Enforcing equal CNN hidden activations for parallel recordings from different microphones improves target-device accuracy in acoustic scene classification and beats MMD-based domain adaptation in these experiments.
Our DA loss is designed to enforce equal hidden layer representations for different devices by exploiting time-aligned recordings
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Exploiting Parallel Audio Recordings to Enforce Device Invariance in CNN-based Acoustic Scene Classification
Enforcing equal CNN hidden activations for parallel recordings from different microphones improves target-device accuracy in acoustic scene classification and beats MMD-based domain adaptation in these experiments.