Kalman Filtering for Low Distortion Speech Enhancement in Mobile Communication
ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97)-Volume 2 - Volume 2
Hi-index | 0.00 |
Oesophageal voice is characterized by its extremely low intelligibility. An algorithm based on Kalman Expectation Maximization (EM) has been developed. The noise presented in the model, state and measurement noise has been optimized in order to improve the algorithm results. The database consists of "a" phonemes of several patients having undergone a total laryngectomy. Additionally, the effect of the algorithm on the UMTS mobile communication context has been tested. The tests show that the algorithm gives the best results when it is used as state noise an oesophageal noise and brown noise as measurement noise. The global percentage enhancement is 75.78%.