
UK microcontroller maker Xmos has launched a complete voice AI system built on xcore platform hardware, software and tools.
The xcore-voice platform voice pipeline example designs using the Xmos audio front-end, incorporating far-field processing and support for third party ISV voice algorithms.
This frees up the voice processing workload from the system on chip host, behaving like a voice processor that enables product designers to deliver voice interfaces quickly and cost effectively, while providing optimal audio quality. Turnkey example designs within the xcore-voice solution leverage the processor capabilities through the xcore platform software, providing the building blocks for a wide range of applications.
The third generation xcore.ai architecture developed in Bristol runs the company’s far-field voice processing algorithms in a cost-effective, customisable package. This provides a natural voice interface that is capable of delivering local command recognition – minimising the need for the cloud that not only enables better user experience through lower latency and enhanced privacy but also further saving cost and power requirements in product designs.
The xcore-voice’s low-power architecture minimises power consumption. By offering a true standby mode, xcore-voice delivers the ‘always-on’ functionality required in many modern electronics at <55mW, and in line with EU energy-saving regulations.
The far-field algorithm enables a flexible voice processing workload in a package that allows for customisation and feature upgrades. Application-specific software components such as tunable algorithms, and core application software components such as FreeRTOS, are also available.
The xcore-voice combination of general purpose DSP, AI, and IO processing can be customised in software – drastically reducing the time-to-market for custom silicon. Integration pathways have been simplified for designers while still offering a complete design for turnkey integration.
The platform can be deployed in a range of home electronics, including appliances, smart TVs and soundbars, smart lighting and power switches, fitness machines, consumer toys, and smart home gateways.
“As consumers, we’re more and more acclimatised to the use of voice technologies in our everyday lives – and that makes us more aware of their limitations,” said Aneet Chopra, EVP Marketing and Product Management at Xmos. “In order to avoid falling behind rival products, product designers need to be able to deliver ‘across-the-room’ voice interfaces that offer the best possible quality, without compromising on speed-to-market or cost.
“xcore-voice is purpose-built to tackle that challenge. The level of customisation we’re delivering in such a cost-effective and power-friendly package accelerates development without cutting corners. It means we can offer designers the capacity to experiment with and reiterate their designs at pace – enabling them to easily respond to a rapidly evolving market. All of this while maintaining the privacy of the end user”
xcore-voice will be generally available from 15 April 2023. The voice evaluation kit and explorer board evaluation kit – the XK-VOICE-L71 and XK-EVK-XU316, respectively are available now via Digi-Key.
