LINGUIST List 8.715

Tue May 13 1997

Qs: Syllabification, Speech Recognition

Editor for this issue: Ljuba Veselinova <ljubalinguistlist.org>


We'd like to remind readers that the responses to queries are usually best posted to the individual asking the question. That individual is then strongly encouraged to post a summary to the list. This policy was instituted to help control the huge volume of mail on LINGUIST; so we would appreciate your cooperating with it whenever it seems appropriate.

Directory

  1. Xu Peng, Syllabification tools
  2. Rusty_Richards, Automatic Speech Recognition

Message 1: Syllabification tools

Date: Fri, 09 May 1997 14:34:21 +0800
From: Xu Peng <xupengsentosa.sas.ntu.ac.sg>
Subject: Syllabification tools

Does anybody know if there is any tool can syllabize people's
English names, e.g., William will be syllabized into wi-llia-m,
and David will be syllabized into da-vi-d. Thanks for your help
in advance. -- Xu Peng

======================================================================
Miss Xu Peng email : xupengsentosa.sas.ntu.ac.sg
Parallel Processing Lab Tel : (65) 7996151
SAS Fax : (65) 7926559
Nanyang Technological Univ. Home : Hall 11 Blk 53-4-1047
Singapore 639798 http://sentosa.sas.ntu.ac.sg:8000/~xupeng/
======================================================================
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Automatic Speech Recognition

Date: Fri, 09 May 1997 11:32 -0500 (EST)
From: Rusty_Richards <Rusty_Richardswycliffe.org>
Subject: Automatic Speech Recognition


 The following question comes to me (Rusty Richards) from Doug Peters, 
 PhD, researcher in automatic speech recognition at NORTEL in Montreal, 
 Canada:
 
 ------------------------------------------------------------------ 
 May 7, 1997
 
 Hi Rusty,
 
 My mail system is finally unstuck, and I will once again attempt to 
 briefly describe the question that I had for you on Sunday:
 
 Automatic Speech Recognition systems need help. Sure, the speaker 
 that unconsciously trains herself to reduce speaking variability while 
 addressing a computer can achieve high-ninety word recognition rates 
 with a high quality microphone in a quiet room. If the same speaker 
 is talking to a friend, however, the performance of the recognition 
 system in the same friendly environment will degrade considerably. 
 And the problem is much worse over the phone. We have tried 
 baseform/surfaceform experiments, essentially permitting multiple 
 pronunciations of each word. Unfortunately, this has the effect of 
 making _every_ word more likely, rather than selectively increasing 
 the likelihood of the correct words. Might you suggest any other 
 information that we could bring to the problem to exploit current 
 linguistic knowledge?
 
 Thanks,
 
 Doug
 ------------------------------------------------------------------
 
 Doug Peters can be reached by email at petersnortel.ca. Any help 
 would be appreciated. Please copy me with any reply. 
 
 Thanks,
 
 Rusty Richards (rusty_richardssil.org)
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue