US20220189462A1 - METHOD OF TRAINING A SPEECH RECOGNITION MODEL OF AN EXTENDED LANGUAGE BY SPEECH IN A SOURCE LANGUAGE