Audio and voice processing have seen immense progress in the past decade, and the introduction of Deep Learning AI machine learning promises that we should expect more in the future. However, it is also evident that what we have today is good but simply not yet good enough.
mpWAV (official site) is a startup that aims to significantly improve voice processing and recognition by making it easier for both humans and machines to hear people’s voices better. Their solution revolves around extracting human voice signals from any noisy environment. Once the voice has been cleanly extracted, the new signal can be utilized to help users with hearing difficulties or provide better input to machine voice interfaces.
mpWav provides both hardware and software to solve this difficult task. Their hardware consists of an array of two to eight microphones. How many there are depends on the maximum device size allowed, but suffice to say that the more, the better. However, it only takes a couple to make things work well.
They use multiple microphones and the principle of beam-forming to zero in on a precise sound emitter (person who speaks) to extract that particular signal while ignoring others from elsewhere. The tuning can be efficiently specialized since human voices are on a relatively known band.
As several microphones in the array pick up sound from slightly different locations, the data can be utilized to reconstruct a clean soundwave and increase the signal strength (audio energy can be integrated/cumulated) as each microphone contributes to the overall sound input power.
mpWAV CEO Hyung-Min Park (left) and Ubergizmo co-founder Hubert Nguyen (right)
The technology works so well that it can extract people’s voices even though a movie sound is in the background. Locations like cafes or concerts are also good use cases. But teleconferencing may become one of their largest businesses as it’s possible to use a Zoom virtual sound driver to remove unwanted noise.
Retail locations might be another avenue for this technology since many places, such as bus or train stations, feature a glass-separated booth with a 2-way microphone/speaker setup to communicate.
How mpWAV can help people with hearing discomfort in their daily lives. These are individuals who have experienced some degree of hearing loss but who do not want to wear medical-grade hearing aids.
Instead, they can now use the ClearSense Audio app with their personal earbuds (Note that the app might not be available in the various app markets yet.). The app will significantly reduce the ambient noise and only capture the voices of people talking to you. I’m convinced that consumer-level devices and software will eventually up-end the current “medical hearing aid” market.
Even if you don’t experience hearing loss, this could be useful because you’re in a noisy place. The live demonstration I got to try was very convincing, and I’m looking forward to testing it on my phone during a conference, flight, or helicopter ride to push it to its limits. However, you don’t need to go to such extremes to benefit from this technology. Smart Homes or Automotive applications are far from perfect today regarding voice commands. They could benefit from a boost in capabilities.
The prospects of this company are very promising as it solves the difficulty of “hearing people” in very challenging environments for humans and machines. Their intellectual property is protected by patents that might make mpWave a larger player in this space or even possibly an acquisition target someday.
Filed in . Read more about CES, CES 2024, Korea, SBA, Seoul, Seoul Business Agency, Startups and Voice Recognition.