click below
click below
Normal Size Small Size show me how
HIT 220 Chapter 11
Speech Recognition
Question | Answer |
---|---|
Artificial neural networks (ANN) | a computational technique based on artificial intelligence and machine learning in which the structure and operation are inspired by the properties and operation of the human brain |
Back- end speech recognition (BESR) | the "specific use of SRT in an environment where the recognition process occurs after the completion of dictation by sending voice files through a server" |
Bigram | example "word pair" world affairs over 'whirled affairs' because world occurs more frequently than 'whirled', giving it a higher unigram probability and b/c it occurs more in the speech & writing of target users resulting in higher bigram probability |
Continuous speech recognition | a computer technology that automatically translates voice patterns into written language in real time |
Discrete speech | older voice recognition applications require each word to be dictated slowly and distinctly by the user, allows "the software to determine where one word begins and the next stops |
Front-end recognition( FESR) | "the specific use of speech recognition technology in an environment where the recognition process occurs in real time(or near real time) as dictation takes place" |
Hidden Markov model | an algorithm used in voice recognition |
Interactive voice response system | an automated call handler that can be configured to automatically dial a log of callers and deliver appointment reminders, lab results, and other information when a person answers the phone |
Key- word spotting | used in call routing by a client who when prompted by the system will speak a name or key word and the system will route the caller to the appropriate area or department for phone assistance, used in companies with large client bases,directory assistance |
Marco | similar to a template in the SRT, uses "series of keystrokes & or commands that are executed on command, suited to generating "large amounts of text using only a few commands that are easily recognized" |
Medical transcriptionists (MTs) | a medical language specialist who types or word processes information dictated by providers into written form |
Natural language processing technology | conversion of human language(structured/unstructured)into data that can be translated and then manipulated by computer systems, branch of artificial intelligence |
Quadgram | a term used in voice recognition to indicate word sequences with four words |
Semantics | the meaning of a word or term; sometimes refers to comparable meaning, usually achieved through a standard vocabulary |
Speaker dependent | a speech recognition system that is "speaker dependent" means that it has been "trained" to a particular user and already has an established vocabulary for the user |
Speech recognition technology (SRT) | technology that translates speech to text |
Syntax | a term that refers to the comparable structure or format of data, usually as they are being transmitted from one system to another |
Template | a pattern used in computer- based patient records to capture data in a structured manner |
Trigram | a term used in voice recognition to indicate word sequences with three words |
Turnaround time (TAT) | defined as "the elapsed time from completion of dictation to the delivery of the transcribed document either in printed medium or electronically to a repository |
VoiceXML (VXML) | the standard "with which voice applications are developed", voice extensible markup language |