+1 for search feature
That sounds natural—but things are definitely getting weirder on the AI front.
We just tested the new Automix feature in the DaVinci Resolve 20 beta, and the results were honestly jaw-dropping. It uses AI to dynamically balance dialogue, music, and effects in real time—something you’d normally expect only from an experienced post-production mixer. It’s not just a gimmick; it feels like a serious shift in how we’ll approach mixes.
Both Pro Tools and DaVinci Resolve rely on OpenAI’s Whisper for transcription, and it’s impressively accurate—not just in American English, but also in Hindi, which is a huge plus for multilingual projects.
What really caught me off guard, though, was Steinberg’s approach in Nuendo 14. Even though it uses its own proprietary transcription model, it’s remarkably good at clean dialogue extraction and speaker segmentation—at least in American English. However, the absence of support for Hindi and other South Asian languages makes it pretty much unusable for our workflows.
Also, I don’t think adding a proper search engine for transcripts should be difficult—Avid has already demonstrated that cleanly in Pro Tools at NAB.
Strange times indeed. Still reeling from the Automix shock.
The weirdest part? It seemed to generate plugins on the fly—encapsulating multiple algorithms and presenting them through a plain, minimalist interface.
But here’s the kicker: these plugins don’t actually exist. They’re virtual constructs, built dynamically by the AI in response to the mix.