USM-Lite: Quantization and Sparsity Aware Fine-Tuning for Speech Recognition with Universal Speech Models.
Shaojin DingDavid QiuDavid RimYanzhang HeOleg RybakovBo LiRohit PrabhavalkarWeiran WangTara N. SainathZhonglin HanJian LiAmir YazdanbakhshShivani AgrawalPublished in: ICASSP (2024)
Keyphrases
- speech recognition
- fine tuning
- speech synthesis
- acoustic models
- speech signal
- automatic speech recognition
- hidden markov models
- speech processing
- speech recognition systems
- speech recognizer
- language model
- noisy environments
- speaker identification
- probabilistic model
- pattern recognition
- speech recognition technology
- speaker independent
- word error rate
- speaker dependent
- isolated word
- keyword spotting
- recognition engine
- handwriting recognition
- speech enhancement
- speaker recognition
- speaker adaptation
- speech recognition errors
- information retrieval
- speech recognizers
- machine learning