Resource-Efficient Fine-Tuning Strategies for Automatic MOS Prediction in Text-to-Speech for Low-Resource Languages
This work addresses resource-efficient development of TTS systems for low-resource languages, offering incremental improvements in MOS prediction and listening test design.
The paper tackled the problem of predicting Mean Opinion Scores (MOS) for text-to-speech systems in low-resource languages, finding that pre-training on BVCC before fine-tuning on SOMOS yields the best accuracy, and that using over 30% of data does not significantly improve results.
We train a MOS prediction model based on wav2vec 2.0 using the open-access data sets BVCC and SOMOS. Our test with neural TTS data in the low-resource language (LRL) West Frisian shows that pre-training on BVCC before fine-tuning on SOMOS leads to the best accuracy for both fine-tuned and zero-shot prediction. Further fine-tuning experiments show that using more than 30 percent of the total data does not lead to significant improvements. In addition, fine-tuning with data from a single listener shows promising system-level accuracy, supporting the viability of one-participant pilot tests. These findings can all assist the resource-conscious development of TTS for LRLs by progressing towards better zero-shot MOS prediction and informing the design of listening tests, especially in early-stage evaluation.