Abstract
Transcutaneous Artificial Larynx (TAL) has been available for over the last 40 years for patients who have had laryngectomies. However their use has several problems since the voice produced by these transducers is of low quality, unnatural and degraded by a steady background noise due to the leakage of acoustic energy from the TAL, its interface with the neck and the surrounding neck tissue, which gives a poorly intelligible and annoying result. In order to significantly enhance the speech as well as to improve the intelligibility of the TAL generated speech, a laryngectomyzed speech enhancement system, is developed in which the voiced parts are determined, by using a pattern recognition stage, and replaced by equivalent voiced parts of normal speech. The evaluation results show that a significant improvement of the artificial larynx generated speech quality and intelligibility is achieved.
Original language | English |
---|---|
Pages (from-to) | III37-III40 |
Journal | Midwest Symposium on Circuits and Systems |
Volume | 3 |
State | Published - 2004 |
Event | The 2004 47th Midwest Symposium on Circuits and Systems - Conference Proceedings - Hiroshima, Japan Duration: 25 Jul 2004 → 28 Jul 2004 |
Keywords
- Alaryngeal speech
- DSP implementation
- Speech enhancement