To demonstrate speech recognition for over a hundred people is an extreme sport! Trond took this on with great success at a seminar arranged by SIKTE (Centre for IKT assistive devices) on September 14th.
Story by: Morten Tollefsen - 22.09.2010
Daniel Scheidegger (SIKTE) began with a presentation of the SMUDI-prosjektet! Scheidegger has been a very active project participant, and his experience with speech recognition and people with disabilities has strongly influenced the development of VOMOTE. The SMUDI project has received support from the VERDIKT program run by the Norwegian Research Council. That we now have a product for Norwegian voice control of the PC is a direct result of this program!
VOMOTE been thoroughly tested by users, one of whom should have demonstrated Norwegian speech recognition. But autumn is here, and this is the time when some of us become sick! Trond Ausland stepped in and took this challenge in his stride.
Almost unbelievable! Recognition during the 10-15 minute long demonstration was 100%. The applause was spontaneous when Ausland finished his demonstration.
Trond answered several questions from the audience.
What about variable speech quality?
VOMOTE is speaker independent. This makes the system very well suited to cope with variable voice quality. People who use VOMOTE must have relatively good pronunciation.
Does VOMOTE support all dialects?
The safest choice is the dialect in Eastern Noray, but many dialects work well with the system. However, we cannot guarantee that all dialects will have equal recognition.
When is the product being launched?
October 15th 2010.
Will the next stage be recognition of sounds?
Sound recognition is quite different from speech recognition. MediaLT is leading a pilot project looking at sound control.
How can one see what the system recognizes?
This appears in a small window and/or with help of synthetic speech.
How do you write text, for example in Word?
With VOMOTE you can say letters, numbers and other characters.
Tips noen om siden