IMaSC -- ICFOSS Malayalam Speech Corpus
This addresses the problem of limited resources for TTS development in Malayalam, benefiting speakers and developers, but it is incremental as it primarily provides a new dataset.
The authors tackled the lack of high-quality speech corpora for Malayalam, a low-resource language, by creating IMaSC, a dataset with 50 hours of recorded speech from 8 speakers and 34,473 text-audio pairs, and used it to train TTS models that achieved an average mean opinion score of 4.50, indicating speech close to human quality.
Modern text-to-speech (TTS) systems use deep learning to synthesize speech increasingly approaching human quality, but they require a database of high quality audio-text sentence pairs for training. Malayalam, the official language of the Indian state of Kerala and spoken by 35+ million people, is a low resource language in terms of available corpora for TTS systems. In this paper, we present IMaSC, a Malayalam text and speech corpora containing approximately 50 hours of recorded speech. With 8 speakers and a total of 34,473 text-audio pairs, IMaSC is larger than every other publicly available alternative. We evaluated the database by using it to train TTS models for each speaker based on a modern deep learning architecture. Via subjective evaluation, we show that our models perform significantly better in terms of naturalness compared to previous studies and publicly available models, with an average mean opinion score of 4.50, indicating that the synthesized speech is close to human quality.