Cleanformer: A Multichannel Array Configuration-Invariant Neural Enhancement Frontend for ASR in Smart Speakers.
Joseph CaroselliArun NarayananNathan HowardTom O'MalleyPublished in: ICASSP (2023)
Keyphrases
- speech recognition
- automatic speech recognition
- network architecture
- back end
- image enhancement
- neural network
- multi channel
- noisy environments
- affine transformation
- bio inspired
- cross channel
- pattern recognition
- feature vectors
- learning rules
- smart environments
- moment invariants
- multiscale
- neural model
- spiking neural networks
- invariant properties
- neural fuzzy
- image processing