Phonetic and Prosody-aware Self-supervised Learning Approach for Non-native Fluency Scoring. Speech fluency/disfluency can be evaluated by analyzing a range of phonetic and prosodic features. Deep neural networks are commonly trained to map fluency-related features into the human scores.
May 19, 2023
In this paper, we propose a self-supervised learn- ing approach that incorporates phonetic and prosodic informa- tion to improve non-native fluency scoring. ...
[Show full abstract] takes into account phonetic and prosody awareness for fluency scoring. Specifically, we first pre-train the model using a reconstruction ...
May 19, 2023 · Speech fluency/disfluency can be evaluated by analyzing a range of phonetic and prosodic features. Deep neural net-.
May 25, 2023 · Bibliographic details on Phonetic and Prosody-aware Self-supervised Learning Approach for Non-native Fluency Scoring.
Deep learning-based pronunciation scoring models highly rely on the availability of the annotated non-native data, which is costly and has scalability issues.
Phonetic and Prosody-aware Self-supervised Learning Approach for Non-native Fluency Scoring ... Specifically, we first pre-train the model using a reconstruction ...
Speech fluency/disfluency can be evaluated by analyzing a range of phonetic and prosodic features. Deep neural networks are commonly trained to map ...
Phonetic and Prosody-aware Self-supervised Learning Approach for Non-native Fluency Scoring ... Specifically, we first pre-train the model using a reconstruction ...
Phonetic and Prosody-aware Self-supervised Learning Approach for Non-native Fluency Scoring ... Non-native Speech in Automated Pronunciation Assessment. S ...