# From Diet to Free Lunch: Estimating Auxiliary Signal Properties using Dynamic Pruning Masks in Speech Enhancement Networks

arXiv: https://arxiv.org/abs/2602.10666 · HTML: https://mlsanity.com/paper/2602.10666/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 36 references that are in the map.
- First posted on arXiv 11 Feb 2026.

## Builds on (most cited first, 23 of 23)

- [SDR - half-baked or well done?](https://mlsanity.com/paper/1811.02508/) (2018)
- [Dynamic Neural Networks: A Survey](https://mlsanity.com/paper/2102.04906/) (2021)
- [FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement](https://mlsanity.com/paper/2010.15508/) (2020)
- [TorchAudio-Squim: Reference-less Speech Quality and Intelligibility measures in TorchAudio](https://mlsanity.com/paper/2304.01448/) (2023)
- [Deep Learning-based Non-Intrusive Multi-Objective Speech Assessment Model with Cross-Domain Features](https://mlsanity.com/paper/2111.02363/) (2021)
- [DeepFilterNet: Perceptually Motivated Real-Time Speech Enhancement](https://mlsanity.com/paper/2305.08227/) (2023)
- [A Perceptually-Motivated Approach for Low-Complexity, Real-Time Enhancement of Fullband Speech](https://mlsanity.com/paper/2008.04259/) (2020)
- [DCCRN+: Channel-wise Subband DCCRN with SNR Estimation for Speech Enhancement](https://mlsanity.com/paper/2106.08672/) (2021)
- [Sparse Mixture of Local Experts for Efficient Speech Enhancement](https://mlsanity.com/paper/2005.08128/) (2020)
- [Speech Enhancement with Zero-Shot Model Selection](https://mlsanity.com/paper/2012.09359/) (2020)
- [Dynamic Gated Recurrent Neural Network for Compute-efficient Speech Enhancement](https://mlsanity.com/paper/2408.12425/) (2024)
- [Scalable Speech Enhancement with Dynamic Channel Pruning](https://mlsanity.com/paper/2412.17121/) (2024)
- [Adaptive Slimming for Scalable and Efficient Speech Enhancement](https://mlsanity.com/paper/2507.04879/) (2025)
- [Investigating the Sensitivity of Pre-trained Audio Embeddings to Common Effects](https://mlsanity.com/paper/2501.15900/) (2025)
- [Dynamic Slimmable Networks for Efficient Speech Separation](https://mlsanity.com/paper/2507.06179/) (2025)
- [Understanding Locally Competitive Networks](https://mlsanity.com/paper/1410.1165/) (2014)
- [Speech enhancement aided end-to-end multi-task learning for voice activity detection](https://mlsanity.com/paper/2010.12484/) (2020)
- [PercepNet+: A Phase and SNR Aware PercepNet for Real-Time Speech Enhancement](https://mlsanity.com/paper/2203.02263/) (2022)
- [Resource-Efficient Speech Quality Prediction through Quantization Aware Training and Binary Activation Maps](https://mlsanity.com/paper/2407.04578/) (2024)
- [Knowing When to Quit: Probabilistic Early Exits for Speech Separation](https://mlsanity.com/paper/2507.09768/) (2025)
- [VSANet: Real-time Speech Enhancement Based on Voice Activity Detection and Causal Spatial Attention](https://mlsanity.com/paper/2310.07295/) (2023)
- [Impairments are Clustered in Latents of Deep Neural Network-based Speech Quality Models](https://mlsanity.com/paper/2504.21528/) (2025)
- [TGIF: Talker Group-Informed Familiarization of Target Speaker Extraction](https://mlsanity.com/paper/2507.14044/) (2025)

## Built on by (most cited first, 2 of 2)

- [Beyond Model Size: Redesigning LiSenNet for embedded speech enhancement](https://mlsanity.com/paper/2609.29866/) (2026)
- [Does per-frame early exit pay? A compute-matched study of dynamic depth for on-device speech enhancement](https://mlsanity.com/paper/2609.29867/) (2026)

## Abstract (arXiv)

Speech Enhancement (SE) in audio devices is often supported by auxiliary modules for Voice Activity Detection (VAD), SNR estimation, or Acoustic Scene Classification to ensure robust context-aware behavior and seamless user experience. Just like SE, these tasks often employ deep learning; however, deploying additional models on-device is computationally impractical, whereas cloud-based inference would introduce additional latency and compromise privacy. Prior work on SE employed Dynamic Channel Pruning (DynCP) to reduce computation by adaptively disabling specific channels based on the current input. In this work, we investigate whether useful signal properties can be estimated from these internal pruning masks, thus removing the need for separate models. We show that simple, interpretable predictors achieve up to 93% accuracy on VAD, 84% on noise classification, and an R2 of 0.86 on F0 estimation. With binary masks, predictions reduce to weighted sums, inducing negligible overhead. Our contribution is twofold: on one hand, we examine the emergent behavior of DynCP models through the lens of downstream prediction tasks, to reveal what they are learning; on the other, we repurpose and re-propose DynCP as a holistic solution for efficient SE and simultaneous estimation of signal properties.