Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!aixssc.uk.ibm.com!watnews.watson.ibm.com!newsgate.watson.ibm.com!news.ans.net!howland.reston.ans.net!math.ohio-state.edu!jussieu.fr!univ-lyon1.fr!swidir.switch.ch!news.unige.ch!dolent!karit
From: karit@idiap.ch (Kari Torkkola)
Subject: Re: Bandwidth, Telephones, and Speech Recognition
Message-ID: <1994Mar11.084108.20474@news.unige.ch>
Sender: usenet@news.unige.ch
Reply-To: karit@idiap.ch
Organization: IDIAP, Switzerland
References: <1994Mar11.040356.8720@Princeton.EDU>
Date: Fri, 11 Mar 1994 08:41:08 GMT
Lines: 19

In article 8720@Princeton.EDU, Devin Hosea <0428870@phoenix.princeton.edu> writes:
> Does anyone know the bandwidth of audio transmission over typical phone
> lines?  I heard that it can be as little as 3,000hz.  If so, how does this
> affect the prospects for good speech recognition over the phone?

I have a copy of a two-page paper "The effects of the telephone network
on phoneme classification" by Benjamin Chigier from NYNEX. Unfortunately
I cannot recall where I copied it from. It has "IEEE Speech Recognition
Workshop" printed as a header, and judged from the references, it has
appeared after November 1990.

Anyway, the paper compares phoneme classification using TIMIT and N-TIMIT.
Using high quality speech the accuracy was 68.2% and with telephone speech
it was 59.9%. There were significant differences in 17 (of 39) phoneme
classes, not only in fricatives as could be expected, but also in some
vowels and liquids.

- Kari

