Keywords
audio signal encoding
speech recognition
multilayer perceptron
surrogate gradient
How to Cite
Abstract
we compared three computationally simple methods for encoding short speech recordings as spike sequences for real-time classification with a multilayer spiking neural network. The methods included integral coding of spectral features, stream spectral encoding, and direct encoding of local signal dynamics. All three methods used the same network of Leaky Integrate-and-Fire neurons trained by surrogate-gradient backpropagation. Experiments on the Google Speech Commands v0.01 (GSC) and Heidelberg Digits (HD) datasets showed that integral rate coding achieved the highest accuracy on GSC (83.1%), whereas all three methods performed comparably on HD. Stream spectral encoding required only four simulation time steps per spectrogram frame, substantially reducing response latency. Speech2Spikes achieved 80.1% accuracy on GSC, only 3% below integral rate coding, while using a more compact input representation. The results provide a basis for selecting spike encoding methods under different computational and latency constraints.
Acknowledgements
References
Neftci E. O., Mostafa H., Zenke F. Surrogate Gradient Learning in Spiking Neural Networks: Bringing the Power of Gradient-Based Optimization to Spiking Neural Networks. IEEE Signal Processing Magazine. 2019;36(6):51–63.
Bellec G., Salaj D., Subramoney A., Legenstein R., Maass W. Spike-Frequency Adaptation Provides a Long Short-Term Memory to Networks of Spiking Neurons. Nature Communications. 2020;11:3690.
Auge D., Hille J., Mueller E., Knoll A. A Survey of Encoding Techniques for Signal Processing in Spiking Neural Networks. Neural Networks. 2021;142:273–301.
Cramer B., Stradmann Y., Schemmel J., Zenke F. The Heidelberg Spiking Data Sets for the Systematic Evaluation of Spiking Neural Networks. IEEE Transactions on Neural Networks and Learning Systems. 2022;33(7):2744–2757.
Rybka R. B., Dyakova E. O., Serenko A. V., Sboev A. G. Spiking Neural Network with Quantized Weights for Audio Stream Classification. Studies in Computational Intelligence. 2025;1179:89–97.
Matsukatova A. N., Prudnikov N. V., Kulagin V. A. et al. Combination of Organic-Based Reservoir Computing and Spiking Neuromorphic Systems for a Robust and Efficient Pattern Classification. Adv. Intell. Syst. 2023;5:2200407.
Warden P. Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition. 2018. arXiv:1804.03209.
Bittar A., Garner P. N. A Surrogate Gradient Spiking Baseline for Speech Command Recognition. Frontiers in Neuroscience. 2022;16:865897.
Wang J., Yu L., Huang L., Zhou C., Zhang H., Song Z., Zhang M., Ma Z., Zhang Z. Efficient Speech Command Recognition Leveraging Spiking Neural Network and Curriculum LearningBased Knowledge Distillation. 2024. arXiv:2412.12858.
Zhang S., Yang Q., Ma C., Wu J., Li H., Tan K. C. TC-LIF: A Two-Compartment Spiking Neuron Model for Long-Term Sequential Modelling. Proceedings of the AAAI Conference on Artificial Intelligence. 2024.
Stewart K. M., Shea T., Pacik-Nelson N., Gallo E., Danielescu A. Speech2Spikes: Efficient Audio Encoding Pipeline for Real-Time Neuromorphic Systems. Proceedings of the 2023 Annual Neuro-Inspired Computational Elements Conference (NICE). 2023:1–7. DOI: 10.1145/3584954.3584995.
Yin B., Corradi F., Bohté S. M. Accurate and Efficient Time-Domain Classification With Adaptive Spiking Recurrent Neural Networks. Nature Machine Intelligence. 2021;3:905–913.
Hammouamri I., Khalfaoui-Hassani I., Masquelier T. Learning Delays in Spiking Neural Networks Using Dilated Convolutions with Learnable Spacings. The Twelfth International Conference on Learning Representations (ICLR). 2024.
This work is licensed under a Creative Commons Attribution 4.0 International License.