Subject: Re: Hypernasality From: Chandan Raghava Narayan <cnarayan@xxxxxxxx> Date: Tue, 26 Sep 2006 11:36:55 -0400I would think speech samples from the northern midwest (i.e., southeast Michigan, Wisconsin, etc.) might be helpful as many speakers from these areas are "always nasal." You might try looking at the Buckeye corpus (buckeyecorpus.osu.edu) from OSU. Chandan Tarun Pruthi wrote: Hi all, I have been working on the automatic detection of nasalization using knowledge based acoustic parameters for sometime. Since, one of the possible applications of such a system can be to detect hypernasality in a person's speech by just looking at the acoustics, I would like to test my algorithm on hypernasal speech and find out if it is able to separate hypernasal speakers from speakers with normal speech. In this respect, I wanted to know if anyone knows of a standard database of speech recorded from normal and hypernasal speakers which I can use for research purposes? It would be fabulous if this database is transcribed and/or has nasalance scores for each of the speakers. Thanks in advance Tarun ----------------------------------------- Tarun Pruthi Graduate Research Assistant, ECE Room 3180, A V Williams Building University of Maryland, College Park MD 20742 USA Email: tpruthi@xxxxxxxx Web: www.ece.umd.edu/~tpruthi Ph: 301-405-1365 ---------------------------------------- ================================ chandan r. narayan dept. of linguistics university of michigan -------------------------------- cnarayan@xxxxxxxx www-personal.umich.edu/~cnarayan ================================