It is an open-source (Apache License) free Text To Speech (TTS) service, that takes the text information from an issue request, as long as the issue title starts with TTS:, TTS-F: or TTS-CN:. After a few minutes, it generates audio, and the audio link would be added into the issue comment.
Samples are Alice’s Adventures in Wonderland, The Tyger By William Blake and Psalms .
Chinese sample 朱自清 背影 叶挺 囚歌 余光中 乡愁 关雎
More ones can be found here
Create a new issue start with TTS: (English best quality), TTS-F: (English quick response) or TTS-CN: (Chinese Mandarin) with a body text. After a few minutes, the issue would be closed with a ready-to-download link of the audio.
You can also choose 20+ different voices for TTS-CN: by using like TTS-CN: 1$, TTS-CN: 30$ more information can be found at TTS-Mandarin
This application uses all open-source projects and free services. There are three TTS engines.
Once an issue is created, GitHub action would be triggered: It first reads the body of the issue; next either use pre-trained data powered by Python Machine Learning (Parakeet), or built-in CPP voice (Mimic) to create the audio; then it compresses wave to mp3 and merges to mp3 branch; Finally, it links to the issue and closes the issue.
As an AI hobbyist, I am always amazed by Google, Alexia, and Siri. Those TTS help me a lot when reading books. However, those are commercial apps that can not be used freely. In addition, I also want to take advantage of GitHub action, a free service that has been out for quite a while, to create new toys.
Since I have played with different TTS engines before, it only took me about two days to wire those up with Github actions. And another couple of days to add Chinese Mandarin. I personally really like this TOY.
bitcoin:bc1qc2rpn0x0lzv8gdgx8nvhsrkx3jm3s0kpql4tfk?time=1614371095