Real-life voice activity detection with LSTM Recurrent Neural Networks and an application to Hollywood movies
Acoustics, Speech and Signal Processing(2013)
摘要
A novel, data-driven approach to voice activity detection is presented. The approach is based on Long Short-Term Memory Recurrent Neural Networks trained on standard RASTA-PLP frontend features. To approximate real-life scenarios, large amounts of noisy speech instances are mixed by using both read and spontaneous speech from the TIMIT and Buckeye corpora, and adding real long term recordings of diverse noise types. The approach is evaluated on unseen synthetically mixed test data as well as a real-life test set consisting of four full-length Hollywood movies. A frame-wise Equal Error Rate (EER) of 33.2% is obtained for the four movies and an EER of 9.6% is obtained for the synthetic test data at a peak SNR of 0 dB, clearly outperforming three state-of-the-art reference algorithms under the same conditions.
更多查看译文
关键词
error statistics,recurrent neural nets,signal detection,speech processing,speech recognition,Hollywood movies,LSTM recurrent neural networks,RASTA-PLP frontend feature,TIMIT and Buckeye corpora,equal error rate,long short-term memory recurrent neural networks,read speech,real life voice activity detection,real long term recording,spontaneous speech,state of the art reference algorithm,Long Short-Term Memory,Neural Networks,Speech Detection,Voice Activity Detection
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络