Accentize is a young start-up based in Darmstadt, Germany which develops cutting-edge audio processing software using machine learning techniques. After some years of academic research we realized that there is a huge gap between state-of-the-art theoretical insights and available productive implementations. Our products aim at filling this gap by applying intelligent algorithms to common audio processing challenges. Most of our currently available products focus on speech enhancing and restoration applications for post-production and content creation.
In order to make our software “intelligent” we combine different machine learning practices with digital signal processing techniques. The core building block of our applications are so-called artificial neural networks (ANNs). ANNs can be trained to model complex behaviours and find solutions to problems which are difficult to come up with using conventional hand-crafted algorithms. After successfully finishing our first research projects, we now have a set of pre-trained algorithms at hand which we constantly try to improve and apply to new challenges. If you happen to face any audio processing problem which you think could be solved using our approaches feel free to send us a mail to mail[at]accentize.com.
Besides the Accentize audio plugins, we also offer licensing of our existing algorithms (Noise Reduction, Reverb Suppression, Automatic Equalization…) and other custom-tailored audio processing solutions for your application. If you are interested please send us a mail to licensing[at]accentize.com and we will get back to you for further analysis and discussion.
Chameleon is being used to match dialogue recorded in a studio to recordings in different locations
Chameleon is being used to match foley to a dialogue recording in an underground parking lot
Chameleon is being used to match a dry hihat-sample to a real drum-set by imitating the reverb of the recording room
Chameleon is being used to increase the reverb amount of a guitar recording. The existing reverb is imitated and boosted.