# Configurable-Bandwidth Time-Frequency Modeling for Efficient Full-Band Speech Enhancement Across Sampling Rates

arXiv: https://arxiv.org/abs/2609.29463 · HTML: https://mlsanity.com/paper/2609.29463/

## 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 25 of its 35 references that are in the map.
- First posted on arXiv 24 Sep 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, 25 of 25)

- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [SDR - half-baked or well done?](https://mlsanity.com/paper/1811.02508/) (2018)
- [FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement](https://mlsanity.com/paper/2010.15508/) (2020)
- [TF-GridNet: Integrating Full- and Sub-Band Modeling for Speech Separation](https://mlsanity.com/paper/2211.12433/) (2022)
- [ICASSP 2022 Deep Noise Suppression Challenge](https://mlsanity.com/paper/2202.13288/) (2022)
- [Siamese Masked Autoencoders](https://mlsanity.com/paper/2305.14344/) (2023)
- [A Hybrid DSP/Deep Learning Approach to Real-Time Full-Band Speech Enhancement](https://mlsanity.com/paper/1709.08243/) (2017)
- [DeepFilterNet: A Low Complexity Speech Enhancement Framework for Full-Band Audio based on Deep Filtering](https://mlsanity.com/paper/2110.05588/) (2021)
- [DPT-FSNet: Dual-path Transformer Based Full-band and Sub-band Fusion Network for Speech Enhancement](https://mlsanity.com/paper/2104.13002/) (2021)
- [A Perceptually-Motivated Approach for Low-Complexity, Real-Time Enhancement of Fullband Speech](https://mlsanity.com/paper/2008.04259/) (2020)
- [DeepFilterNet2: Towards Real-Time Speech Enhancement on Embedded Devices for Full-Band Audio](https://mlsanity.com/paper/2205.05474/) (2022)
- [Toward Universal Speech Enhancement for Diverse Input Conditions](https://mlsanity.com/paper/2309.17384/) (2023)
- [Deep Filtering: Signal Extraction and Reconstruction Using Complex Time-Frequency Filters](https://mlsanity.com/paper/1904.08369/) (2019)
- [TF-Locoformer: Transformer with Local Modeling by Convolution for Speech Separation and Enhancement](https://mlsanity.com/paper/2408.03440/) (2024)
- [Separate and Reconstruct: Asymmetric Encoder-Decoder for Speech Separation](https://mlsanity.com/paper/2406.05983/) (2024)
- [Improving Design of Input Condition Invariant Speech Enhancement](https://mlsanity.com/paper/2401.14271/) (2024)
- [Sampling Frequency Independent Dialogue Separation](https://mlsanity.com/paper/2206.02124/) (2022)
- [UL-UNAS: Ultra-Lightweight U-Nets for Real-Time Speech Enhancement via Network Architecture Search](https://mlsanity.com/paper/2503.00340/) (2025)
- [FastEnhancer: Speed-Optimized Streaming Neural Speech Enhancement](https://mlsanity.com/paper/2509.21867/) (2025)
- [DMF-Net: A decoupling-style multi-band fusion model for full-band speech enhancement](https://mlsanity.com/paper/2203.00472/) (2022)
- [Query-Based Asymmetric Modeling with Decoupled Input-Output Rates for Speech Restoration](https://mlsanity.com/paper/2509.21003/) (2025)
- [Deep Filter Estimation from Inter-Frame Correlations for Monaural Speech Dereverberation](https://mlsanity.com/paper/2603.14986/) (2026)
- [Asymmetric Encoder-Decoder Based on Time-Frequency Correlation for Speech Separation](https://mlsanity.com/paper/2603.29097/) (2026)
- [One Model, Many Latencies: Universal Speech Enhancement for Diverse Real-Time Applications](https://mlsanity.com/paper/2606.25621/) (2026)

## Abstract (arXiv)

Speech enhancement systems are often developed for a fixed sampling rate, while time-frequency models become more expensive as the number of frequency bins increases. We propose TF-Refiner, a sampling-frequency-independent model that decouples the deep analysis bandwidth from the full-band input and output. A deep encoder processes the band below a configurable cutoff, while a shallow decoder combines the encoded features with input-dependent high-band queries and predicts local complex filters applied to the original noisy STFT. A single parameter set trained at 16 and 48 kHz is evaluated at various sampling rates. On VoiceBank+DEMAND, the universal model outperforms the rate-specific counterparts in PESQ, STOI, and log-spectral distance across the evaluated rates, including rates unseen in training. Random-cutoff training enables inference-time selection of cost-quality operating points without retraining or changing the output bandwidth. These results support configurable analysis bandwidth as a practical design choice for multi-rate full-band enhancement.