Reference designs overcome interfering ambient noise
Article By : Steve Taranovich
DSP Concepts CTO Paul Beckmann demonstrated how the company's Voice UI works in the presence of high levels of interfering ambient noise.
Recently, I visited with DSP Concepts in Silicon Valley and spoke to CTO Paul Beckmann. Their lab has a variety of acoustical chambers for testing their solutions as well as competitor’s solutions in order to improve the performance of their Voice UI in practical applications (Figure 1).
Figure 1 DSP Concepts sound room (Image courtesy of Loretta Taranovich)
DSP Concepts has a reference design they did with Amlogic. Beckmann showed me an awesome demo of Voice UI responding to human voice commands in the presence of high levels of interfering ambient noise. This would make Voice UI so much more practical in the home, but especially in noisy industrial environments.
The demo had a series of multi-microphone reference designs; two had two microphones, and two had six microphones. For comparison, there was an Amazon Echo Dot with seven microphones. The DSP Concepts six-microphone solution was the only one with a speaker connected and is what you’ll hear in the demo.
There were also two low-end Voice UI systems, also reference designs done with Ambiq Micro, which operated with low power capabilities for wearables that we would discuss separately after the demo (Figure 2).
Figure 2 Ambiq Micro low-power wearable designs (Image courtesy of Loretta Taranovich)
This demo showed me two areas of marked improvement in real-life situations. The first showed how the products reacted with interfering noise such as in the home kitchen with a fan/air-conditioner running or while watching TV. Can these products still understand your voice commands? The Google Home made the interfering sound in the first demo.
In the second demo, what happens when the product itself is playing music, but you still want to be able to control it with only your voice being heard? You need an algorithm known as an echo canceller to get rid of that music/sound being played out of the speaker.
See the demo video below. In the background of the video, you will see Cynthia Hoye of Hoye Communications. Without her I would not have known of DSP Concepts efforts with Voice UI so early on. I was introduced to Paul Beckmann early on as well because of her.
Beckmann told me that this in-ear, personal assistant wearable design architecture is the lowest power solution in the market. This enables the device to be ‘always-on’ and listening without having to go into a sleep mode at any time. The device will run for a week powered only off a small coin cell. This device also does not need a ‘push-to-talk’ like the Apple EarPods.
Figure 3 Beckmann says that his vision is to put voice into TVs, set-top boxes, and even wearables. (Image courtesy of Loretta Taranovich)
There is no doubt that Voice UI will be a critical part of electronics technology going forward and creative software, along with good microphone technology, will make it an enabling force within many areas of the huge IoT growth to come.
Steve Taranovich is a senior technical editor at EDN with 45 years of experience in the electronics industry.
Related articles:
- Alexa, can you hear me now? Low power voice interface technology evolves
- Enhancing audio system designs with digital signal processing
- Teardown: Amazon’s Echo voice-activated virtual assistant
- The fundamental flaw of too many listening devices
- New technology provides breakthrough speech recognition in noisy environments
- A soft whisper heard clearly: Getting automatic speech recognition right