# Dual Knowledge Distillation for Efficient Sound Event Detection

arXiv: https://arxiv.org/abs/2402.02781 · HTML: https://mlsanity.com/paper/2402.02781/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Speech and audio distillation, one of 12 directions in Knowledge distillation.
- Builds on 16 of its 32 references that are in the map.
- First posted on arXiv 5 Feb 2024.

## Placed in

- Knowledge distillation / [Speech and audio distillation](https://mlsanity.com/distill/speech-and-audio-distillation/)

## Builds on (most cited first, 16 of 16)

- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift](https://mlsanity.com/paper/1502.03167/) (2015)
- [Squeeze-and-Excitation Networks](https://mlsanity.com/paper/1709.01507/) (2017)
- [BEATs: Audio Pre-Training with Acoustic Tokenizers](https://mlsanity.com/paper/2212.09058/) (2022)
- [Sound Event Detection: A Tutorial](https://mlsanity.com/paper/2107.05463/) (2021)
- [Convolutional Recurrent Neural Networks for Polyphonic Sound Event Detection](https://mlsanity.com/paper/1702.06286/) (2017)
- [A Framework for the Robust Evaluation of Sound Event Detection](https://mlsanity.com/paper/1910.08440/) (2019)
- [Frequency Dynamic Convolution: Frequency-Adaptive Pattern Recognition for Sound Event Detection](https://mlsanity.com/paper/2203.15296/) (2022)
- [Low-complexity acoustic scene classification in DCASE 2022 Challenge](https://mlsanity.com/paper/2206.03835/) (2022)
- [Description and analysis of novelties introduced in DCASE Task 4 2022 on the baseline system](https://mlsanity.com/paper/2210.07856/) (2022)
- [A Multi-Task Learning Framework for Sound Event Detection using High-level Acoustic Characteristics of Sounds](https://mlsanity.com/paper/2305.10729/) (2023)
- [Impact of Sound Duration and Inactive Frames on Sound Event Detection Performance](https://mlsanity.com/paper/2102.01927/) (2021)
- [The impact of non-target events in synthetic soundscapes for sound event detection](https://mlsanity.com/paper/2109.14061/) (2021)
- [Leveraging Audio-Tagging Assisted Sound Event Detection using Weakified Strong Labels and Frequency Dynamic Convolutions](https://mlsanity.com/paper/2304.12688/) (2023)
- [Frequency & Channel Attention for Computationally Efficient Sound Event Detection](https://mlsanity.com/paper/2306.11277/) (2023)
- [Post-Processing Independent Evaluation of Sound Event Detection Systems](https://mlsanity.com/paper/2306.15440/) (2023)

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

- [TF-Mamba: A Time-Frequency Network for Sound Source Localization](https://mlsanity.com/paper/2409.05034/) (2024)
- [Advancing Airport Tower Command Recognition: Integrating Squeeze-and-Excitation and Broadcasted Residual Learning](https://mlsanity.com/paper/2406.18313/) (2024)

## Abstract (arXiv)

Sound event detection (SED) is essential for recognizing specific sounds and their temporal locations within acoustic signals. This becomes challenging particularly for on-device applications, where computational resources are limited. To address this issue, we introduce a novel framework referred to as dual knowledge distillation for developing efficient SED systems in this work. Our proposed dual knowledge distillation commences with temporal-averaging knowledge distillation (TAKD), utilizing a mean student model derived from the temporal averaging of the student model's parameters. This allows the student model to indirectly learn from a pre-trained teacher model, ensuring a stable knowledge distillation. Subsequently, we introduce embedding-enhanced feature distillation (EEFD), which involves incorporating an embedding distillation layer within the student model to bolster contextual learning. On DCASE 2023 Task 4A public evaluation dataset, our proposed SED system with dual knowledge distillation having merely one-third of the baseline model's parameters, demonstrates superior performance in terms of PSDS1 and PSDS2. This highlights the importance of proposed dual knowledge distillation for compact SED systems, which can be ideal for edge devices.