With the probabilistic sampling, the first experiment got a 6.3% relative phone error rate (PER) reduction compared to the conventional DNN baseline; the second ...
ABSTRACT. In HMM/DNN automatic speech recognition (ASR) system- s, the DNNs model the posterior probabilities for triphone states.
Improving HMM/DNN in asr of under-resourced languages using probabilistic sampling. M Song, Q Zhang, J Pan, Y Yan. 2015 IEEE China Summit and International ...
Improving HMM/DNN in ASR of under-resourced languages using probabilistic sampling. M. Song, Q. Zhang, J. Pan, и Y. Yan. ChinaSIP, стр. 20-24. IEEE, (2015 ). 1.
Documenting endangered languages supports the historical preservation of diverse cultures. Automatic speech recognition. (ASR), while potentially very ...
We investigate two strategies to improve the contextdependent deep neural network hidden Markov model (CDDNN- HMM) in low-resource speech recognition.
Missing: probabilistic | Show results with:probabilistic
We describe a novel way to implement subword language models in speech recognition systems based on weighted finite state transducers, hidden Markov models, ...
Nov 10, 2023 · This paper reviews the research status of feature extraction and acoustic models, and conducts research on resource expansion.
Abstract. We investigate two strategies to improve the context- dependent deep neural network hidden Markov model (CD-. DNN-HMM) in low-resource speech ...
Jun 21, 2024 · ASR was trained in four languages without native transcripts. Adaptation using mismatched crowdsourcing significantly outperformed self-training ...