Personal Noise Reduction – out of the box deployment
- On November 19, 2016
- personal noise reduction
The personal noise reduction software has proven to increase audio quality and MOS score in different noisy environments. This technology is using information on the speaker’s voice in order to separate it from the background noise and attenuate the latter. This powerful technology had one major drawback that prevented it from reaching mass deployment. This drawback is the initial setup that was required from the end-user prior to using it. The initial setup was the need to provide proper training data which includes capturing few minutes of the user’s voice in a relatively quiet environment. These days, SoliCall is proud to introduce the next level in its revolutionary technology which enables bypassing the initial training step.
Automatic tuning to the speaker
In order to avoid the initial setup, SoliCall developed a version that enables it to automatically tune to the speaker without the need to provide initial training data.
The practical outcome is that the powerful personal noise reduction can be kicked in instantly at any call and at any time regardless the identification of the participants. For example, if you get a call from a person that is located in a noisy location (e.g. in a noisy street or in a dance club), you can immediately use the powerful personal noise reduction algorithm to attenuate the noise that you hear and improve your ability to concentrate on the voice itself. This automatic tuning technology can also be used by network operator or voice solution provider. For example, a conference bridge provider can monitor the incoming audio ports that dial into the conference bridge. Once an incoming audio port is generating noise, it can automatically be filtered by the personal noise reduction algorithm in order to improve the quality of the conference call.
Under the hood
The powerful noise reduction algorithm was combined with out-of-the-box activation capabilities using two distinct technologies:
1. The personal reduction supports a tolerant mode that does not enforce a “biometric” identification to a specific person. It works with a “representative” group of people with similar acoustic behavior.
2. The system is pre-loaded with few representative acoustic profiles and a unique technology can automatically selected, on the fly, the best match.
The program execution is very simple. When a call starts, the system automatically scans the pre-loaded profiles. Almost instantly, it locates the best profile that can be used to provide superb personal noise reduction filtering. The filter’s quality can be enhanced further by enriching the pre-loaded profiles with a personal acoustic profile but this is no longer a prerequisite for using the personal noise reduction algorithm.