[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference trlian::voice_products_v2

Title:	Voice Products V2
Notice:	Announcements, General Info, Overviews: Notes 1 - 15
Moderator:	CSSRUS::OUYANG

Created:	Tue Aug 14 1990
Last Modified:	Fri Apr 18 1997
Last Successful Update:	Wed May 07 1997
Number of topics:	607
Total number of notes:	1972

605.0. "UNIX and speech recog?" by VELENO::MICHIELI (Is IBM DEClining?) Mon Mar 24 1997 12:47

    Hi all,
    hope this is the right conference; if not plese give me the right
    pointer.
    
    My customer works in the Air Traffic Control systems. They works under
    Digital UNIX. The need is to send/receive messages from tower to
    aircraft and viceversa. The operator say a command which is displayed
    in a window and then sent this to the aircraft. The same from aircraft,
    where both text will be shown and the speech recognized.
    
    Q: is such already available under UNIX with MME or DECtalk, with some 
    programming extensions, or do we need special HW/SW?
    
    Regards,
     Gabriele.

T.R Title User Personal
Name Date Lines

605.1 I'd concentrate on non-verbal I/O XDELTA::HOFFMAN Steve, OpenVMS Engineering Tue Mar 25 1997 13:24 20

T.R	Title	User	Personal Name	Date	Lines
605.1	I'd concentrate on non-verbal I/O	XDELTA::HOFFMAN	Steve, OpenVMS Engineering	`Tue Mar 25 1997 13:24`	20
	We have not yet reached the "science fiction" voice recognition system. I cannot imagine including a speech recognizer into this particular environment, given the accents of the speakers, the distortion likely on the various radio links, and the life-critical safety requirements. DECtalk is a voice synthesizer. It converts text into sound. It's a "voice printer". DECvoice is/was one commercially-available package that performed synthesis, recognition, and digitization. (text-to-speech, speech-to-text, record-and-playback, respectively.) If the customer insists on this, they will need one of the hardware or software speaker independent voice recognition packages, and they will need to create and stick with a particular set of words -- a set vocabulary. (Commercial packages such as DragonDictate still have interesting problems with accents, and with various homonyms, etc.) Speaking as someone who occasionally flies, I'd concentrate on other solutions.

   We have not yet reached the "science fiction" voice recognition system.

   I cannot imagine including a speech recognizer into this particular
   environment, given the accents of the speakers, the distortion likely
   on the various radio links, and the life-critical safety requirements.

   DECtalk is a voice synthesizer.  It converts text into sound.  It's
   a "voice printer".  DECvoice is/was one commercially-available
   package that performed synthesis, recognition, and digitization.
   (text-to-speech, speech-to-text, record-and-playback, respectively.)

   If the customer insists on this, they will need one of the hardware
   or software speaker independent voice recognition packages, and they
   will need to create and stick with a particular set of words -- a set
   vocabulary.  (Commercial packages such as DragonDictate still have
   interesting problems with accents, and with various homonyms, etc.)

   Speaking as someone who occasionally flies, I'd concentrate on other
   solutions.