Cambridge Consultants adds new IP to Audio Weaver, pushes audio boundaries, brings decades of integrated systems expertise
This gives DSP Concepts’ customers access to decades of integrated systems expertise, providing the strategic counsel they need to build and deploy product-differentiating features at scale. In addition, Cambridge Consultants gains access to the Audio Weaver platform, where it will also add its own Voice Activity Detector (VAD) IP, in order to accelerate the development of breakthrough products and services for its clients.
Michael Maia, Vice President of Marketing at DSP Concepts, added: “Cambridge Consultants has a stellar reputation across a wide range of industry verticals, with a proven track record of helping brands produce cutting-edge product and human interface designs. We’re excited to work with Cambridge Consultants and harness their technology strategy and development expertise for product makers building on Audio Weaver, accelerating innovation and time to market for our customers.”
As a significant expansion to the engineering and innovation expertise available within the DSP Concepts ecosystem, Cambridge Consultants will extend support to help product makers build and deploy complex and disruptive new audio, signal processing and machine learning features.
With a six-decade history across a range of market segments, the company has developed a range of cutting-edge audio technologies – from the world’s first CMOS Bluetooth chip, which kickstarted the Bluetooth audio market now worth tens of billions of dollars, to Sapphyre, the digital signal processing core that underpins many of Cambridge Consultants’ ultra-low power audio applications.
Cambridge Consultants pushes audio product design beyond convention, using a ‘standards plus’ approach to create unique and industry-leading technologies, while applying novel techniques in wireless AV, sound beamforming, wake-on-word, echo cancellation, machine learning, deep learning and voice detection.
Robert Tan, Head of Signal Processing and AI at Cambridge Consultants, said: “Audio Weaver is an impressive platform for embedded audio development that allows both platform abstracted designs and platform specific optimizations. Crucially, it includes the automated real-time systems checks and monitoring for audio tuning, which is critical for mitigating production delays and accelerating time to market. The environment allows product developers to focus on the differentiating features that drive sales, which is exactly where our product and service development capability provides most value.”
In the relationship first milestone, Cambridge Consultants is making its Voice Activity Detector( VAD) available on the Audio Weaver platform, where it can be integrated into always on/always available voice-enabled products in the home and elsewhere.
Cambridge Consultants’ signal processing and Artificial Intelligence team has been working closely with DSP Concepts to demonstrate the first results of their collaboration, a custom module based on the widely deployed WebRTC VAD, a powerful and flexible VAD for offline applications. Cambridge Consultants previously used the same technology to demonstrate Ecoutez, the world’s lowest power VAD design. Ecoutez requires just 11µW – about 100th of the power consumed by a modern hearing aid.
The WebRTC VAD for Audio Weaver is available as a single custom module block that can be dragged and dropped via the intuitive interface into an audio workflow. For each input block it returns a zero or non-zero value indicating whether non-speech or speech is present. It accepts float input data of block sizes of 10, 20 or 30 ms and an audio sample rate of 8 kHz, which allows trade-offs in accuracy and responsiveness for specific applications. It also supports real time sensitivity control which makes it more resilient to background noise.
Audio Weaver already has several wake word engines optimized for Google Voice Assistant (GVA) and Alexa Voice Service (AVS), and Cambridge Consultants’ WebRTC VAD was specifically added for deployment in voice-enabled products that don’t run on a cloud service. Like wake word engines built for GVA and AVS devices, this can be combined with TalkTo™, DSP Concepts’ highly flexible audio front end that eliminates noise for high-performance voice control in noisy and unpredictable environments.