Optimized Power Normalized Cepstral Coefficients towards Robust Deep Speaker Verification
This is an incremental improvement for speaker verification systems, enhancing robustness in both in-domain and cross-domain scenarios.
The paper tackled the problem of PNCC features suppressing speaker variations in deep speaker verification by optimizing them through ablating the medium-time processor and introducing channel energy normalization, resulting in a 5.8% relative improvement on VoxCeleb1 and 61.2% on VoxMovies.
After their introduction to robust speech recognition, power normalized cepstral coefficient (PNCC) features were successfully adopted to other tasks, including speaker verification. However, as a feature extractor with long-term operations on the power spectrogram, its temporal processing and amplitude scaling steps dedicated on environmental compensation may be redundant. Further, they might suppress intrinsic speaker variations that are useful for speaker verification based on deep neural networks (DNN). Therefore, in this study, we revisit and optimize PNCCs by ablating its medium-time processor and by introducing channel energy normalization. Experimental results with a DNN-based speaker verification system indicate substantial improvement over baseline PNCCs on both in-domain and cross-domain scenarios, reflected by relatively 5.8% and 61.2% maximum lower equal error rate on VoxCeleb1 and VoxMovies, respectively.