This project is designed to make English listening audio for English learners in China, but it can also serve other purposes. Microsoft uses deep neural networks, to create a Text to Speech service makes the voices of computers expressive and nearly indistinguishable from natural spoken voice.
- Uses Microsoft Azure Neural Text-to-speech voices to synthesize
- Partial SSML support, which includes the Microsoft supported tags and a custom
<repeat>that will repeat the designated text. - Tailors the input text to meet the requirements of Microsoft, so that you don't need to do it yourself.
- Friendly SSML editor. You don't need to type the tags yourself.
- Dialog Maker to help you transform speaker indicators like "Tony:", "Betty:", etc. to the
<voice>tags so that these lines will be read by desired voices. - Automatically detects lines written in Chinese and have them read by a Chinese voice so that you won't have to manually insert the
<voice>tags. - Optimizes the requests to reduce the cost.
- Export the speech to WAV files.
- Clone this repo.
- Follow the guide in the "Getting started" section of Microsoft Azure Text to Speech and obtain your subscription key. Be sure to choose a region that supports neural voices.
- Edit the HTML file, search for
YOUR_SUBSCRIPTION_KEYand replace it with your subscription key. - Find
YOUR_REGIONand replace it with the region you've chosen, e.g.:westusfor West US. - Enjoy!
This project uses Microsoft Azure Services, jQuery, audiobuffer-to-wav.js.