Stepping into the Sony Studios event felt like a familiar journey back to early days cutting tracks with Pro Tools systems across Fox, Sony, Disney, and Warner Brothers. Surrounded by top industry professionals, I paused for a deep-dive demo at the Focusrite table, where experts from Focusrite and AudioShake were breaking down a technological leap completely changing post-production: advanced source separation.
The presentation dove straight into practical audio workflows, examining how these powerhouse brands tackle everything from historical preservation to everyday broadcast challenges.
Brand Spotlights and Demonstrations
Focusrite & AudioShake Partnership: Focusrite hosted the session alongside AudioShake, showcasing how their cutting-edge platform handles automated source separation and stem extraction.
Warner Brothers Post-Production Collaboration: Warner Brothers post-production teams provided valuable insight and audio-visual clips from The Wizard of Oz, demonstrating how tools like AudioShake were deployed on Stage Five for pre-dubbing and finalling.
Real-World Industry Trust: The brands highlighted that their platforms are trusted across major music labels and sports entities to handle catalog remixes, rights clearance, and clean re-cuts without copyright infringement.
Practical Workflow Solutions Discussed
Handling Environmental Distractions: If an actor’s emotional delivery or dialogue is disrupted by an unwanted sound like a plane flying overhead, engineers can cleanly pull that noise out to salvage original audio and streamline ADR mixes.
Localization and Re-Cuts: Sports broadcasting and catalog content can isolate dialogue or remove music to handle re-cuts without triggering copyright infringement or to cleanly manage expired music licenses.
De-Noise and De-reverb Modules: Specialized models are trained specifically on pre-digital, low-quality analog recordings or heavy environmental noise (such as phone calls or car audio), while separate de-reverb models target room acoustics as an independent object rather than breaking a stem.
Granular Instrument Stems: Alongside vocals, workflows support isolating musical instrument groups like bass, guitar, and drums, with beta tests expanding into individual drum kit components (separating kicks, snares, toms, and hi-hats).
Non-Generative Precision vs. Generative AI
Deterministic Output: Unlike generative AI—which guesses outcomes and can change with every pass—source separation uses precise spectrogram analysis and machine learning models trained on human voices and instruments to create a consistent, exact mask.
Interactive Adjustments: While the separation itself is fixed out of the system, engineers can quickly pull background tracks down or push dialogue up in a mix down without needing to completely rebuild a session in Pro Tools.
It is an incredible time to be in audio engineering. With brands like Focusrite and AudioShake pushing the boundaries, tasks that once took hours of painstaking manual cleanup can now be achieved instantaneously, letting professionals focus entirely on the creative process rather than the tedious hurdles of the past.