Received-SPF: none (google.com: me@v4hn.de does not designate permitted sender hosts) client-ip=85.13.134.113;
Date: Mon, 15 Dec 2014 18:13:01 +0100
From: v4hn <me@v4hn.de>
To: lojban@googlegroups.com
Subject: Re: [lojban] Speech Recognition for Lojban
Message-ID: <20141215171301.GB753@kuebelreiter.informatik.Uni-Osnabrueck.DE>
References: <940da156-9cea-4bba-8745-a61b119b84e2@googlegroups.com>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature"; boundary="jq0ap7NbKX2Kqbes"
Content-Disposition: inline
In-Reply-To: <940da156-9cea-4bba-8745-a61b119b84e2@googlegroups.com>
User-Agent: Mutt/1.5.23 (2014-03-12)
Reply-To: lojban@googlegroups.com
Precedence: list
Mailing-list: list lojban@googlegroups.com; contact lojban+owners@googlegroups.com
Sender: lojban@googlegroups.com
X-Spam_score: -1.9
X-Spam_score_int: -18
X-Spam_bar: -

--jq0ap7NbKX2Kqbes
Content-Type: text/plain; charset=UTF-8
Content-Disposition: inline

coi do

> *But first we need a speech recognition tool for Lojban.* Any ideas? Has
> anyone in Lojban community done this before? And any resources related to
> this?

This idea pops up every now and then, yes.
However, I don't know of any really successful approach.

In 2008 there has been a project by a couple of cognitive science students
to implement speech recognition for lojban at the university of Osnabrueck.
After (quite) some nagging they provided me with a folder of (really few)
training data and some scripts they used back then.

It doesn't look like they had a lot of fun with it though:
> We decided to try the impossible:
> Use the speech recognition toolkit of CMU to build a speech recognizer for a non-English language.

Also, it seems a lot has changed w.r.t. open source
speech recognition engines in the last couple of years.
ti'e CMU Sphinx got rewritten and other (potentially capable)
speech recognition engines like julius appeared.

I didn't get around to play with any of these, although I'm still
interested in a working/usable FOSS speech recognition tool for Lojban.

Btw: In theory it is, by design, much more easy to do speech recognition with Lojban
than with most natural human languages because of the audio-visual isomorphism.
It prevents hard questions like "Why is there a 'k' in knight?" .u'i


mi'e la .van. mu'o

-- 
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+unsubscribe@googlegroups.com.
To post to this group, send email to lojban@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban.
For more options, visit https://groups.google.com/d/optout.

--jq0ap7NbKX2Kqbes
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=Xrea
-----END PGP SIGNATURE-----

--jq0ap7NbKX2Kqbes--