IEEE Journal of Solid-State Circuits · 2020 · 100 citations · 20 references
EngineeringComputer EngineeringComputer ScienceEmbedded SystemsLow-power DesignSystem-on-chip DesignEnergy-efficient Embedded SystemsHardware AccelerationLow-power Digital Circuit DesignMixed-signal IntegrationPower-aware DesignSpeech RecognitionAcceleratorsKeyword SpottingPower ConsumptionPower Consumption OptimizationPower-aware Hardware ArchitectureSpeaker RecognitionSystem-on-chip ArchitectureUltra-low-power Cmos
The use of speech-triggered wake-up interfaces has grown significantly in the last few years for use in ubiquitous and mobile devices. Since these interfaces must always be active, power consumption is one of their primary design metrics. This article presents a complete mixed-signal system-on-chip, capable of directly interfacing to an analog microphone and performing keyword spotting (KWS) and speaker verification (SV), without any need for further external accesses. Through the use of: 1) an integrated single-chip digital-friendly design; b) hardware-aware algorithmic optimization; and c) memoryand power-optimized accelerators, ultra-low power is achieved while maintaining high accuracy for speech recognition tasks. The 65-nm implementation achieves 18.3-μW worst case power consumption or 10.6-μW power for typical real-time scenarios, 10× below state of the art (SoA).
20
Pattern Recognition and Machine Learning
Journal of Electronic Imaging · 2007 · 22.1K citations
Front-End Factor Analysis for Speaker Verification
Najim Dehak, Patrick Kenny, Réda Dehak et al. · IEEE Transactions on Audio Speech and Language Processing · 2010 · 3.6K citations