Integer-Only Approximated MFCC for Ultra-Low Power Audio NN Processing on Multi-Core MCUs.
Marco FariselliManuele RusciJoel CambonieEric FlamandPublished in: AICAS (2021)
Keyphrases
- neural network
- nearest neighbor
- feature extraction
- information processing
- real time
- artificial neural networks
- ultra low power
- speaker identification
- neural network model
- multimedia
- signal processing
- data processing
- linear combination
- mel frequency cepstral coefficients
- knn
- audio features
- speech recognition
- emotion recognition
- multilayer perceptron
- gaussian mixture model
- speaker recognition
- genetic algorithm