Z3000PA@AWITUW01.BITNET (01/30/90)
Recently, there has been some discussion about the keyboard being
replaced by voice input, and about voice output, too.
1. For which languages will voice systems become generally available
within the next ten years? US-English? British English? Japanese?
English spoken with scottish or german or japanese accent? Gaelic?
French? Italian? German German, Austrian German, Swiss German?
Castillian Spanish, Catalane Spanish, Latin American Spanish? ...
All the European languages?
All African and Asiatic languages?
How many languages are there one earth, even if we consider only
those parts of those countries where PCs are likely to be used?
2. One advantage of written e-mail over spoken voice-mail is that
it's much easier to read and write foreign languages rather than
to listen to them or to pronounce them.
Let's take myself as an example:
I am in regular e-mail connection with various people both within
and outside my home country (Austria). With people in Austria and
Germany, I use German. With people in other countries and with
international newsletters and mailing lists, I mostly use English,
but I am also subscribed to a french and a dutch mailing list, and
I have even happened to recieve portions of e-mail in Italian,
Swedish and Spanish.
With German, voice input and output would be no problem for me
(except with genuine Swiss German).
With spoken English, I would probably understand less than 60 % of the
texts, whereas I currently understand about 90 % of the written
English I recieve; and I doubt that any computer system would be able to
understand my near-English pronounciation and my strong german accent.
(On the other hand, I assume that all of you are able to understand
this written message, in spite of my spelling errors.)
With french, italian or dutch, I manage to understand about 60 % of
what I read, but I would not understand enough if I had to listen to
these languages spoken as fast as is usual for native speakers, and I
am unable to pronounce anything correctly in these languages.
So, what will voice input and output make better for me? Or for
other people who also face the existence of more than one language
on earth?
Hubert Partl, Vienna, Austria <z3000pa@awituw01.bitnet>unccab@calico.med.unc.edu (Charles Balan) (02/22/90)
In article <wZsNchz01EuM1LMncG@twain> urban%rcc@RAND.ORG ("Michael Urban/fdfTo:") writes: >Even if this were the appropriate forum, which is dubious, it seems >unproductive to argue the technical merits of Esperanto versus Papiamento (or >Loglan, or creoles, or Ido, or Interlingua, or Sindarin, or ...) as an I submit that followups be directed to sci.lang in which one will probably be slightly singed for re-hashing this old argument in that forum :-) >I suppose that trying to relate this discussion to future computing is entirely >in vain, but I will observe again that the only language that real people are >going to want to use to talk to their voice-typewriters or voice-mail systems >is their native language; Which was my point entirely, in response to <someone@somewhere> contention that voice input would be too difficult because not everyone spoke the same language that he does. Ergo, I responded that voice interpretation and translation software would bypass this problem. >but the machine will still have to handle >the native tongues or else it will just be a laboratory curiosity. As the current voice-input devices are currently; interesting, but too expensive and non-versatile for common use. Charles Balan UNCCAB@med.unc.edu , UNCCAB@uncmed.uucp , UNCCAB@unc.bitnet %%%%%%%%%%%%% A Witty Saying Proves Nothing - Voltaire %%%%%%%%%%%%
jrk@sys.uea.ac.uk (Richard Kennaway) (02/22/90)
In article <1990Feb17.035719.18228@mthvax.cs.miami.edu> wb8foz@Mthvax.cs.Miami.Edu (David Lesher) writes: >> Article <654@ajpo.sei.cmu.edu> From: wellerd@ajpo.sei.cmu.edu (David Weller) ># Esperanto. Being rule-based, it holds the greatest potential ># for being _the_ medium from which to base all electronic communications. ># Its potential as a spoken language for computers is already being explored ># at the University of Illinois. The completely regular and phonetic Then of course, there's Loglan, which is even more regular and rule-based. Getting people to speak it may be more of a problem. BTW, what is the current state of the art in speech recognition? I get the impression that it's at least five years away, and has been for a long time... -- Richard Kennaway SYS, University of East Anglia, Norwich, U.K. Internet: jrk@sys.uea.ac.uk uucp: ...mcvax!ukc!uea-sys!jrk
hollombe@ttidca.TTI.COM (The Polymath) (02/27/90)
In article <1313@sys.uea.ac.uk> jrk@uea-sys.UUCP (Richard Kennaway) writes: }BTW, what is the current state of the art in speech recognition? I get }the impression that it's at least five years away, and has been for a }long time... Natural language speech recognition was ten years in the future for about thirty years. It's now about five years in the future and I expect it to stay there for at least another ten years. (-: -- The Polymath (aka: Jerry Hollombe, hollombe@ttidca.tti.com) Illegitimis non Citicorp(+)TTI Carborundum 3100 Ocean Park Blvd. (213) 450-9111, x2483 Santa Monica, CA 90405 {csun | philabs | psivax}!ttidca!hollombe
josh@klaatu.rutgers.edu (J Storrs Hall) (02/28/90)
"}BTW, what is the current state of the art in speech recognition? I get "}the impression that it's at least five years away, and has been for a "}long time... "Natural language speech recognition was ten years in the future for about "thirty years. It's now about five years in the future and I expect it to "stay there for at least another ten years. (-: A good overview can be found in Bristow, "Electronic Speech Recognition", McGraw Hill, 1986. Everything below the level of "understanding" seems to be theoretically ready, and hardware systems capable of putting it together are in the lab. The major stumbling block seems to be raw processing power, and of course we all expect that ship to be coming in "any day now". --JoSH