# Towards Unified All-Neural Beamforming for Time and Frequency Domain Speech Separation

arXiv: https://arxiv.org/abs/2212.08348 · HTML: https://mlsanity.com/paper/2212.08348/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Dec 2022.

## Built on by (most cited first, 5 of 5)

- [ReZero: Region-customizable Sound Extraction](https://mlsanity.com/paper/2308.16892/) (2023)
- [SonicSieve: Bringing Directional Speech Extraction to Smartphones Using Acoustic Microstructures](https://mlsanity.com/paper/2504.10793/) (2025)
- [SelectTSL: Prompt-Guided Selective Target Sound Localization in Complex Scenarios](https://mlsanity.com/paper/2607.02343/) (2026)
- [RIR-SF: Room Impulse Response Based Spatial Feature for Target Speech Recognition in Multi-Channel Multi-Speaker Scenarios](https://mlsanity.com/paper/2311.00146/) (2023)
- [SpatialEmb: Extract and Encode Spatial Information for 1-Stage Multi-channel Multi-speaker ASR on Arbitrary Microphone Arrays](https://mlsanity.com/paper/2601.18037/) (2026)

## Abstract (arXiv)

Recently, frequency domain all-neural beamforming methods have achieved remarkable progress for multichannel speech separation. In parallel, the integration of time domain network structure and beamforming also gains significant attention. This study proposes a novel all-neural beamforming method in time domain and makes an attempt to unify the all-neural beamforming pipelines for time domain and frequency domain multichannel speech separation. The proposed model consists of two modules: separation and beamforming. Both modules perform temporal-spectral-spatial modeling and are trained from end-to-end using a joint loss function. The novelty of this study lies in two folds. Firstly, a time domain directional feature conditioned on the direction of the target speaker is proposed, which can be jointly optimized within the time domain architecture to enhance target signal estimation. Secondly, an all-neural beamforming network in time domain is designed to refine the pre-separated results. This module features with parametric time-variant beamforming coefficient estimation, without explicitly following the derivation of optimal filters that may lead to an upper bound. The proposed method is evaluated on simulated reverberant overlapped speech data derived from the AISHELL-1 corpus. Experimental results demonstrate significant performance improvements over frequency domain state-of-the-arts, ideal magnitude masks and existing time domain neural beamforming methods.