AudioTelligence unveils software to spell the end of unwanted audio in video conferencing and smartphone videos
We have all become accustomed to unfortunate interruptions during video conferencing calls, or when recording video on our phones.
But now Cambridge audio software specialist AudioTelligence has unveiled an astonishing solution that for the first time will allow you to remove the voice and images of interlopers - in real time, or on recordings - without losing what the main speaker is saying. And it is easy as cropping out an unwanted area of a picture.
The revolutionary technology is embedded in a family of products for smartphones, released under the company’s new consumer technology brand Aiso.
It promises to help make it easier to achieve the perfect soundtrack at the first attempt and could spell the end of those conferencing interruptions for people working from home.
AudioTelligence CEO Ken Roberts said: “The technology is compatible with any video call or video meeting app such as Zoom and Microsoft Teams, with all sounds originating from outside the camera's field of view automatically discarded in real time.
“And it makes sound editing your TikTok or YouTube video as easy as framing the subject in a photo. The user simply pinches and zooms to select a rectangular area of the video containing the target sound sources. Interfering voices from outside this area are then removed from the audio.”
Under the Aiso brand, the company - based in Broers Building on JJ Thomson Avenue - is initially launching two product lines for smartphones, each with two features.
AudioCrop and AudioTag are designed for smartphone videos and can be used to select the desired audio either in real time or during post-processing.
AudioCrop allows a specific part of a video frame to be selected, with audio and/or visual interruptions from outside that area removed. AudioTag, meanwhile, allows you to select specific sound sources based on their location, so that all other sound sources can be discarded.
The second product line is for video calls or video conferences and will provide a new CallCrop or CallTag sound option in video meeting apps, alongside the standard internal microphone and Bluetooth headset options.
As in the first product line, CallCrop focuses on a particular part of the video frame, while CallTag focuses on an individual sound source.
While AudioTelligence is not the first to attempt to clean up such background distractions, existing 'audio zoom' solutions are limited because based on imprecise ‘beamforming’ technology, which is only capable of focusing audio capture within a range of tens of degrees.
Aiso products use technology known as blind source separation (BSS), which is more effective at separating target sound sources from interfering ones.
Unlike noise suppression technology, this is capable of handling overlapping speech signals, meaning that if new sources appear, they can also be eliminated. Impressively, BSS still works even if the source of interest isn’t dominant.
AudioTelligence says its BSS technology improves the signal-to-interference ratio by up to 25 decibels (dB) on a three-microphone smartphone.
It is now focused on licensing its Aiso family of products to smartphone and tablet manufacturers and developers.
Designed as flexible, software-only solutions, they work on any multi-microphone Android smartphone or tablet.
AudioTelligence says the lightweight embedded software is simple for smartphone OEMs to integrate as it does not require a dedicated hardware codec or DSP and works with standard microphones, without the need for special microphone positioning.