Story perspectives

Breakthrough Speech Tech Boosts Efficiency with Neuromorphic Computing

9/19/2025

35 5

Summary

Story summary

Researchers created an analog speech recognition system using Resistive Neural Processing Units (RNPUs), improving efficiency at room temperature.
This technology minimizes complex digital preprocessing, allowing ultra-low-power acoustic processing for voice-command devices.
RNPU preprocessing achieves over 90% classification accuracy, surpassing traditional digital methods.
The UK is launching the Neuroware Innovation and Knowledge Centre with £12.8 million funding to advance neuromorphic computing.
The centre will focus on neuromorphic photonics for faster, energy-efficient computation, collaborating with top UK universities and tech firms.

1 of 1

Story summary

Researchers created an analog speech recognition system using Resistive Neural Processing Units (RNPUs), improving efficiency at room temperature.
This technology minimizes complex digital preprocessing, allowing ultra-low-power acoustic processing for voice-command devices.
RNPU preprocessing achieves over 90% classification accuracy, surpassing traditional digital methods.
The UK is launching the Neuroware Innovation and Knowledge Centre with £12.8 million funding to advance neuromorphic computing.
The centre will focus on neuromorphic photonics for faster, energy-efficient computation, collaborating with top UK universities and tech firms.