# AmbiSep: Ambisonic-to-Ambisonic Reverberant Speech Separation Using Transformer Networks

arXiv: https://arxiv.org/abs/2206.06184 · HTML: https://mlsanity.com/paper/2206.06184/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Jun 2022.

## Built on by (most cited first, 1 of 1)

- [Multi-Microphone Speaker Separation by Spatial Regions](https://mlsanity.com/paper/2303.07143/) (2023)

## Abstract (arXiv)

Consider a multichannel Ambisonic recording containing a mixture of several reverberant speech signals. Retreiving the reverberant Ambisonic signals corresponding to the individual speech sources blindly from the mixture is a challenging task as it requires to estimate multiple signal channels for each source. In this work, we propose AmbiSep, a deep neural network-based plane-wave domain masking approach to solve this task. The masking network uses learned feature representations and transformers in a triple-path processing configuration. We train and evaluate the proposed network architecture on a spatialized WSJ0-2mix dataset, and show that the method achieves a multichannel scale-invariant signal-to-distortion ratio improvement of 17.7 dB on the blind test set, while preserving the spatial characteristics of the separated sounds.