# Iterative autoregression: a novel trick to improve your low-latency speech enhancement model

arXiv: https://arxiv.org/abs/2211.01751 · HTML: https://mlsanity.com/paper/2211.01751/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 29 references that are in the map.
- First posted on arXiv 3 Nov 2022.

## Builds on (most cited first, 20 of 20)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [WaveNet: A Generative Model for Raw Audio](https://mlsanity.com/paper/1609.03499/) (2016)
- [Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks](https://mlsanity.com/paper/1506.03099/) (2015)
- [SDR - half-baked or well done?](https://mlsanity.com/paper/1811.02508/) (2018)
- [Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation](https://mlsanity.com/paper/1809.07454/) (2018)
- [UTMOS: UTokyo-SaruLab System for VoiceMOS Challenge 2022](https://mlsanity.com/paper/2204.02152/) (2022)
- [Efficient Neural Audio Synthesis](https://mlsanity.com/paper/1802.08435/) (2018)
- [Professor Forcing: A New Algorithm for Training Recurrent Networks](https://mlsanity.com/paper/1610.09038/) (2016)
- [Real Time Speech Enhancement in the Waveform Domain](https://mlsanity.com/paper/2006.12847/) (2020)
- [DNSMOS P.835: A Non-Intrusive Perceptual Objective Speech Quality Metric to Evaluate Noise Suppressors](https://mlsanity.com/paper/2110.01763/) (2021)
- [Wave-U-Net: A Multi-Scale Neural Network for End-to-End Audio Source Separation](https://mlsanity.com/paper/1806.03185/) (2018)
- [FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement](https://mlsanity.com/paper/2010.15508/) (2020)
- [HiFi++: a Unified Framework for Bandwidth Extension and Speech Enhancement](https://mlsanity.com/paper/2203.13086/) (2022)
- [Chunked Autoregressive GAN for Conditional Waveform Synthesis](https://mlsanity.com/paper/2110.10139/) (2021)
- [Bunched LPCNet : Vocoder for Low-cost Neural Text-To-Speech Systems](https://mlsanity.com/paper/2008.04574/) (2020)
- [STFT-Domain Neural Speech Enhancement with Very Low Algorithmic Latency](https://mlsanity.com/paper/2204.09911/) (2022)
- [Fast Wavenet Generation Algorithm](https://mlsanity.com/paper/1611.09482/) (2016)
- [FFC-SE: Fast Fourier Convolution for Speech Enhancement](https://mlsanity.com/paper/2204.03042/) (2022)
- [Deep neural network Based Low-latency Speech Separation with Asymmetric analysis-Synthesis Window Pair](https://mlsanity.com/paper/2106.11794/) (2021)
- [Speech Quality Assessment in Crowdsourcing: Comparison Category Rating Method](https://mlsanity.com/paper/2104.04371/) (2021)

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

- [Self-Steering Deep Non-Linear Spatially Selective Filters for Efficient Extraction of Moving Speakers under Weak Guidance](https://mlsanity.com/paper/2507.02791/) (2025)
- [Adaptive Rotary Steering with Joint Autoregression for Robust Extraction of Closely Moving Speakers in Dynamic Scenarios](https://mlsanity.com/paper/2601.12345/) (2026)
- [Speech Boosting: Low-Latency Live Speech Enhancement for TWS Earbuds](https://mlsanity.com/paper/2409.18705/) (2024)
- [Autoregressive Guidance of Deep Spatially Selective Filters using Bayesian Tracking for Efficient Extraction of Moving Speakers](https://mlsanity.com/paper/2603.23723/) (2026)
- [MoMuSE: Momentum Multi-modal Target Speaker Extraction for Real-time Scenarios with Impaired Visual Cues](https://mlsanity.com/paper/2412.08247/) (2024)

## Abstract (arXiv)

Streaming models are an essential component of real-time speech enhancement tools. The streaming regime constrains speech enhancement models to use only a tiny context of future information. As a result, the low-latency streaming setup is generally considered a challenging task and has a significant negative impact on the model's quality. However, the sequential nature of streaming generation offers a natural possibility for autoregression, that is, utilizing previous predictions while making current ones. The conventional method for training autoregressive models is teacher forcing, but its primary drawback lies in the training-inference mismatch that can lead to a substantial degradation in quality. In this study, we propose a straightforward yet effective alternative technique for training autoregressive low-latency speech enhancement models. We demonstrate that the proposed approach leads to stable improvement across diverse architectures and training scenarios.