A 14uj/decision Keyword Spotting Accelerator With In-sram-computing And On Chip Learning For Customization

Abstract

Keyword spotting has gained popularity as a natural way to interact with consumer devices in recent years. However, because of its always-on nature and the variety of speech, it necessitates a low-power design as well as user customization. This paper describes a low-power, energy-efficient keyword spotting accelerator with SRAM based in-memory computing (IMC) and on-chip learning for user customization. However, IMC is constrained by macro size, limited precision, and non-ideal effects. To address the issues mentioned above, this paper proposes bias compensation and fine-tuning using an IMC-aware model design. Furthermore, because learning with low-precision edge devices results in zero error and gradient values due to quantization, this paper proposes error scaling and small gradient accumulation to achieve the same accuracy as ideal model training. The simulation results show that with user customization, we can recover the accuracy loss from 51.08% to 89.76% with compensation and f

Stats

• citations4
• S2 citations—
• github stars0
• HF likes0
• heat score5.24
• arxiv keychiang2022a

Related papers

• Boosting Keyword Spotting Through On-device Learnable User Speech Characteristics (2024)0.00
• Broadcasted Residual Learning For Efficient Keyword Spotting (2021)18.60
• Autokws: Keyword Spotting With Differentiable Architecture Search (2020)9.92
• Small-footprint Open-vocabulary Keyword Spotting With Quantized LSTM Networks (2020)0.00
• Bifocal Neural ASR: Exploiting Keyword Spotting For Inference Optimization (2021)7.50
• Few-shot Open-set Learning For On-device Customization Of Keyword Spotting Systems (2023)8.60
• Convmixer: Feature Interactive Convolution With Curriculum Learning For Small Footprint And Noisy Far-field Keyword Spotting (2022)12.61
• DCCRN-KWS: An Audio Bias Based Model For Noise Robust Small-footprint Keyword Spotting (2023)5.24