|
Echo Cancellation and Multi-Microphone Noise Cancellation
The echo cancellation algorithm developed for host telephony applications, holds two patents filed during 1998. The algorithm was implemented as a standard DLL, operated by any host application under Unix and Windows OS. For more details, see the echo cancellation document.
The group developed algorithms for beam-forming and adaptive noise cancellation using recordings from microphone arrays. For details, see the multi-microphone document.
Speech Anaylsis, Synthesis, Modification, and Enhancement
The group has developed algorithms for time domain and frequency domain speech synthesis. Natural sound, high quality synthesis is achievable using a database of speaker wave forms. Speech analysis methods include accurate, high resolution pitch detection. Speech modification activities include pitch modification and Time Scale Modification (TSM - for speeding up and slowing down audio without pitch change). Speech enhancement includes improving the intelligibility of speech recorded in reverberant conditions.
| 
