Bypass Temporal Classification: Weakly Supervised Automatic Speech Recognition with Imperfect Transcripts.
Dongji GaoMatthew WiesnerHainan XuLeibny Paola GarcíaDaniel PoveySanjeev KhudanpurPublished in: INTERSPEECH (2023)
Keyphrases
- automatic speech recognition
- weakly supervised
- speech recognition
- broadcast news
- speech signal
- pattern recognition
- hidden markov models
- spontaneous speech
- image classification
- machine learning
- superpixels
- object class
- unsupervised learning
- feature vectors
- feature space
- support vector
- relation extraction
- graph cuts
- topic models
- training examples
- language model
- supervised learning
- multi class
- training set
- feature extraction