Why dereverberation is a game-changing feature for the semiconductor industry | accusonus

Why dereverberation is a game-changing feature for the semiconductor industry

"With our patented dereverberation solution, semiconductor companies gain a significant competitive advantage. However, it's the end user who will experience the real change, since her everyday aural experiences will dramatically improve!"

Alex Tsilfidis- CEO, Accusonus

Why dereverberation is a game-changing feature for the semiconductor industry

Congratulations! You just bought a brand new apartment. After all those years of hard work, you can finally enjoy the spacious living room, the large bedroom and the huge kitchen where your culinary talent will shine! Your new furniture will arrive in 4 weeks though and until then you have to spend all day in your large new home. Well, I bet that in a few days you'll be desperately missing your old cosy studio, the main reason being that you cannot stand the sound of your empty apartment anymore!!

Every room has a sound signature which is called reverberation. You can't hear a room if you don't trigger it with a sound (for example with your voice when you call your best friend to moan about the delay in the delivery of your furniture!). When you finally trigger it with a sound, the room's reverberation is always there! And since it's always there since the prehistoric ages (imagine the reverberation of those big caves that our ancestors used to live in), our auditory system can easily adapt to it and we are rather used to its presence. However, listening in reverberant environments significantly reduces what we call "listening comfort". This is the reason you will miss your cosy studio until your furniture arrives to reduce the reverberation of your new apartment. In reverberant conditions, people quickly get tired and distracted by conversations. Non-native speakers face big problems also in understanding speech. Both the reduction of speech intelligibility and the acoustic discomfort are even more pronounced for people with hearing difficulties!

Our auditory system has been evolving for millions of years, but our Automatic Speech Recognition (ASR) engines have been struggling to keep up during the last 60 years. Therefore, reverberation is a nightmare for SIRI and her friends (i.e. all other speech recognition engines). It appears that the recognition performance can easily drop down to 30-40 percent in everyday environments and the decrease is more dramatic in highly reverberant spaces (think of a church or a theatre foyer). Reverberation is one of the main reasons automatic speech recognition is not working as users expect: it undermines the user experience, rendering the technology less appealing for frequent use.

All these problems can be (partially or totally) solved by using dereverberation processing. Despite the fact that dereverberation is a hot research topic for 4 decades, it appears to be a very difficult task. Traditional dereverberation methods are computationally expensive and introduce severe processing artifacts.

Until now, that is. Accusonus has recently presented the first truly robust solution to the dereverberation problem, in the form of two patented products:Focus-MDR and Focus-BDR. After many years of research and development, our embedded software solutions overcome the flaws of traditional approaches, being the first products that perform high-quality dereverberation, without sacrificing speed and efficiency.

Our Focus-MDR can be a game changer in:

  • Mobile applications (i.e. smartphones and tablets): In typical cases of mobile device use such as video calls, conference calls, hands-free applications, media consuming, etc, consumers are holding their devices away from their mouth at a minimum of arm's length (in what is called the “free-field” condition). Therefore, reverberation becomes really intrusive and Focus-MDR will be an important feature that significantly enhances the user experience!
  • Smart home devices: In quiet environments (like our homes) reverberation can be a bigger problem for ASR engines than noise! Recently both Samsung and LG have released a remote TV control that can be used as microphone in order to control the TV with speech. This can be a bit awkward, even if you are a big karaoke fan. Using Focus-MDR and talking directly to the TV (or any other device) can create a really immersive and natural user experience.
  • Teleconference and blue tooth accessories: In professional environments people have to take fast (and smart) decisions. Focus-MDR will significantly reduce the cognitive load of communicating under reverberant conditions. Moreover, Focus-MDR will enhance the performance of dictation and other office applications that are based on ASR!
  • Hi-end toys/ robots: Toys are becoming increasingly sophisticated, embracing ASR technologies to achieve a high level of interaction. Focus-MDR can enhance their ability to understand us, making their use more fun and intuitive!
    On the other hand, our Focus-BDR has the potential to be a game changer in the Hearing Aids industry. Focus-BDR performs binaural dereverberation processing by preserving sound localization cues for the hearing aid users. Our first feedback from the clinical tests is impressive!

    At accusonus, we are very proud to have a final solution for the reverberation problem. We are confident that semiconductor companies will gain a significant advantage by embracing our breakthrough solutions. However, it's the end user (every single one of us!) who will experience the real change. Although furniture will not arrive on time in our new apartment, our everyday aural experiences will dramatically improve!

  • 0 Comments - Why dereverberation is a game-changing feature for the semiconductor industry

    Leave a comment

    This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.