30-05-2013, 04:05 PM
Characteristics of Speech Recognition Applications &Methods of Handling Recognition Errors
Recognition Applications.ppt (Size: 100 KB / Downloads: 19)
Real
Increase Productivity, easy to use, better machine human interface, mode of communication.
User Friendly
Make user comfortable with Voice dialogue.
Effective means of communication.
Accurate
Achieve desired performance on task associated with recognition decision.
Responsive
Response to query comes immediately.
Application Scope
Interactive voice response system (IVRS).
PIN and numeric password entry modules.
Automated teller machines (ATMs).
Mobile operators.
Home and office security device providers.
Function Description
The project converts speech into text.
It contains-
Speech acquisition.
Speech preprocessing.
HMM training.
HMM-based recognition.
Text storage.
Methods of Handling Recognition Errors
Fail Soft Methods
The “Cost” of recognition error in terms of user and revenue is low.
The Error is detected and corrected at later stage.
User enter a correction mode.
Rejection/Pass on to Operator
Recognition system fails to defer a decision on inputs.
It passes string to the human operator who makes final decision by listening to spoken input.