Focus Voice Module | accusonus

Far-Field Voice Enhancement

Accusonus has created a unique voice pre-processing module that enhances speech quality and intelligibility for far-field applications. The Focus Voice Module effectively reduces the perceived user-to-device distance when the user is away from the microphone. Our patented Focus Voice Module delivers a clear and comprehensible speech signal in applications such as speakerphones, smart devices, microphone bars and IOT appliances.

Reverberation and Noise are the two main mechanisms that contribute to the complex far-field problem:
Reverberation: In an indoor environment, speech bounces off walls and nearby surfaces and creates reflections, known as reverberation. When the user is close to the device, speech is captured clearly and reflections are minimal. However, as the user-to-device distance grows reverberation becomes significant and speech quality suffers. Studies show that speech recognition performance drops dramatically for distances longer than 2m. These are typically called far-field applications.
Noise: Noise coming from open windows, air conditioners, vacuum cleaners and other domestic appliances is a common problem in indoor voice applications. In the far-field case noise becomes an even bigger problem, since the energy of noise sources is increased as it bounces off walls and surfaces.

Accusonus has developed the Focus Voice Module specifically for far-field voice enhancement. The Focus Voice Module consists of two patented embedded software products (Focus-MDR and Focus-DNR) that simultaneously address the far-field problem and are the result of many years of intensive R&D. The Focus Voice Module essentially brings the effective position of a user virtually closer to the microphone and therefore the device operates as if the user was within a few centimeters. The Focus Voice Module offers greater speech quality, better speech intelligibility and more robust speech recognition performance.

Note: The Focus Voice Module shares similar technological background with ERA-D, our general purpose sound restoration tool. If you are interested in Focus Voice Module and want to perform a quick evaluation of the underlying technology, please download the ERA-D trial and contact us to arrange an evaluation license.

Focus-MDR and Focus-DNR have been optimized
for Cadence® Tensilica® HiFi Audio/Voice DSPs and for Qualcomm® BlueCore CSR8670/75


Focus Voice Module

Embedded products based on patented technology

Focus-MDR (single and multi-channel dereverberation)

Patented reverberation suppression, that effectively reduces the perceived user-to-microphone distance solving the far-field voice problem in a unique way

Focus-DNR (single and multi-channel denoise)

Novel solution to the noise suppression problem, takes advantage of multi-microphone configurations and removes noise in reverberant conditions

Focus-MDR&DNR (joint single and multi-channel denoise and dereverberation)

Tight and efficient integration of Focus-MDR and Focus-DNR, offering optimal voice quality in challenging far-field cases

Focus-BDR (binaural dereverberation and ambient noise suppression)

Binaural dereverberation and ambient noise suppression, preserving the binaural localization cues targeting the hearing aids market


Please contact us for further details on the specifications of this product.
Please contact us for further details on the specifications of this product.