WERD: Using social text spelling variants for evaluating dialectal speech recognition.
Ahmed AliPreslav NakovPeter BellSteve RenalsPublished in: ASRU (2017)
Keyphrases
- speech recognition
- hidden markov models
- automatic speech recognition
- language model
- speech processing
- speech recognizer
- speech signal
- speech recognition technology
- speech synthesis
- pattern recognition
- information retrieval
- speaker identification
- noisy environments
- handwriting recognition
- text mining
- speaker dependent
- speech retrieval
- speech recognition errors
- text data
- speaker independent
- semantic information
- speech recognition systems
- conversational speech
- speech recognizers
- keyword spotting
- english text
- speaker recognition
- probabilistic model
- computer vision