Multilingual Neural Network Acoustic Modelling for ASR of Under-Resourced English-isiZulu Code-Switched Speech.
Astik BiswasFebe de WetEwald van der WesthuizenEmre YilmazThomas NieslerPublished in: INTERSPEECH (2018)
Keyphrases
- automatic speech recognition
- broadcast news
- neural network
- language resources
- speech sounds
- speech recognition
- acoustic features
- speech recognizers
- acoustic models
- speech signal
- cross language
- speech segments
- speech recognizer
- speech recognition systems
- cross lingual
- speech retrieval
- cross language ir
- spontaneous speech
- speaker independent
- spoken language
- word error rate
- cross language information retrieval
- speaker identification
- hidden markov models
- text to speech
- conversational speech
- multi lingual
- pattern recognition
- prosodic features
- parallel corpus
- machine translation
- language specific
- speech corpus
- formant frequencies
- recognition errors
- language identification
- back propagation
- multilingual information retrieval
- language model
- speech synthesis
- source code
- cross lingual information retrieval
- english text
- language independent
- spoken document retrieval
- text retrieval
- spoken words
- noisy environments
- comparable corpora
- cl sr
- out of vocabulary
- indian languages
- artificial neural networks
- sound source
- question answering
- document images
- vocal tract
- chinese english
- document collections
- non stationary
- audio stream
- document retrieval
- natural language
- language learning
- target language
- machine translation system
- information retrieval