Syntiant Raises $35M for On-Edge AI and Voice Processors
Artificial intelligence and voice-focused computer chip startup Syntiant has closed a $35 million funding round led by Applied Ventures and Microsoft’s M12 Ventures. Syntiant’s chips are designed to operate on the edge, without needing to send data to the cloud. That makes them far more power-efficient than standard chipsets when it comes to processing voice assistant wake words and commands. Keeping the process on the device also helps protect user privacy, a highly valuable trait in a market where privacy concerns make many reluctant to purchase voice-activated devices.
Syntiant is best known for its Neural Decision Processors (NDP), which can process audio or other sensory data and determine if a wake word or other activation signal has been used internally, without verification from a cloud server. Less power is consumed by keeping the process on the device, with Syntiant claiming its chips process ten times the data of standard designs using just a hundredth of the energy despite being half the size. Manufacturers can program the chips to recognize 64 different sounds, assigning functions to each. So, no only could a smart speaker might understand different wake words, but a smart baby monitor could send a text to a user whenever the sound of crying is heard, or a smart burglar alarm could turn on a siren when it hears glass breaking.
There is an appetite for these kinds of chips. Syntiant noted in its funding announcement that it has shipped more than a million of its NDP chips to customers. Some of that is likely a result of Amazon certifying Syntiant’s chips for integration into Alexa-powered devices almost exactly a year ago, but the chips are used in devices running Google Assistant, Siri and Baidu’s DuerOS as well. The new round brings Syntiant’s total funding to about $65 million, including previous investment from M12 Ventures as well as Intel and the Amazon Alexa Fund. The new capital will go to expanding Syntiant’s production capacity to get the chips into many more devices.
“It is a tremendous honor to know that some of the world’s leading tech investors are supporting our growth stage, as we deliver our deep learning voice solution to customers across the globe,” Syntaint CEO Kurt Busch said in a statement. “We are especially thrilled that production volumes of applications using our neural decision processors are increasing and expect orders to ramp even higher throughout the remainder of 2020, as our NDPs continue to set the standard for always on voice as the new interface.”
Though Syntiant has a lot of deep-pocketed backing, it isn’t alone in providing on-edge processors. Startups like Picovoice and Sensory offer on-edge voice platforms, while ID R&D’s vocal identification engine small enough to bring vocal security to devices without needing to transmit data. Meanwhile, Apple spent a reported $200 million to acquire AI startup Xnor to add the same kind of technology to its portfolio. The increased privacy of on-edge processors can be a significant part of their appeal in addition to the reduction in cost. Though voice assistant developers are eager to tout their focus on addressing privacy concerns, companies who want voice controls may prefer using chips from a third party that keeps everything on a device.