Toji – 東寺

Toji is standalone web application for collecting a large amount of voices read aloud from the manuscript efficiently.

Collect voice readings easily! All you need is the manuscripts to read out. The manuscript will appear on the screen, and then press the Record button to record the voice corresponding to the manuscript. Finally, you can download the recorded voices as a zip file.

Toji (東寺) is the name of Japanese temple. The name of this repository is derived from a phrase that appears in the lead of the first sentence of Voice Actress Corpus.

Usage

Start-up

You can use docker,

$ docker build -t toji .
$ docker run -p 8501:8501 --rm -t -i toji

or start locally.

$ poetry install
$ poetry run streamlit run toji/app.py

And access to http://localhost:8501/

Use Web App

  1. Prepare the manuscripts with line breaks and paste it into the text area.
  2. Press START button to start recording.
  3. When you have finished reading, press the STOP button.
  4. Press Next to move on to the next utterance.
  5. When all tasks are finished, click Proceed to Download button to download the wav files.

GitHub

View Github