| 講演抄録/キーワード |
| 講演名 |
2023-10-14 17:05
Electrolaryngeal Speech Enhancement through Strong Linguistic Encoding Methods ○Lester Phillip Violeta・Wen-Chin Huang・Ding Ma・Ryuichi Yamamoto・Kazuhiro Kobayashi・Tomoki Toda(Nagoya Univ.) SP2023-33 WIT2023-24 |
| 抄録 |
(和) |
Although pretraining and fine-tuning approaches have proven to work well in speech intelligibility enhancement, various mismatches, such as the speech type mismatch or speaker mismatches between the datasets used in each stage, can deteriorate the conversion performance of this framework. We propose a linguistic encoder robust enough to project both EL and typical speech in the same latent space, while still being able to extract accurate linguistic information, creating a unified representation to reduce the speech type mismatch. Furthermore, we introduce HuBERT output features to the proposed framework for reducing the speaker mismatch. Such a framework makes it possible to effectively use a large-scale parallel dataset during pretraining. We show that compared to the conventional framework using mel-spectrogram input and output features, using the proposed framework enables the model to synthesize more intelligible and naturally sounding speech, as shown by a significant 16% improvement in character error rate and 0.83 improvement in naturalness score. |
| (英) |
Although pretraining and fine-tuning approaches have proven to work well in speech intelligibility enhancement, various mismatches, such as the speech type mismatch or speaker mismatches between the datasets used in each stage, can deteriorate the conversion performance of this framework. We propose a linguistic encoder robust enough to project both EL and typical speech in the same latent space, while still being able to extract accurate linguistic information, creating a unified representation to reduce the speech type mismatch. Furthermore, we introduce HuBERT output features to the proposed framework for reducing the speaker mismatch. Such a framework makes it possible to effectively use a large-scale parallel dataset during pretraining. We show that compared to the conventional framework using mel-spectrogram input and output features, using the proposed framework enables the model to synthesize more intelligible and naturally sounding speech, as shown by a significant 16% improvement in character error rate and 0.83 improvement in naturalness score. |
| キーワード |
(和) |
Intelligibility enhancement / Electrolaryngeal speech / Atypical speech / / / / / |
| (英) |
Intelligibility enhancement / Electrolaryngeal speech / Atypical speech / / / / / |
| 文献情報 |
信学技報, vol. 123, no. 212, SP2023-33, pp. 33-38, 2023年10月. |
| 資料番号 |
SP2023-33 |
| 発行日 |
2023-10-07 (SP, WIT) |
| ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
| PDFダウンロード |
SP2023-33 WIT2023-24 |
|