# CabinSep: IR-Augmented Mask-Based MVDR for Real-Time In-Car Speech Separation with Distributed Heterogeneous Arrays

arXiv: https://arxiv.org/abs/2509.01399 · HTML: https://mlsanity.com/paper/2509.01399/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 40 references that are in the map.
- First posted on arXiv 1 Sep 2025.

## Builds on (most cited first, 22 of 22)

- [Conformer: Convolution-augmented Transformer for Speech Recognition](https://mlsanity.com/paper/2005.08100/) (2020)
- [SDR - half-baked or well done?](https://mlsanity.com/paper/1811.02508/) (2018)
- [AISHELL-2: Transforming Mandarin ASR Research Into Industrial Scale](https://mlsanity.com/paper/1808.10583/) (2018)
- [WeNet: Production oriented Streaming and Non-streaming End-to-End Speech Recognition Toolkit](https://mlsanity.com/paper/2102.01547/) (2021)
- [FunAudioLLM: Voice Understanding and Generation Foundation Models for Natural Interaction Between Humans and LLMs](https://mlsanity.com/paper/2407.04051/) (2024)
- [Continuous Speech Separation with Conformer](https://mlsanity.com/paper/2008.05773/) (2020)
- [End-to-end Microphone Permutation and Number Invariant Multi-channel Speech Separation](https://mlsanity.com/paper/1910.14104/) (2019)
- [Recognizing Overlapped Speech in Meetings: A Multichannel Separation Approach Using Neural Networks](https://mlsanity.com/paper/1810.03655/) (2018)
- [Uformer: A Unet based dilated complex & real dual-path conformer network for simultaneous speech enhancement and dereverberation](https://mlsanity.com/paper/2111.06015/) (2021)
- [Embedding and Beamforming: All-neural Causal Beamformer for Multichannel Speech Enhancement](https://mlsanity.com/paper/2109.00265/) (2021)
- [Multi-microphone Complex Spectral Mapping for Utterance-wise and Continuous Speech Separation](https://mlsanity.com/paper/2010.01703/) (2020)
- [Audio-visual Multi-channel Integration and Recognition of Overlapped Speech](https://mlsanity.com/paper/2011.07755/) (2020)
- [Neural Blind Source Separation and Diarization for Distant Speech Recognition](https://mlsanity.com/paper/2406.08396/) (2024)
- [Multi-channel Multi-frame ADL-MVDR for Target Speech Separation](https://mlsanity.com/paper/2012.13442/) (2020)
- [The NPU-ASLP System for Audio-Visual Speech Recognition in MISP 2022 Challenge](https://mlsanity.com/paper/2303.06341/) (2023)
- [Distil-DCCRN: A Small-footprint DCCRN Leveraging Feature-based Knowledge Distillation in Speech Enhancement](https://mlsanity.com/paper/2408.04267/) (2024)
- [DualSep: A Light-weight dual-encoder convolutional recurrent network for real-time in-car speech separation](https://mlsanity.com/paper/2409.08610/) (2024)
- [Model-based estimation of in-car-communication feedback applied to speech zone detection](https://mlsanity.com/paper/2210.03363/) (2022)
- [Neural Speech Enhancement with Very Low Algorithmic Latency and Complexity via Integrated Full- and Sub-Band Modeling](https://mlsanity.com/paper/2304.08707/) (2023)
- [Automatic channel selection and spatial feature integration for multi-channel speech recognition across various array topologies](https://mlsanity.com/paper/2312.09746/) (2023)
- [ICMC-ASR: The ICASSP 2024 In-Car Multi-Channel Automatic Speech Recognition Challenge](https://mlsanity.com/paper/2401.03473/) (2024)
- [An audio-quality-based multi-strategy approach for target speaker extraction in the MISP 2023 Challenge](https://mlsanity.com/paper/2401.03697/) (2024)

## Abstract (arXiv)

Separating overlapping speech from multiple speakers is crucial for effective human-vehicle interaction. This paper proposes CabinSep, a lightweight neural mask-based minimum variance distortionless response (MVDR) speech separation approach, to reduce speech recognition errors in back-end automatic speech recognition (ASR) models. Our contributions are threefold: First, we utilize channel information to extract spatial features, which improves the estimation of speech and noise masks. Second, we employ MVDR during inference, reducing speech distortion to make it more ASR-friendly. Third, we introduce a data augmentation method combining simulated and real-recorded impulse responses (IRs), improving speaker localization at zone boundaries and further reducing speech recognition errors. With a computational complexity of only 0.4 GMACs, CabinSep achieves a 17.5% relative reduction in speech recognition error rate in a real-recorded dataset compared to the state-of-the-art DualSep model. Demos are available at: https://cabinsep.github.io/cabinsep/.