Real-life voice activity detection with LSTM Recurrent Neural Networks and an application to Hollywood movies.
Florian EybenFelix WeningerStefano SquartiniBjörn W. SchullerPublished in: ICASSP (2013)
Keyphrases
- recurrent neural networks
- voice activity detection
- real life
- long short term memory
- action recognition
- noisy environments
- feed forward
- neural network
- recurrent networks
- artificial neural networks
- complex valued
- neural model
- speech recognition
- echo state networks
- feedforward neural networks
- nonlinear dynamic systems
- reservoir computing
- video data
- human activities
- cascade correlation
- pattern recognition
- multiscale
- multimedia
- real time
- hidden markov models