The NISQA-TTS model weights can be used to estimate the Naturalness of synthetic speech generated by a Voice Conversion or Text-To-Speech system (Siri, Alexa, etc.). NISQA can be used to train new ...