Recognition (1) Voice Input Analog to Digital Acoustic Model Language Model Display Speech Engine Feedback
Spoken words: “I think there are…..”
Phonemes: ‘ ay th-in-nk-kd dh-eh-r aa-r’
H.M.M.’s: 5 state representation
Voice Training (1)
Can be done by:
Predetermined text segments
Compare new acoustic with old and combines
More training = better recognition
Voice Training (2)
User specific Voice file
Patterns of word use
Move cursor by voice command
Memorize edit commands
List of possible alternatives
Make correction manually
Program or Link name/number
Recent Improvements in SR
Faster training ~10 min.
Better recognition ~95%
More compatible software
Better system control/command
Current Software Options for PC
Dragon Systems – Naturally Speaking
Philips – FreeSpeech
IBM – ViaVoice
Lernout & Hauspie – Voice Xpress
How well do the work? Training Dictation Correct. App. Integrat. Command - Control Dragon Excellent Excellent Good Good Philips Fair Fair Good Good IBM Excellent Good Good Excellent L & H Good Good Good Good
Future of SR
SUI – Speech-based User Interface
- Greater accuracy
- Greater system control/command
- More compatible software
How does it work?
Problems of SR
More SR coming soon….
1. Alwang, Greg. “Speech Recognition,” PC Magazine, December 1 1999
2. Hauptmann, Alexander G. Jang, Photina Jaeyun. Carnegie Mellon University. “Learning to Recognize Speech by Watching Television,” IEEE Intelligent Systems, September/October 1999.
3. Miastkowski, Stan. “Latest Speech Software Gets You Up and Running Faster,” PC World, November 1999.