Project Icon

NISQA

Deep Learning-Based Model for Evaluating Speech Quality and Naturalness

Product DescriptionNISQA v2.0 offers an approach to speech quality and naturalness prediction with enhanced accuracy across multiple dimensions, including noisiness and loudness. It supports naturalness estimation for synthetic speech and allows for model training using CNN, Self-Attention, and LSTM. Access over 14,000 speech samples for various speech communication applications.
Project Details