In the digital audio device industry, designing algorithms that enable processing often requires considerable engineering effort. This has given rise to a cottage industry of startups, including DSP Concepts, that collaborate with algorithm providers to reduce audio development and go-to-market time. Five years after its founding, DSP today announced that it has secured fresh capital to expand its customer base and hire employees across its offices in Santa Clara, California; Boston, Massachusetts; Stuttgart, Germany; and soon Taiwan. (It currently has 48 employees, 38 of whom are engineers.)
Taiwania Capital led the $14.5 million series B round, with participation from Sony Innovation Fund, Porsche Ventures, BMW iVentures, Arm IoT Fund, and more, bringing DSP’s total raised to over $25 million. CTO Paul Beckmann said the infusion follows a year in which the number of audio devices customers have deployed eclipsed 40 million.
“We like to think of Audio Weaver as building blocks for audio: We provide over 400 of these audio ‘Legos’ that include modules to support everything from hyper-efficient [internet of things] devices to ultra-premium automotive systems,” said Beckmann, who led R&D engineering at Bose prior to cofounding DSP with Chin Beckmann in 2015. “What few people realize is that audio is so much more than woofers and tweeters, but the ability to innovate with audio has been so stifled that only now [will] we start to see dramatic leaps forward in everything from automotive audio to edge [internet of things] devices to even AI-based medical monitoring.”
The company offers a range of audio product development solutions tailored to automotive, home audio, set-top boxes, and smart home manufacturers, which include Zoox, Bugatti, Panasonic, Sony, Baidu, Bose, Logitech, Tesla, Braun, BMW, GoPro, and Porsche.
DSP’s flagship is Audio Weaver, a cross-platform audio-processing engine featuring a library of over 400 audio building blocks optimized for embedded Arm, Broadcom, Qualcomm, Texas Instruments, Intel, MediaTek, NXP, and other processors. With Audio Weaver, engineers can change signal-processing layouts without recompiling or reflashing hardware, and they’re able to draw on memory profiling to make decisions about processor selection. Plus, they can tune product performance and add features on the fly, all without any need to learn coding skills or a programming language.
Audio Weaver Designer has a drag-and-drop interface with dashed lines representing non-audio control data, or volume and balance; system state information like battery level; and derived values, like event-detection flags. Users can tap modules to ensure such control information is exchanged with the surrounding firmware and adjust mapping, refine control signals, and query the execution time and memory usage of modules in the layout.
DSP also provides TalkTo, a suite of machine learning and microphone processing algorithms that deliver a “quiet-room” experience in noisy environments. It comprises an audio front end that serves as a “microphone cleaner,” using raw audio from a device’s microphone to detect, extract, and clean speech activity in ambient sound. Wake-word engines continuously scan the provided voice stream, looking for the presence of a specific phrase (e.g., “Alexa”). After the wake word is detected, the extracted voice command is passed to a cloud-based or local service, which converts speech-to-text and determines user intent.
“We have been continually impressed by DSP Concepts through our due diligence process,” said Taiwania Capital Fund managing partner Lee Huang in a statement. “The company has executed technology development, demonstrated product-market fit, and … built an impressive list of customers that have validated the value of the technology. We are excited about the opportunity to help the company scale globally, and particularly in Taiwan, where DSP Concepts technology can help domestic companies move up the value chain.”