HanTTS

Chinese Text-to-Speech(TTS)

汉字 => ["han4", "zi4"] => .wav audio

Read this page in 简体中文

Libraries Used

For turning Chinese characters into pinyin

For processing .wav files

Web backend

Express

A list of all Chinese characters is obtained from 倉頡平台2012, a Chinese input software.

Build and Run

git clone https://github.com/junzew/HanTTS.git
cd HanTTS
pip install --user -r requires.txt

Download syllables.zip from SourceForge, and decompress under the directory HanTTS.

Either run locally: python main.py
Or through web interface: cd into the server folder
```
 npm install
 node app.js
```
Navigate to localhost:3000 in a browser

Use your own voice

Record five tones of each pinyin listed in mapping.json
Group them by the first letter (a,b,c,d, etc.), and save under folder ./recording as {letter}.wav
Then run python process.py {letter}for each letter to split recordings into individual pinyin
Move valid .wav files from ./pre to the folder ./syllables

Deploy to Heroku

heroku create
heroku git:remote -a <app-name>
heroku buildpacks:set https://github.com/heroku/heroku-buildpack-nodejs#v170 -a <app-name>
heroku buildpacks:add --index 1 heroku-community/apt
heroku buildpacks:add --index 3 heroku/python

$heroku buildpacks
=== hantts Buildpack URLs
1. heroku-community/apt
2. https://github.com/heroku/heroku-buildpack-nodejs#v170
3. heroku/python

heroku apps:rename <newname>
heroku ps:scale web=1
git push heroku master

Issues Encountered during Deployment:

Failed to bind to port https://stackoverflow.com/questions/15693192/heroku-node-js-error-web-process-failed-to-bind-to-port-within-60-seconds-of app.listen(process.env.PORT || 3000, function () {...}
node Procfile
apt Aptfile
python requirements.txt

Based on the text-to-speech-sample project by @alexram1313