mlsanityData to 19 Oct 2026

Lyric: Wave-Domain Computing for Efficient Spoken-Digit Recognition

arXiv 2610.06433 ↗First posted 5 Oct 2026eess.AScited by 0

On the map: Speech and audio language models → Speech recognition models and training also fits: Speaker recognition and diarization

Abstract (arXiv)

Low-power speech recognition requires reducing audio acquisition and feature-processing costs as well as neural inference. We present Lyric, a speech-recognition front end that uses wave-domain computing to extract features before digitization. Our proposed system uses passive acoustic resonators to separate speech by frequency and supplies their slowly varying envelopes to a compact temporal neural network. This moves spectral filtering into the acoustic structure, reducing the data and processing needed for recognition. We build a prototype and evaluate spoken-digit recognition on a speakerdisjoint AudioMNIST split across nine classifier families. The front end acquires 32 times fewer scalar samples than a 16-kHz waveform. In the EdgeSpeechNet-A comparison on a Raspberry Pi 4, we reduce preprocessing-plus-inference latency and estimated processor energy per inference by 98.6%, with a 3.58-percentage-point decrease in accuracy to 95.70%.

Key facts. No citing papers in the map yet (data through 19 Oct 2026). Placed in Speech recognition models and training, one of 11 directions in Speech and audio language models. Builds on 12 of its 23 references that are in the map. First posted on arXiv 5 Oct 2026.

Builds on

12 of 23 references.

References · 12

  1. 2015Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Codingcited by 2,364
  2. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,620
  3. 2021MLPerf Tiny Benchmarkcited by 126
  4. 2018AudioMNIST: Exploring Explainable Artificial Intelligence for Audio Analysis on a Simple Benchmarkcited by 76
  5. 2019Temporal Convolution for Real-time Keyword Spotting on Mobile Devicescited by 55
  6. 2021Broadcasted Residual Learning for Efficient Keyword Spottingcited by 53
  7. 2019Wave Physics as an Analog Recurrent Neural Networkcited by 44
  8. 2020MatchboxNet: 1D Time-Channel Separable Convolutional Neural Network Architecture for Speech Commands Recognitioncited by 40
  9. 2018EdgeSpeechNets: Highly Efficient Deep Neural Networks for Speech Recognition on the Edgecited by 7
  10. 2023Physics-inspired Neuroacoustic Computing Based on Tunable Nonlinear Multiple-scatteringcited by 5
  11. 2024In-Materia Speech Recognitioncited by 3
  12. 2023How Tiny Can Analog Filterbank Features Be Made for Ultra-low-power On-device Keyword Spotting?cited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2021Exploiting Beam Search Confidence for Energy-Efficient Speech Recognition
  2. 2023Deep Photonic Reservoir Computer for Speech Recognition
  3. 2025A 71.2-μW Speech Recognition Accelerator with Recurrent Spiking Neural Network
  4. 2022ASRPU: A Programmable Accelerator for Low-Power Automatic Speech Recognition
  5. 2026O'ido: Open-Vocabulary Speech Recognition on a 5 Microcontroller without a Neural Accelerator
  6. 2021Bifocal Neural ASR: Exploiting Keyword Spotting for Inference Optimization
  7. 2022SpeechNet: Weakly Supervised, End-to-End Speech Recognition at Industrial Scale
  8. 2026Bridging Biological Hearing and Neuromorphic Computing: End-to-End Time-Domain Audio Signal Processing with Reservoir Computing