it is more accurate and less sensitive to noise. The other advantage is that it requires a smaller time frame to estimate the pitch compared to other methods. Therefore, it is more suitable for tracking the pitch of a pitch-varying signal or a quasi-periodic signal. Both the synthesized and natural tones are used in the computer simulation.

[1]  John S. Collura,et al.  MELP: the new Federal Standard at 2400 bps , 1997, 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[2]  Jing Peng,et al.  An Efficient Gradient-Based Algorithm for On-Line Training of Recurrent Network Trajectories , 1990, Neural Computation.

[3]  Tamás D. Gedeon,et al.  Simulated annealing and weight decay in adaptive learning: the SARPROP algorithm , 1998, IEEE Trans. Neural Networks.

[4]  Wolfgang Hess,et al.  Pitch Determination of Speech Signals: Algorithms and Devices , 1983 .

[5]  Fernando J. Pineda,et al.  Recurrent Backpropagation and the Dynamical Approach to Adaptive Neural Computation , 1989, Neural Computation.

[6]  Andrzej Czyzewski,et al.  Representing Musical Instrument Sounds for Their Automatic Classification , 2001 .

[7]  S. Haykin,et al.  Adaptive Filter Theory , 1986 .

[8]  Alan V. Oppenheim,et al.  Discrete-Time Signal Pro-cessing , 1989 .

[9]  John E. Markel,et al.  Linear Prediction of Speech , 1976, Communication and Cybernetics.

[10]  Matti Karjalainen,et al.  Towards High-Quality Sound Synthesis of the Guitar and String Instruments , 1993, ICMC.

[11]  Wolfgang Hess,et al.  Pitch Determination of Speech Signals , 1983 .

[12]  Sheng-Fu Liang,et al.  A new automatic IIR analysis/synthesis technique for plucked-string instruments , 2001, IEEE Trans. Speech Audio Process..

[13]  Manfred R. Schroeder,et al.  Code-excited linear prediction(CELP): High-quality speech at very low bit rates , 1985, ICASSP '85. IEEE International Conference on Acoustics, Speech, and Signal Processing.