基於原始波形的端到端DNN中避免說話人過擬合的文本無關說話人驗證

Avoiding Speaker Overfitting in End-to-End DNNs using Raw Waveform for Text-Independent Speaker Verification   基於原始波形的端到端DNN中避免說話人過擬合的文本無關說話人驗證 Avoiding Speaker Overfitting in End-to-End DNNs using Ra

>>阅读原文<<