Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Innovative isolation: Audionamix drives ADX technology forward

Following the 2010 release of a refreshed version of Vera Lynn's We'll Meet Again, momentum behind Audionamix's ADX technology is growing.

French technology company Audionamix has quietly been making inroads into the post-production and music communities since working on its first major project, director Olivier Dahan’s Edith Piaf biopic La Vie en Rose in 2007.

PSNEurope readers may recall encountering the name in February 2010, in a cover story about a Vera Lynn recording. Momentum has since been building for the company’s ADX technology, which at its essence enables content owners to monetise their assets by isolating key elements of a mixed master audio track where the individual tracks are not available.

“The technology was born out of the desire to identify the DNA of audio so that, instead of transferring a music file, you’d transfer the DNA and recompose it at the other end,” explains Arnaud Dudemaine (pictured), VP operations of the company’s US office. “In the process, we discovered this way of being able to differentiate and isolate and basically teach a computer to listen to music and emulate how we can do this naturally in our mind.”
By enabling a computer to replicate what is known as the “cocktail party effect”, whereby humans can focus on a conversation across a crowded room while ignoring any other noise, Audionamix is able to isolate vocals and individual instruments from any mixed format, from mono to stereo and beyond.

For La Vie en Rose, the company created vocal and instrument stems from both mono and stereo sources that allowed the filmmaker to reposition them in the soundtrack mix. In the 2010 PSNEurope story, the voice of a 22-year-old Vera Lynn was extracted from a 1939 mono recording so it could be supplanted into a fresh orchestration and arrangement of her iconic song We’ll Meet Again.
This is not phase-based technology. “It’s a non-destructive process where the sum of the parts is exactly equal to the original. So what a layman might hear as an artifact is actually not an artifact, it’s impure separation. We get to 98, 99% separation, and then that last percent, we can’t quite get automatically, so we have to go in and try to manipulate it. But it’s not about cleanup; it’s about perfecting the separation,” says Dudemaine.
@page_break@ Audionamix is hands-off with regard to the creative process. “It’s a very deliberate choice on our behalf that we don’t manipulate the audio or enhance the audio. We provide a raw track which is the separation of the original content, but the creative decisions to enhance with EQ, delay and time stretch and ‘make it sound better’ according to that creative vision is not one we take on.”
Audionamix divides its services into five basic areas: music dissociation, where music licensed for a specific territory, for example, is removed, leaving dialogue and effects intact, allowing new licensed music to be laid in; dialogue isolation, where dialogue is removed from content or isolated for use elsewhere; vocal/instrument isolation, isolating and preserving musical elements similarly to dialogue; sync track/instrumental creation, where lead vocals are removed or replaced; and stem creation, where individual elements are isolated for remixing.
Composer Hans Zimmer also had Audionamix extract stems from Piaf’s La Vie en Rose, remixing it for 5.1 presentation and manipulating the horns into a repeating motif throughout Christopher Nolan’s film, Inception. That inspired the company to further develop its tools for music dissociation, says Dudemaine.
“There was a request from CBS for the treatment of foreign versions of TV shows where the original elements, the foreign dubs, had never been preserved. So when they came to renegotiate the music licensing rights, the cost was too expensive for the return on investment for those territories, so they decided to change the music. Changing the music domestically was very easy because they had the separate elements, but in these foreign territories, they didn’t.”
Another early driver of the technology was a 50th anniversary surround remix of Psycho in 2010, where the dialogue was replaced with French actors, and a French language release of The Blues Brothers. “The voice actors there are just as iconic as our actors are here, so to pull out that dialogue and put it into a new surround mix enhances the product for those international territories,” Dudemaine explains.
In the US, demand has recently increased for music-only projects. Rod Stewart sang a duet with an ADX-extracted Ella Fitzgerald vocal on his Christmas 2012 album. In January 2013, Andrea Bocelli released a duet with Piaf using her ADX-isolated vocal.
The company is now anticipating public reaction to Verve Remixed: The First Ladies, an album released on 9 July by the jazz label that is wholly dependent on ADX. “The clear artistic direction was, ‘We want to use this technology to see if we can broaden the audience of our classic Verve catalogue and bring it to younger generations’,” reports Dudemaine.
For now, Audionamix performs the ADX work, pricing each project according to complexity and the desired quality level. But the potential for partnerships and software products is significant.
“We’re really on that cusp of transitioning from being a research laboratory into a technology company,” observes Dudemaine. “By the end of the year, I’m sure we’re going to have some pretty exciting things to talk about.”

Steve Harvey