Voice Recognition

1
Voice Recognition

• By Scott Orphan

2
Overview

• What is voice/speech recognition?
• Types of recognition
• Advantages disadvantages of different
  recognition approaches
• Applications of voice recognition
• Difficulties of recognition
• Future of voice recognition
• Demonstration

3
Speech Recognition Must

• Identify the sound of a human voice
• Uses the physics of sound
• Factor out environmental noise
• Convert the acoustic signal to a stream of words
• Accept messages as input for controlling the
  system

4
Two Categories of Voice Recognition

• Discrete speech recognition
• Isolated word and phrase recognition
• Connected word recognition
• Continuous speech recognition

5
Isolated Word Recognition

• Most simple form
• Uses pattern matching
• Single words, separated by pauses
• Speech compared to list of word templates
• Used by automated operator systems

6
Connected Word Recognition

• Continuation of isolated word recognition
• System learns fluid sequences of its vocabulary
  words
• Examples Credit card numbers
• Telephone numbers

7
Analysis Of Discrete Voice Recognition

• Speech not natural or easy
• Specific commands (limited vocabulary)
• No grammatical or syntactic interpretation
• Rely only on phonological input
• Accept vs. Except

8
Continuous Voice Recognition

• A more complex system
• Ability to speak in an everyday manner
• Tries to recognize and understand speech
• No specific or learned commands
• May use hidden Markov modeling, neural networks,
  dynamic time warping

9
Continuous Voice Recognition

• Error prone
• Expensive
• Requires a lot of computational power
• Two types
• speaker dependent
• speaker independent

10
Speaker Dependent Systems

• Text read, voice speech pattern analyzed
• Lacks flexibility, cannot be shared
• Less costly
• More accurate
• Used by most commercial software

11
Speaker Independent Systems

• Understands multiple users of a certain language
  type
• No enrollment period
• Greater flexibility
• More error prone expensive
• Tends to be used for specialized, single-task
  systems

12
Components Of A Speech Recognition System
13
Voice Interactive System
14
Applications Of Voice Recognition

• 40 billion market
• Post office for speed mail delivery
• Wal-Mart warehouse facilities
• Word processing / Dictation
• ViaVoice
• Voice Xpress
• Speechworks
• Voice Verification
• Many more

15
Voice Verification
16
Difficulties In Voice Recognition

• Unpredictable errors, signal acoustic
  variability
• Phonetic variability
• Within speaker variability
• Across speaker variability
• Peoples fears and expectations
• Multi-modal communication
• Spoken language is different

17
Future Of Voice Recognition

• Better rejection of extraneous speech
• Better recognition of embedded commands
• Better efficiency on low cost processors
• Standards for performance evaluation
• Increased portability
• Lower error rates
• Improve overall robustness