BROWSE
Detail
Conference Paper
Harp-Net: Hyper-Autoencoded Reconstruction Propagation for Scalable Neural Audio Coding
Cited 16 time in 
Share 
Share
- Authors
- Darius Petermann, Seungkwon Beack, Minje Kim
- Issue Date
- 2021-10
- Citation
- Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA) 2021, pp.1-5
- Language
- English
- Type
- Conference Paper
- Abstract
- We propose a novel autoencoder architecture that improves the architectural scalability of general-purpose neural audio coding models. An autoencoder-based codec employs quantization to turn its bottleneck layer activation into bitstrings, a process that hinders information flow between the encoder and decoder parts. To circumvent this issue, we employ additional skip connections between the corresponding pair of encoder-decoder layers. The assumption is that, in a mirrored autoencoder topology, a decoder layer reconstructs the intermediate feature representation of its corresponding encoder layer. Hence, any additional information directly propagated from the corresponding encoder layer helps the reconstruction. We implement this kind of skip connections in the form of additional autoencoders, each of which is a small codec that compresses the massive data transfer between the paired encoder-decoder layers. We empirically verify that the proposed hyper-autoencoded architecture improves perceptual audio quality compared to an ordinary autoencoder baseline.
- KSP Keywords
- Audio coding, Audio quality, Data transfer, Encoder and Decoder, Feature Representation, Information Flow, Massive Data, additional information, coding models, skip connections