Neural Network Based Pitch Tracking in Very Noisy Speech
Authors:
- 14 October 2014
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 22 (12) , 2158-2168
Abstract
Pitch determination is a fundamental problem in speech processing, which has been studied for decades. However, it is challenging to determinate pitch in strong noise because the harmonic structure is corrupted. In this paper, we estimate pitch using supervised learning, where the probabilistic pitch states are directly learned from noisy speech data. We investigate two alternative neural networks modeling pitch state distribution given observations. The first one is a feedforward deep neural network (DNN), which is trained on static frame-level acoustic features. The second one is a recurrent deep neural network (RNN) which is trained on sequential frame-level features and capable of learning temporal dynamics. Both DNNs and RNNs produce accurate probabilistic outputs of pitch states, which are then connected into pitch contours by Viterbi decoding. Our systematic evaluation shows that the proposed pitch tracking algorithms are robust to different noise conditions and can even be applied to reverberant speech. The proposed approach also significantly outperforms other state-of-the-art pitch tracking algorithms.Keywords
This publication has 35 references indexed in Scilit:
- A Fast Learning Algorithm for Deep Belief NetsNeural Computation, 2006
- A multipitch tracking algorithm for noisy speechIEEE Transactions on Speech and Audio Processing, 2003
- The fundamental and medical impacts of recent progress in research on hereditary hearing lossHuman Molecular Genetics, 1998
- Long Short-Term MemoryNeural Computation, 1997
- An Efficient Gradient-Based Algorithm for On-Line Training of Recurrent Network TrajectoriesNeural Computation, 1990
- Finding structure in timeCognitive Science, 1990
- Learning representations by back-propagating errorsNature, 1986
- A semiautomatic pitch detector (SAPD)IEEE Transactions on Acoustics, Speech, and Signal Processing, 1975
- Automatic Speaker Recognition Based on Pitch ContoursThe Journal of the Acoustical Society of America, 1972
- Period Histogram and Product Spectrum: New Methods for Fundamental-Frequency MeasurementThe Journal of the Acoustical Society of America, 1968