(Message inbox:22)
Received: from cs.orst.edu by ICS.UCI.EDU id aa27726; 27 Jun 90 9:39 PDT
Received: from turing.CS.ORST.EDU by CS.ORST.EDU (5.59/1.15)
	id AA04900; Wed, 27 Jun 90 09:40:03 PDT
Received: from tesla.CS.ORST.EDU by turing.CS.ORST.EDU (4.1/SMI-4.0)
	id AA00338; Wed, 27 Jun 90 09:39:22 PDT
Date: Wed, 27 Jun 90 09:39:22 PDT
From: Tom Dietterich <tgd@turing.cs.orst.edu>
Message-Id: <9006271639.AA00338@turing.CS.ORST.EDU>
To: aha@ICS.UCI.EDU
Subject: [tgd@turing.CS.ORST.EDU: Errors in the nettalk database]
Reply-To: tgd@turing.cs.orst.edu

David,

There are a few errors in the cmu nettalk file:

--Tom
======

Date: Tue, 26 Jun 90 12:48:06 PDT
From: Tom Dietterich <tgd@turing.CS.ORST.EDU>
To: Scott.Fahlman@cs.cmu.edu
Subject: Errors in the nettalk database
Reply-To: tgd@turing.CS.ORST.EDU

I recently compared the nettalk.data file from the nnbench directory
at CMU with the version of the data file that we obtained from
Sejnowski (and that we have corrected slightly).  I noticed a few
errors, and I see that you are the maintainer of the file:

(These are extracted from a diff.  Lines beginning with < are the
corrected lines.  Lines beginning with > are from nettalk.data at
cmu.) 

< barroom	barru-m	>1<>2<<	0
---
> barroom	barru-m	>1<>2<<	

The foreign/domestic information is missing.

< fiddle	fId-L-	>1<>>>	0
---
> fiddle	fId-L-	>1<>>)	0

Typo in the stress field.

< urbanization	-RbxnxzeS-xn	2<>0>1>0>>0<	0
---
> urbanization	-RbxnxzeS-xn	2<0>1>0<<	0

The stress information is incomplete--it doesn't line up with the
phonemes. 

--Tom

