Feedback Recurrent AutoEncoder
This work addresses efficient speech compression for applications requiring low bitrate transmission or storage, though it appears incremental as it builds on existing autoencoder and neural vocoder techniques.
The paper tackles online compression of sequential data with temporal dependencies by proposing a Feedback Recurrent AutoEncoder (FRAE) architecture, which achieves high-quality speech waveform compression at low fixed and variable bitrates when paired with a neural vocoder and entropy coder.
In this work, we propose a new recurrent autoencoder architecture, termed Feedback Recurrent AutoEncoder (FRAE), for online compression of sequential data with temporal dependency. The recurrent structure of FRAE is designed to efficiently extract the redundancy along the time dimension and allows a compact discrete representation of the data to be learned. We demonstrate its effectiveness in speech spectrogram compression. Specifically, we show that the FRAE, paired with a powerful neural vocoder, can produce high-quality speech waveforms at a low, fixed bitrate. We further show that by adding a learned prior for the latent space and using an entropy coder, we can achieve an even lower variable bitrate.