We all know that you can't un-bake a cake. Once everything has been mixed together and baked in the oven, it is physically impossible to go back in and remove some of the ingredients or change the recipe. In a similar way, it’s not possible to take a pot of green paint and separate it back into its blue and yellow components. Once things are mixed, they are inextricably mixed. For decades, broadcast and audio professionals have operated under this exact same unwritten law: once the dialog, music and effects have been combined into a final mix they are locked together forever.

Or are they?

What if you could unbake that cake?

The Missing Ingredients

Content owners and studios are sitting on enormous libraries of finished programming. But in today's global, multi-platform media landscape, a single finished mix is not always enough. Assets constantly need to be repurposed, adapted and modified for new distribution opportunities, however the original isolated multitracks are often unavailable or lost.

This presents several frustrating challenges:

• The Localization Dilemma: Suppose you have a hit television show or a classic film that you want to release in a new international market. To dub the program into a new language you need to remove the original dialog component but keep the music and effects (the “M&E track”) intact. If you only have the final mix available, this would have traditionally meant hours of painstaking, expensive manual editing, or entirely recreating the music and effects from scratch.

• The Intelligibility Issue: We’ve all experienced the frustration of watching a movie where the background noise, explosions, or dramatic musical score completely drown out what the characters are actually saying. If a broadcaster receives a final mix with poor dialog intelligibility, there is very little they can do to fix it using traditional EQ or compression methods, since any change made to the dialog will also be applied to the underlying music and effects.

• The Compliance and Promo Pivot: Imagine you want to cut a 30 second promotional trailer for a film, but the music in the final mix jumps around awkwardly when you make your edits. Or perhaps you want to distribute a classic series on a new streaming platform, but the rights to the song playing in the background of a key scene have expired. You need to strip away the music track so you can replace it with a royalty-free alternative, but you have to preserve the dialog and effects. Again, with a baked mix, you are stuck.

But what if you could un-bake the cake?

A New Module for the Menu

Introducing AudioShake: sound separation for your ATS workflow

Telos Alliance has partnered with AudioShake, an industry leader in AI-driven sound separation technology, to integrate their powerful tools directly into the AudioTools Server environment.

The new AudioTools Source Separation by AudioShake module enables AudioTools Server users to extract the dialog, music and effects components from finished program audio. To use the baking analogy, it effectively hands you back the raw ingredients of your mix.

Need to prep an old movie for international dubbing? The module can effortlessly generate a new M&E track by stripping away the dialog. Need to improve intelligibility? You can isolate the dialog stem, clean it up and balance it against the background audio. Need to edit a promo or swap out an expired music license? Simply extract the music stem and drop in your replacement. It turns static, finished media archives into practical, flexible assets ready for modern distribution workflows.

The AudioTools Advantage

AudioShake integrates right into AudioTools Server. Click for a closer view.

While the concept of AI audio separation is generating a lot of buzz in the industry, the true power of this technology relies on how it is implemented. Standalone tools are great for one-off projects, but for enterprise-level broadcasters and studios processing massive content libraries, integration and security are paramount. This is where the AudioTools Server integration truly shines.

Firstly, because Source Separation by AudioShake is a module within AudioTools Server, it doesn't have to exist in a vacuum and can instead form a part of an existing, wider automated workflow. For example, a profile that automatically ingests a watch-folder of legacy stereo television episodes, removes the dialog component to generate a new M&E track, upmixes the audio to 5.1, loudness processes it and then combines the final audio with the video.

Secondly, and crucially for enterprise customers, all processing is done locally.

In an era where many AI tools require you to upload your files to the cloud for processing, data security has become a massive concern. When you are dealing with highly sensitive, pre-release theatrical content or proprietary corporate media, sending unencrypted files to an external server is simply not an option. Because the AudioShake integration lives inside your AudioTools Server environment, your content never leaves your facility. The processing utilizes your own local computing power, ensuring strict compliance with your internal security protocols and giving you total peace of mind.

Having Your Cake and Eating It Too

The Source Separation by AudioShake module is available now for AudioTools Server users on a flexible monthly or annual subscription basis, bringing the magic of AI un-mixing directly into the automated, file-based workflows you already trust.

In life, we are often told that we have to accept things as they are. We are told that you can't un-mix the paint and you certainly can't un-bake that cake. But in the world of file-based audio processing, the rules have officially changed. With AudioTools Server and AudioShake, you finally have the power to take back your ingredients, tweak the recipe, and serve up exactly what your audience is hungry for.

To learn more about how AudioTools Server can help make intelligibility issues a thing of the past, please contact your AudioTools Server representative. We’d love to hear from you!