Learnable MFCCs for Speaker Verification
Xuechen Liu, Md Sahidullah, Tomi Kinnunen
Abstract
We propose a learnable mel-frequency cepstral coefficients (MFCCs) front-end architecture for deep neural network (DNN) based automatic speaker verification. Our architecture retains the simplicity and interpretability of MFCC-based features while allowing the model to be adapted to data flexibly. In practice, we formulate data-driven version of four linear transforms in a standard MFCC extractor - windowing, discrete Fourier transform (DFT), mel filterbank and discrete cosine transform (DCT). Results reported reach up to 6.7% (VoxCeleb1) and 9.7% (SITW) relative improvement in term of equal error rate (EER) from static MFCCs, without additional tuning effort.
Topics & Concepts
Mel-frequency cepstrumDiscrete cosine transformComputer scienceInterpretabilityFilter bankSpeech recognitionModified discrete cosine transformArtificial neural networkWord error rateArtificial intelligencePattern recognition (psychology)Feature extractionTransform codingComputer visionFilter (signal processing)Image (mathematics)Speech Recognition and SynthesisSpeech and Audio ProcessingMusic and Audio Processing