CACRN-Net: A 3D log Mel spectrogram based channel attention convolutional recurrent neural network for few-shot speaker identification.
Banala SarithaMohammad Azharuddin LaskarAnish Monsley K.Rabul Hussain LaskarMadhuchhanda ChoudhuryPublished in: Comput. Electr. Eng. (2024)
Keyphrases
- recurrent neural networks
- speaker identification
- speech signal
- speech recognition
- neural network
- speaker recognition
- recurrent networks
- feed forward
- gaussian mixture model
- automatic speech recognition
- complex valued
- noisy environments
- feature extraction
- broadcast news
- hidden layer
- echo state networks
- non stationary
- video sequences
- artificial neural networks
- video shots
- video data
- hidden markov models
- pattern recognition
- audio signal
- visual attention
- visual features
- audio features
- support vector
- computer vision
- genetic algorithm