I am pleased to announce the source code availability of a new end-to-end speech synthesis system. The project’s website is https://tiberiu44.github.io/TTS-Cube/index.html, and the code can be freely accessed on GitHub https://github.com/tiberiu44/TTS-Cube. The examples are built using a Romanian speech synthesis corpus, but the system can be trained for any other language.
It has a dual architecture, composed of a network that generates Mel-spectrograms from raw text (character/phoneme list) and a vocoder that is conditioned to produce RAW signal.
Hope you find this useful, Tiberiu
-------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 5069 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20180723/dccefe05/attachment.txt>