Redefining possibilities within the semiconductor and audio design sectors.
By Aneet Chopra, EVP Marketing & Product Management at XMOS.
The past decade has seen rapid evolution in the capabilities and sophistication of the electronic devices that we increasingly take for granted. Artificial intelligence (AI) and the internet of things (IoT) have been at the forefront of this revolution, unlocking a new era of innovation for device manufacturers.
But what lies at the heart of every ‘AI-powered’ or ‘smart’ product? The answer is semiconductors. Not only have the capabilities of the humble semiconductor been pushed to their limits by manufacturers eager to innovate and leverage AI or IoT-powered technologies; these tech trends have also upended the traditional business model of the semiconductor sector.
Previously, manufacturers could confidently make major investments in a single chip design or solution, safe in the knowledge that there would be a suitably large end market for that device. Consumer- facing brands were producing a relatively limited number of high-volume products, so you could afford to design a chip with a relatively singular focus.
However, with the rise of AI and the IoT, the focus has shifted. Manufacturers want to be able to target increasingly small niches, and conventional markets are constantly under threat from both the pace of technological development, and the emergence of challenger technologies looking to outperform or outthink them.
As a result, semiconductor companies can’t throw huge investments behind monolithic platforms that might be perfect for a camera, but not so good for a smartphone. Instead, there is a significant increase in demand from manufacturers for new ‘one-chip-fits-all’ solutions that can address multiple markets simultaneously.
Given that state of play, flexibility has never been more important. Versatile components that can do many things and that can be changed and altered and finessed without physically removing the chips is crucial.
This shift presents a significant opportunity for semiconductor designers to innovate—but also some significant challenges. Developing solutions for a wide range of hardware requirements demands considerable expertise and a different mindset to the ‘traditional’ approach of the semiconductor industry.
XMOS excellence
So, what’s the solution? For XMOS, it is called xcore.ai. This is a revolutionary chip that offers versatility and flexibility to device manufacturers to efficiently configure a multitude of crucial parameters such as AI capabilities, digital signal processing (DSP) functionality, control, and input/output (I/ O)—all from one small device. What makes xcore.ai truly special, though, is that its unique selling point (USP) doesn’t lie within the physical chip, but rather within the ability to be programmed entirely through software. This allows for complete transformation of a device’s functionality without the need to physically replace or remove the semiconductor from within the device.
The ability to adapt the semiconductor design on the fly and having access to countless combinations and pairings with other elements that are also in the device is something that cannot be ignored. This is big.
Enter DSP Concepts
Emphasising just how flexible xcore.ai can be as a platform, XMOS has recently partnered with DSP Concepts. The announcement shines a light on DSP Concept’s decision to integrate xcore.ai with its flagship software, Audio Weaver.
Audio Weaver is a platform for engineers and designers to graphically design and debug both audio and voice solutions. The primary use case of Audio Weaver is constructing entire audio processing pipelines while almost eliminating the need to dabble with complex manual coding, lowering the barrier to entry for people wanting to enter the market and making it faster, easier, and more convenient for engineers to craft the solution they are aiming for.
The Audio Weaver platform does this by offering a user interface (UI) in which the coding of the device is performed through a visual drag and drop interface while producing a highly optimised target run-time library and making all the advanced debugging features available during the product development phase. Doing this not only makes the design phase faster andmore efficient, it also greatly minimises user error, user exhaustion, and takes care of compatibility issues as and when they arrive.
So why is the Audio Weaver platform such a perfect accompaniment for xcore.ai? Well, let’s remind ourselves of the fact that xcore.ai offers a fully adaptable combination of AI, DSP, control, and I/O. It is the ‘all-in-one-chip’ that design engineers are looking for, and its capabilities perfectly match what audio engineers require.
Part of this requirement is the capacity to ‘parallelise’ workloads—that is to say, split a large task up into smaller parts to be executed at the same time by multiple cores. Audio Weaver has parallelisation built-in, which dovetails beautifully with xcore’s multicore architecture. Engineers are thus empowered to scale the capabilities of DSP according to the end product’s requirements, blending it with the need for AI, control, and/ or I/O without compromising on performance.
That process is accelerated by Audio Weaver’s powerful, simple drag-and-drop design UI. Being able to revise your designs through a graphical interface offers an extremely fast time-to-market compared to manual coding, at a reasonable cost, and with the flexibility to respond to changing demands and new opportunities.
By integrating XMOS’s xcore.ai with DSP Concepts’ Audio Weaver platform, both companies are ready to redefine the possibilities within the semiconductor and audio design sectors. As industries continue to seek innovative solutions that combine efficiency with versatility, partnerships like this will likely become a blueprint for future technological advancements, driving further innovation and setting new standards for what is achievable.