# HALO: Half-Frame-Rate Adaptive Learnable Operator for Lightweight STFT-Based Speech Enhancement

arXiv: https://arxiv.org/abs/2606.12328 · HTML: https://mlsanity.com/paper/2606.12328/

## Key facts

- No citing papers in the map yet (data through 19 Oct 2026).
- Placed in Speech enhancement and noise suppression, one of 11 directions in Speech and audio language models.
- Builds on 22 of its 30 references that are in the map.
- First posted on arXiv 10 Jun 2026.

## Placed in

- Speech and audio language models / [Speech enhancement and noise suppression](https://mlsanity.com/speech/speech-enhancement-and-noise-suppression/)

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

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [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)
- [Dynamic Convolution: Attention over Convolution Kernels](https://mlsanity.com/paper/1912.03458/) (2019)
- [DNSMOS P.835: A Non-Intrusive Perceptual Objective Speech Quality Metric to Evaluate Noise Suppressors](https://mlsanity.com/paper/2110.01763/) (2021)
- [DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancement](https://mlsanity.com/paper/2008.00264/) (2020)
- [CondConv: Conditionally Parameterized Convolutions for Efficient Inference](https://mlsanity.com/paper/1904.04971/) (2019)
- [Dual-path RNN: efficient long sequence modeling for time-domain single-channel speech separation](https://mlsanity.com/paper/1910.06379/) (2019)
- [TF-GridNet: Integrating Full- and Sub-Band Modeling for Speech Separation](https://mlsanity.com/paper/2211.12433/) (2022)
- [Interspeech 2021 Deep Noise Suppression Challenge](https://mlsanity.com/paper/2101.01902/) (2021)
- [TSTNN: Two-stage Transformer based Neural Network for Speech Enhancement in the Time Domain](https://mlsanity.com/paper/2103.09963/) (2021)
- [DiDiSpeech: A Large Scale Mandarin Speech Corpus](https://mlsanity.com/paper/2010.09275/) (2020)
- [DeepFilterNet2: Towards Real-Time Speech Enhancement on Embedded Devices for Full-Band Audio](https://mlsanity.com/paper/2205.05474/) (2022)
- [DPCRN: Dual-Path Convolution Recurrent Network for Single Channel Speech Enhancement](https://mlsanity.com/paper/2107.05429/) (2021)
- [STFT-Domain Neural Speech Enhancement with Very Low Algorithmic Latency](https://mlsanity.com/paper/2204.09911/) (2022)
- [LiSenNet: Lightweight Sub-band and Dual-Path Modeling for Real-Time Speech Enhancement](https://mlsanity.com/paper/2409.13285/) (2024)
- [Beyond Performance Plateaus: A Comprehensive Study on Scalability in Speech Enhancement](https://mlsanity.com/paper/2406.04269/) (2024)
- [Adaptive Convolutions with Per-pixel Dynamic Filter Atom](https://mlsanity.com/paper/2108.07895/) (2021)
- [Improving Frame-Online Neural Speech Enhancement with Overlapped-Frame Prediction](https://mlsanity.com/paper/2204.07566/) (2022)
- [UL-UNAS: Ultra-Lightweight U-Nets for Real-Time Speech Enhancement via Network Architecture Search](https://mlsanity.com/paper/2503.00340/) (2025)
- [Causal Signal-Based DCCRN with Overlapped-Frame Prediction for Online Speech Enhancement](https://mlsanity.com/paper/2309.03684/) (2023)
- [Adaptive Convolution for CNN-based Speech Enhancement Models](https://mlsanity.com/paper/2502.14224/) (2025)

## Abstract (arXiv)

STFT-based speech enhancement typically adopts overlapping analysis frames. While overlap is essential for stable STFT processing, it makes adjacent frames highly correlated, causing redundant computation in lightweight models. We propose Half-frame-rate Adaptive Learnable Operator (HALO), a causal plug-in module that halves the internal frame rate without altering the STFT procedure. Broadly applicable to many lightweight models, HALO applies adaptive rate reduction before the backbone and restoration afterward, reconstructing the full-rate spectrum on the original STFT grid. Both reduction and restoration are implemented with lightweight dynamic convolutions. By halving the processed frame rate, HALO reduces backbone compute cost with no added algorithmic latency, freeing budget for channel widening. Experiments on the DNS3 dataset show consistent gains across diverse lightweight models under matched complexity, demonstrating the effectiveness of reducing overlap-induced redundancy.