This script is a prototyp for an automated tutorial generator which can be used on any website to create tutorial videos. The voices can be generated with PIPER or coqui TTS a text to speech system. The actions are performed by Playwright.
- Create virtual environment:
python3 -m venv venv - Activate environment:
source venv/bin/activate sudo apt install espeak-ngpip install -r requirements.txtplaywright install
If you are using piper:
- Install piper under
./piper, see https://github.com/rhasspy/piper#installation. The executable should be located under./piper/piper. You can also set an alternative path with the-pargument. - Download a voice from https://github.com/rhasspy/piper/releases/tag/v0.0.2. When executing the script, pass the
-margument with the voice model path of the.onnxmodel file.
The following steps show how to use this script.
First you need to replace the highlighted Playwright code in the run(..) method. You can manually implement the code or use the Playwright code generator by the command playwright codegen [url]. If you use the generator you can replace the example code with the generated code. Note: Only the page related code should be copied. Preferably, you should orientate on the example.
You can insert speech between actions by calling the page.voice(..) function. You can wait for the end of a speech by page.wait_for_voice().
The voices can be translated into different languages by translation files which can be passed by the -t argument. The translation file contains a json object with key value pairs. Each key identifies a voice string and its value contains the corresponding translation. The voice string <key> passed to the page.voice(<key>) function will be replaced by the matching translation.
Important elements on the page can be highlighted by a red border with the Locator's mark(..) function. For example, the code page.get_by_role("button", name="Save").mark() highlights the save button.
Timeouts between actions can be set with page.wait_for_timeout(..). The general timeout between actions is set to one second and can be set with the -s argument in milliseconds.
Finally, run the script with python3 tutorial_generator.py -m <path-to-voice-model>. Run python3 tutorial_generator.py -h to see all available arguments.