Abstract: In this work, we propose CleanMel, a single-channel Mel-spectrogram denoising and dereverberation network for improving both speech quality and automatic speech recognition (ASR) performance ...

Abstract: Neural vocoders often struggle with aliasing in latent feature spaces, caused by time-domain nonlinear operations and resampling layers. Aliasing folds high-frequency components into the low ...

Diffusion Speech is a diffusion-based text-to-speech model. Our speech synthesis pipeline is quite simple. We use a diffusion transformer model (DiT) to predict the duration of each phoneme. Then we ...

Creates unit tests using the MATLAB Testing Framework. Generates test classes, test methods, and test suites following best practices: ...