A CHiME-3 challenge system: Long-term acoustic features for noise robust automatic speech recognition.
Niko MoritzStephan GerlachKamil AdilogluJörn AnemüllerBirger KollmeierStefan GoetzePublished in: ASRU (2015)
Keyphrases
- automatic speech recognition
- acoustic features
- noisy environments
- speech recognition
- speech signal
- speaker verification
- hidden markov models
- broadcast news
- conversational speech
- visual features
- speech retrieval
- speaker identification
- music information retrieval
- noise reduction
- language model
- image processing
- multimedia
- computer vision