Improving Speech Intelligibility in Noise Using a Binary Mask That Is Based on Magnitude Spectrum Constraints.
Gibak KimPhilipos C. LoizouPublished in: IEEE Signal Process. Lett. (2010)
Keyphrases
- speech recognition
- spectral subtraction
- signal to noise ratio
- speech signal
- noisy environments
- non binary
- speech enhancement
- power spectrum
- noise reduction
- missing data
- additive noise
- hidden markov models
- restrictive assumptions
- constraint programming
- information retrieval
- automatic speech recognition
- image processing
- constraint satisfaction
- geometric constraints
- median filter
- noise model