Ran for 1000 epochs using custom dataset and encountered non english voice

Have been running on jupyterlab with a custom dataset in the style of LIJspeech, hence have been using all Lijspeech configuration of TTS. But after 1000 epochs it gives an unnatural non english voice… I have 1 hour of english audio.The inference text is given in the pic below

Link to audio (Please listen) - https://drive.google.com/file/d/1aYj910oZsuKlUXYLEwkj_bK_Z4vYwhdk/view?usp=sharing