From rob@twcny.rr.com Fri Jan 04 13:31:21 2002 Return-Path: X-Sender: rob@twcny.rr.com X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-8_0_1_3); 4 Jan 2002 21:31:19 -0000 Received: (qmail 87907 invoked from network); 4 Jan 2002 21:31:19 -0000 Received: from unknown (216.115.97.167) by m10.grp.snv.yahoo.com with QMQP; 4 Jan 2002 21:31:19 -0000 Received: from unknown (HELO mailout5.nyroc.rr.com) (24.92.226.122) by mta1.grp.snv.yahoo.com with SMTP; 4 Jan 2002 21:31:20 -0000 Received: from mail1.twcny.rr.com (mail1-1 [24.92.226.139]) by mailout5.nyroc.rr.com (8.11.6/Road Runner 1.12) with ESMTP id g04LVIq12177 for ; Fri, 4 Jan 2002 16:31:18 -0500 (EST) Received: from riff ([24.92.246.4]) by mail1.twcny.rr.com (Post.Office MTA v3.5.3 release 223 ID# 0-59787U250000L250000S0V35) with ESMTP id com for ; Fri, 4 Jan 2002 16:31:18 -0500 Received: from rob by riff with local (Exim 3.33 #1 (Debian)) id 16Mbvq-0000KT-00 for ; Fri, 04 Jan 2002 16:31:18 -0500 Date: Fri, 4 Jan 2002 16:31:18 -0500 To: lojban@yahoogroups.com Subject: Re: [lojban] Lojban Text to Speech Message-ID: <20020104213118.GC1109@twcny.rr.com> References: <20020103205206.B23288@cc96364-a.hwrd1.md.home.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.24i X-Is-It-Not-Nifty: www.sluggy.com From: Rob Speer Reply-To: rob@twcny.rr.com X-Yahoo-Group-Post: member; u=2572649 X-Yahoo-Profile: squeekybobo On Thu, Jan 03, 2002 at 09:03:22PM -0500, Invent Yourself wrote: > On Thu, 3 Jan 2002 jspickes@etrademail.com wrote: > > > I believe the dialog8k.wav you've discovered is the file to which I'm > > referring. Yes, I'll agree that it could stand some improvement, but I'm > > impressed with its intelligibility at this early stage. Even with my > > (extremely limited) lojban vocabulary, I was able to pick out parts of what > > it was saying. > > > I can pick out the occasional word but not enough to get any context. Once I got used to the 'accent' (not the Southern accent, but the fact that, for example, a "whirring" sound represented {x}) by reading along with the voice, I could understand about half of the words without reading along. Then again, I had read the English version of this conversation before, so that provided the context. I tried not reading it at all at first, and didn't understand a bit. The main problem was "le skami" and "le prenu"; not only were they not grammatically attached, only set apart by pauses ({.le prenu.} signified {.i sei le prenu cusku}), but the pronounciation of "prenu" was totally unrecognizable. Still, I was generally impressed with the Lojban speech because it's quite an improvement over nothing, which is what we had before. (Yes, there is the anthem, but that was made from piecing together sounds from Spanish, Italian, and other languages' synthesizers.) festvox is the basis for the 'Festival' synthesizer, right? It's great to see that can be adapted for Lojban. If someone recorded diphones (not a trivial task, I know), I think we could have some relatively clear Lojban speech come out of it. -- la rab.spir noi tirna le voksa be lo skami