# Towards efficient keyword spotting using spike-based time difference encoders

arXiv: https://arxiv.org/abs/2503.15402 · HTML: https://mlsanity.com/paper/2503.15402/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 78 references that are in the map.
- First posted on arXiv 19 Mar 2025.

## Builds on (most cited first, 19 of 19)

- [Surrogate Gradient Learning in Spiking Neural Networks](https://mlsanity.com/paper/1901.09948/) (2019)
- [A Survey on Neural Network Interpretability](https://mlsanity.com/paper/2012.14261/) (2020)
- [A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay](https://mlsanity.com/paper/1803.09820/) (2018)
- [The Computational Limits of Deep Learning](https://mlsanity.com/paper/2007.05558/) (2020)
- [The Heidelberg spiking datasets for the systematic evaluation of spiking neural networks](https://mlsanity.com/paper/1910.07407/) (2019)
- [Machine Learning at the Network Edge: A Survey](https://mlsanity.com/paper/1908.00080/) (2019)
- [A scalable multi-core architecture with heterogeneous memory structures for Dynamic Neuromorphic Asynchronous Processors (DYNAPs)](https://mlsanity.com/paper/1708.04198/) (2017)
- [An Analytical Estimation of Spiking Neural Networks Energy Efficiency](https://mlsanity.com/paper/2210.13107/) (2022)
- [Benchmarking Keyword Spotting Efficiency on Neuromorphic Hardware](https://mlsanity.com/paper/1812.01739/) (2018)
- [Optimizing the energy consumption of spiking neural networks for neuromorphic applications](https://mlsanity.com/paper/1912.01268/) (2019)
- [Neuromorphic electronic circuits for building autonomous cognitive systems](https://mlsanity.com/paper/1403.6428/) (2014)
- [Neuromorphic Artificial Intelligence Systems](https://mlsanity.com/paper/2205.13037/) (2022)
- [Low-power event-based face detection with asynchronous neuromorphic hardware](https://mlsanity.com/paper/2312.14261/) (2023)
- [Braille Letter Reading: A Benchmark for Spatio-Temporal Pattern Recognition on Neuromorphic Hardware](https://mlsanity.com/paper/2205.15864/) (2022)
- [Synthesizing Images from Spatio-Temporal Representations using Spike-based Backpropagation](https://mlsanity.com/paper/1906.08861/) (2019)
- [Synaptic Delays for Temporal Feature Detection in Dynamic Neuromorphic Processors](https://mlsanity.com/paper/1906.12282/) (2019)
- [Efficient spike encoding algorithms for neuromorphic speech recognition](https://mlsanity.com/paper/2207.07073/) (2022)
- [Impact of spiking neurons leakages and network recurrences on event-based spatio-temporal pattern recognition](https://mlsanity.com/paper/2211.07761/) (2022)
- [Finding the Gap: Neuromorphic Motion Vision in Cluttered Environments](https://mlsanity.com/paper/2102.08417/) (2021)

## Abstract (arXiv)

Keyword spotting in edge devices is becoming increasingly important as voice-activated assistants are widely used. However, its deployment is often limited by the extreme low-power constraints of the target embedded systems. Here, we explore the Temporal Difference Encoder (TDE) performance in keyword spotting. This recent neuron model encodes the time difference in instantaneous frequency and spike count to perform efficient keyword spotting with neuromorphic processors. We use the TIdigits dataset of spoken digits with a formant decomposition and rate-based encoding into spikes. We compare three Spiking Neural Networks (SNNs) architectures to learn and classify spatio-temporal signals. The proposed SNN architectures are made of three layers with variation in its hidden layer composed of either (1) feedforward TDE, (2) feedforward Current-Based Leaky Integrate-and-Fire (CuBa-LIF), or (3) recurrent CuBa-LIF neurons. We first show that the spike trains of the frequency-converted spoken digits have a large amount of information in the temporal domain, reinforcing the importance of better exploiting temporal encoding for such a task. We then train the three SNNs with the same number of synaptic weights to quantify and compare their performance based on the accuracy and synaptic operations. The resulting accuracy of the feedforward TDE network (89%) is higher than the feedforward CuBa-LIF network (71%) and close to the recurrent CuBa-LIF network (91%). However, the feedforward TDE-based network performs 92% fewer synaptic operations than the recurrent CuBa-LIF network with the same amount of synapses. In addition, the results of the TDE network are highly interpretable and correlated with the frequency and timescale features of the spoken keywords in the dataset. Our findings suggest that the TDE is a promising neuron model for scalable event-driven processing of spatio-temporal patterns.