When scanning the CES 2017 agenda for panels, speeches and so forth, it is difficult to see that barriers have been recently broken in one category that will affect consumer electronics dramatically. What is not widely represented is the importance and the reality of a new brand of AI. Artificial intelligence? No, audio intelligence.
I recently had the pleasure of talking to management at three companies in the audio/voice technology segment, that will attend CES 2017 this year. DSP Concepts, Vesper and Kopin will demonstrate new products and capabilities. I imagine that a flurry of design activity will result from the products they will be showcasing.
At CES, DSP Concepts will demonstrate its Audio Weaver modular programming software used on system-on-chip (SoC) devices that have multiple cores of different types, and which now has multicore support. Audio Weaver enables engineers to build and optimize complex signal-processing chains.
The software is ideal for such applications as the complex, noisy automotive environment and includes microphones, sensor input, voice commands, road noise cancelling, engine sound synthesis, and audio and video entertainment.
The company will also display its recently announced Far-Field Microphone Array Prototyping Kit for MEMS microphone and Internet of Things (IoT) developers, which simplifies the creation of microphone arrays for voice-command and IoT products. Voice- recognition accuracy in voice-command products depends on sophisticated front-end microphone processing. The kit enables improved microphone and voice-recognition performance with echo cancelling, filtering, wind noise suppression, dynamics processing and other functions, implemented via Audio Weaver modular programming software.
Paul Beckmann, CEO and founder of DSP Concepts, walked me through the evolution of audio—from its roots in playback, through its use in talking to another human via cell and speaker phones, and now in talking to a computer. Today computers are listening to what is happening in the environment, have contextual awareness or employ smart listening. There is a wealth of nonverbal information around us in the form of audio. While the automotive industry will be a major player in audio intelligence, so will the medical industry.
“We’re working with Quvium, a start-up in the UK, on a wearable asthma detector for children. When a child has an asthma attack, they need to be treated immediately. Research shows that asthma is triggered well before the attack, after exposure to a pollutant. The child exhibits a pattern of coughing that escalates and manifests over time,” said Beckmann. “We’re working on a wearable cough detector that detects a cough and sends data to the cloud, where it’s determined if it is serious. DSP Concepts provides the algorithm software to identify the 40 types of coughs that exist.”
When asked what he would like CES attendees to understand about audio and its role today, Beckmann shared that audio is here now, and it is playing a much more important role than ever. He offers this advice to design engineers:
- Don’t be cheap: put a minimum of two microphones in your system. If you have at least two microphones, you will be able to unlock advanced audio features.
- Sensor fusion is also very important. The combination of audio, vision and other sensors builds a complete contextual awareness capability.
- Audio will also become important in machine-learning applications.
Check out DSP Concepts demonstrations at the Cadence booth.
Matt Crowley, CEO of Vesper, is convinced that for audio technology, 2017 will be the year of pervasive voice interface, in terms of the breadth of applications. Vesper, an acoustic-sensing start-up company based in Boston, offers piezoelectric MEMS microphones that are natively waterproof, dustproof, particle-resistant and shockproof. Vesper is showcasing its microphone arrays and its wake-on-sound microphone, a sound-powered voice-interface system that eliminates the need to press a button to activate the voice interface.
The technology and product were so compelling, in fact, that Vesper’s co-founder and CTO, Bobby Littrell, was recently awarded Innovator of the Year; and Vesper’s VM1000 piezoelectric MEMS microphone won the Ultimate Product award at the ACE Awards at the Embedded Systems Conference Silicon Valley in early December.
“A lot of the things that we’re demonstrating at CES are removing several design constraints that existed. We’re making possible systems that were simply impossible to build until our technology came along,” said Crowley. “I want to open up peoples’ imaginations and say, 'if you take this technology, you can start building all kinds of voice-interfaced systems.' Our technology can survive in harsh environments indoors and outdoors, so we remove the geographic constraints. And you don’t have to worry about power, so you don’t have to think about these systems. You can now run them off a battery and place them anywhere you want, and put them in the house. Now that the constraints are removed on geography and power, what kind of systems can you now build?”
In a home environment, the system will not record anything unless a keyword is used. The microphone sleeps, waiting for the sound to occur. Instead of hitting a power button, the solution uses energy from the sound itself to turn on the microphone. Crowley expects that a new class of sound-enabled products will be seen, based on this technology, by the second half of next year. Vesper is alone in making it possible to build battery-powered voice- interface systems. The company will have a meeting room at CES and will also display their technology at the Qualcomm booth.
Until now Kopin has worked to transition technologies it perfected for the military, aerospace and enterprise markets—display, audio and voice—and bring them to wearables. Stuart Nixdorff, Sr., vice president of sales and marketing at Kopin, says that what they are seeing this year is particularly interesting. “We think that this is the year of smartphones and mobile phones going into a wearable state because of the change in the user interface from touch to voice. Now, voice can go mobile. The big catalyst going forward is that voice can represent the user interface for virtually everything. This past year, Amazon and then Google have been successful making voice as a primary interface. With that capability applied to wearables, we think that wearables will finally become a mass market.”
Paul Baker, vice president of business development at Kopin, explained what, until now, had stood in the way. “The reality of a voice interface so far has been on the frustrating side. It’s great when it works, but it doesn’t work too much of the time. That’s the focus for Kopin. We make the voice interface reliable. Our Whisper chip effectively handles speech recognition in any type of noise environment, even those that are extremely loud, where you wouldn’t expect to be able to hear. The Whisper chip works in any type of system regardless of the platform, operating system or what speech recognition engines you use, and is designed for any wearable device.”
“If you’re in the wearables business, you probably need to talk to Kopin before undertaking new product development and make sure that you’re getting the most advanced technology,” said Baker. “We’ve been pretty quiet over the past several years while developing new technologies, and many in the wearables space don’t really know what we’ve developed. Now we’re showing our technology to the world and it’s impressive.”
Baker added that voice, as a mass market, was impossible until now. Voice user-interface reliability is necessary, and voice must work as well on the street as it does in a home. "Today people don't believe that it's available because they haven't seen it. That’s what we bring to CES."
I’ll bet the farm that by CES 2018, we will see a wealth of panels and presentations on how to add audio intelligence to your design. Until then, you will have to search out gems like DSP Concepts, Vesper and Kopin—companies ready to make a huge difference in your designs without breaking the proverbial bank.