Noise-robust automatic speech recognition using orthogonalized distinctive phonetic feature vectors.
Takashi FukudaTsuneo NittaPublished in: INTERSPEECH (2003)
Keyphrases
- feature vectors
- speech recognition
- noisy environments
- image noise
- background noise
- salt pepper
- greater robustness
- speaker identification
- speech signal
- speech music discrimination
- speech enhancement
- speech synthesis
- similarity measure
- feature space
- audio visual
- imaging artifacts
- speech segments
- cepstral coefficients
- speech recognizer
- fully automatic
- signal to noise ratio
- noise reduction
- face images
- spoken document retrieval
- language model
- support vector machine
- gaussian mixture model
- texture features
- speech sounds
- feature selection