# Separate and Reconstruct: Asymmetric Encoder-Decoder for Speech Separation

arXiv: https://arxiv.org/abs/2406.05983 · HTML: https://mlsanity.com/paper/2406.05983/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Jun 2024.

## Built on by (most cited first, 14 of 14)

- [SoloSpeech: Enhancing Intelligibility and Quality in Target Speech Extraction through a Cascaded Generative Pipeline](https://mlsanity.com/paper/2505.19314/) (2025)
- [Towards AI-Driven Policing: Interdisciplinary Knowledge Discovery from Police Body-Worn Camera Footage](https://mlsanity.com/paper/2504.20007/) (2025)
- [Learning to Highlight Audio by Watching Movies](https://mlsanity.com/paper/2505.12154/) (2025)
- [MGE-LDM: Joint Latent Diffusion for Simultaneous Music Generation and Source Extraction](https://mlsanity.com/paper/2505.23305/) (2025)
- [SAO-Instruct: Free-form Audio Editing using Natural Language Instructions](https://mlsanity.com/paper/2510.22795/) (2025)
- [Knowing When to Quit: Probabilistic Early Exits for Speech Separation](https://mlsanity.com/paper/2507.09768/) (2025)
- [Efficient Audio-Visual Speech Separation with Discrete Lip Semantics and Multi-Scale Global-Local Attention](https://mlsanity.com/paper/2509.23610/) (2025)
- [When AVSR Meets Video Conferencing: Dataset, Degradation, and the Hidden Mechanism Behind Performance Collapse](https://mlsanity.com/paper/2603.22915/) (2026)
- [TF-MLPNet: Tiny Real-Time Neural Speech Separation](https://mlsanity.com/paper/2508.03047/) (2025)
- [Robust Target Speaker Diarization and Separation via Augmented Speaker Embedding Sampling](https://mlsanity.com/paper/2508.06393/) (2025)
- [TISDiSS: A Training-Time and Inference-Time Scalable Framework for Discriminative Source Separation](https://mlsanity.com/paper/2509.15666/) (2025)
- [Source Separation for A Cappella Music](https://mlsanity.com/paper/2509.26580/) (2025)
- [A Knowledge-Driven Approach to Music Segmentation, Music Source Separation and Cinematic Audio Source Separation](https://mlsanity.com/paper/2602.21476/) (2026)
- [Sommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language Models](https://mlsanity.com/paper/2603.25750/) (2026)

## Abstract (arXiv)

In speech separation, time-domain approaches have successfully replaced the time-frequency domain with latent sequence feature from a learnable encoder. Conventionally, the feature is separated into speaker-specific ones at the final stage of the network. Instead, we propose a more intuitive strategy that separates features earlier by expanding the feature sequence to the number of speakers as an extra dimension. To achieve this, an asymmetric strategy is presented in which the encoder and decoder are partitioned to perform distinct processing in separation tasks. The encoder analyzes features, and the output of the encoder is split into the number of speakers to be separated. The separated sequences are then reconstructed by the weight-shared decoder, which also performs cross-speaker processing. Without relying on speaker information, the weight-shared network in the decoder directly learns to discriminate features using a separation objective. In addition, to improve performance, traditional methods have extended the sequence length, leading to the adoption of dual-path models, which handle the much longer sequence effectively by segmenting it into chunks. To address this, we introduce global and local Transformer blocks that can directly handle long sequences more efficiently without chunking and dual-path processing. The experimental results demonstrated that this asymmetric structure is effective and that the combination of proposed global and local Transformer can sufficiently replace the role of inter- and intra-chunk processing in dual-path structure. Finally, the presented model combining both of these achieved state-of-the-art performance with much less computation in various benchmark datasets.