After the first days of quarantine we've noticed a lot of complaints from people working from home due to noisy environments and we realized how lucky we are to either live in spacious homes or to be able to afford expensive hardware such as Bose or Plantronics headsets.
Since our team has become quite specialized in sound technology while developing SONIC Tag, we knew that we can develop a tool that can help home workers during the Pandemic. When we found out about the #EUvsVirus initiative we decided to enter the Pan-European Hackathon and join our efforts to create this free noise cancelling solution that can empower people to be productive while working from home.
We are confident NoiseBlock will make life easier for people who are not lucky enough to have privacy and silence around while they work from home.
What it does
NoiseBlock is a Windows Tray application that filters the surrounding noise feeding into the device microphone, allowing you to enjoy a crystal clear conversation with your team.
- NoiseBlock will be Free to use for any user world wide
- NoiseBlock can be scaled on potentially billions of devices
- The learning algorithm we are using allows our solution to evolve over time and has the potential to become one of the best noise cancelling solutions in the world, one that can be used by the people who need it the most.
How we built it
Having 3+ years of experience while developing real time sound processing technologies for SONIC Tag allowed us to quickly develop a mass market solution while using deep learning algorithms for noise suppression.
In the process of creating NoiseBlock, we were guided by the VIRUS design principles:
- Real Time Processing
- User Friendly
- Seamless Experience
- NoiseBlock works directly with the device microphone driver using low level C programming to process the sound input.
- This allows NoiseBlock to have an excellent performance and is needed because we are processing the microphone in real time with almost no latency. Excessive audio latency has the potential to degrade call quality in telecommunications applications and NoiseBlock was specifically created to avoid just that.
- The NoiseBlock learning algorithm is capable of evolving over time and has the potential to become one of the best noise cancelling solutions in the world.
NoiseBlock was created for the EUvsVirus Hackathon and was developed using a mix of coding languages and technologies:
- Visual Studio
- Machine Learning
- HTML & CSS
- VB-CABLE Virtual Audio Device
- NAudio (.NET audio library)
Challenges we ran into
Working with deep learning and recurrent neural network algorithms for audio noise suppression was one of the biggest technical challenges we had to overcome. Although tired and with limited time, we remained motivated to add value to people's lives during and hopefully after the Pandemic as well.
Having developed a solution that needs to be installed on Windows devices and that also includes an audio Driver, we are required to purchase a Code Signing EV Certificate approved by Microsoft.
Accomplishments that we are proud of
In such a short period of time we have created a quality product that will improve the quality of working from home. The level of noise reduction exceeded our expectations and we are happy to see very good initial results in different testing environments. Furthermore, our team is now even stronger and inspired to bring quality experiences to our users. We are proud to say that we already have the beta version ready and we estimate that in the next 2 weeks we will have the final product ready.
What we learned
Hackathons bring people together and, in a short period of time, the possibility to learn new skills, find new collaborators, and plan future work that can be surprisingly fulfilling. This will also be useful after the Hackathon ends, as we plan to remain active in the EUvsVirus Community and continue to elaborate on ideas and inspire new, creative ways of thinking about projects.
Furthermore, it was both humbling and insightful to learn more about the various working from home conditions people have and to be able to care for that need. Small apartments, washing machines, loud neighbors and all sorts of other unthinkable noise generators combined with non ideal environments go to show some of the social inequalities humanity is still struggling with.
What's next for SONIC NoiseBlock
In the following months we plan to go ahead with the next steps in the development of the product:
- launch the product and drive social impact worldwide while gathering feedback
- improve the learning algorithm to take into account even more scenarios
- create a NoiseBlock version to cover all devices (MacOS, iOS, Android etc.)
- partner with European Companies and integrate our technology in other platforms, software and hardware
- develop a one stop shop solution suite for the future of remote work
At the same time, we are trying to develop a Business Plan that will enable us to:
- Make NoiseBlock available for Free and train the algorithm at the same time in order to provide a better user experience moving forward
- Use a Donationware licensing model and, on the long term, offer a Pro Version for companies and enterprises