After the application starts up, you will see the very UI with the 3 Test Parts shown vertically.

Click the button on the bottom to start configeration.

If the connection is NOT established, there is small red dot on the most bottom shows that 'There is no connection" to your Local server on which LM Studio is running.

Modify the 'localhost" to the Local server's IP provided by LM Studio, for example, if IP address in LM Stuido 192.168.1.230, please replace the 'localhost' with this IP adress. 

The Model Name is the LLM you chose to run in the LM Studio, please refer to the API Usage infomation in the Info Tab. In this case, we chose the google/gemma-3-12b as a default.

Click the 'Test la connecxion' button in red rectangle. If the local server's IP address coulb be reachable , you will see a pop-up lists the  enumeration of all the Modles have been downloaded in the LM Studio, which identifies a successful connection establishment.

Then click the button 'Fermer' (in gree rectangle) to go back to the main page.

A successful connection setting would be automatically saved for the next time use. To ensure the connection, you could test the connection again when the app relaunched.

Choosing any of those 3 parts could launch a speaking exam practice, but you have to wait a moment for the LLM to generate a corresponding question.

For exmple in part1, it's a conversation lasting for 2 minutes.

When it's your turn to speak, the app will automatically start the timer and a voice-to-text recording, but your could control either the timer (by click the Pause button in yellow rectangle) or the recording (by click the 'Stop recording' button in gree rectangle).

By Clicking the 'Terminer l'examen' button or 'x' (both in red rectangles), the practice would be terminated and then a scoring and evaluation pages would be shown to inform your score you got and the level you are in.