vosk model gives empty responses using gradio.

I'm trying to use gradio with Vosk but it seems the model is not predicting speech. I'm using `test_gradio.py` to predict the speech.

Here is the app console output.

I have added `localhost` to save the origin

`partial` is always empty.