Speech Magnitude-Spectrum Information-Entropy (MSIE) for Automatic Speech Recognition in Noisy Environments.
Juan Arturo Nolazco-FloresRoberto A. Aceves L.L. Paola García-PereraPublished in: ICPR (2010)
Keyphrases
- information entropy
- automatic speech recognition
- noisy environments
- speech recognition
- speech signal
- spectral subtraction
- word error rate
- speech enhancement
- speaker verification
- broadcast news
- rough sets
- conversational speech
- digit recognition
- background noise
- speaker identification
- decision table
- hidden markov models
- recognition errors
- speech corpus
- spontaneous speech
- acoustic features
- spoken words
- speech synthesis
- voice activity detection
- speech retrieval
- rough set theory
- speech sounds
- speech recognizer
- word recognition
- attribute values
- linear prediction
- noisy speech
- language model
- non stationary
- attribute reduction
- vocal tract
- handwriting recognition
- pattern recognition
- image coding
- probabilistic model
- data mining