AudioShake at NAB 2026: Real-Time Dialogue Isolation, ESPN, and What Broadcast Is Asking For Next
AudioShake arrived in Vegas with a new real-time model (Dialogue RT), a partnership with ESPN, and a booth built around the live broadcast workflows we've been working towards for two years. We left with the clearest signal yet of where the industry is going — and what to build next.
The conversations this year were noticeably more specific. Less "is AI ready for broadcast?" and more, "here's the latency budget I'm working with, where do you fit?" AudioShake came prepared with our fastest, highest quality models yet.
Here's what we launched, who we partnered with, what we showed on the floor, and where it’s all heading.
Dialogue RT: real-time dialogue isolation, built for live broadcast
The headline launch was — AudioShake's real-time dialogue isolation model, running end-to-end at 11ms latency.
11ms isn't a benchmark. It's the threshold where dialogue isolation can sit inside a live chain without breaking lip-sync. That's the difference between a post-production tool and a live tool, and Dialogue RT is the first high-quality dialogue isolation model to clear it.
A few things that make it distinct from anything else available right now:
- It separates rather than suppresses. Most existing tools reduce noise by suppressing what isn't speech. Dialogue RT outputs two independent streams (a clean dialogue stem and a separate background stem) giving engineers full control over both, rather than forcing a tradeoff.
- No tuning required. It adapts to noisy environments in real time. No threshold management, no per-venue calibration.
- It runs natively in the chain. Available via the AudioShake SDK, Dialogue RT is built to deploy inside existing broadcast infrastructure.
Use cases we walked through at the booth:
- Live sports — isolating commentators, sideline mics, and on-field audio before it goes to air
- Breaking news — cleaning field reporter and anchor feeds in real time
- Stadium and live events — serving international distribution, in-venue audio, and accessibility from a single feed
- Real-time captioning and ASR — feeding cleaner dialogue into transcription pipelines for higher accuracy without perceptible delay
Read the full Dialogue RT announcement →
The ESPN partnership: AI audio separation at sports media scale
The other major announcement was our partnership with ESPN.
The first piece of work to surface publicly: pulling Phil Simms' "I'm going to Disney World!" clean from a 35-year-old mixed master for ESPN's 2026 Super Bowl ad, without licensing the music baked into the original recording.
That's one example of a much larger problem we're solving across ESPN's catalog. A huge portion of sports media sits in exactly this state: music cleared once, for one distribution window, now making decades of footage unusable across the platforms that came after. The Ocho, ESPN's archive of T-Rex races, stein-holding championships, and competitive sign spinning, is a case in point. Decades of compelling content, locked behind expired music licenses, with no stems to work from.
The work has grown beyond archival. ESPN now uses AudioShake to clear music from sports highlights for social and digital distribution, and to isolate dialogue from sideline interviews, locker rooms, and mic'd-up moments using Dialogue RT.
Read the full ESPN announcement →
Live broadcast demos with AI-Media, Ortana, and Telos Alliance
A new model is only as useful as the workflows it slots into. That's why we anchored our booth around end-to-end live broadcast demos with three of the most respected names in broadcast tech.
AI-Media — live captioning and multilingual translation AI-Media's LEXI Voice platform powers live captioning and real-time AI translation for sports, corporate events, and news. With Dialogue RT running upstream of LEXI, broadcast teams get sharper transcription, more accurate multilingual translation, and cleaner output from messy live audio.
Ortana — broadcast media management and orchestration Ortana's Cubix platform handles media workflow orchestration across some of the world's most demanding broadcast environments. With AudioShake integrated into Ortana's automation flows, broadcast teams can apply dialogue isolation, music removal, and audio cleanup as standard processing steps, not as manual interventions.
Telos Alliance — broadcast audio infrastructure Telos Alliance is foundational broadcast audio infrastructure — the backbone of live audio for radio and TV operations globally. Demonstrating AudioShake inside the Telos signal chain is one of the clearest possible signals that AI audio separation has crossed the threshold from interesting technology to production-ready broadcast tooling.
Watch the full NAB 2026 demo reel
If you missed us in Vegas, the booth demo reel walks through everything we showed: dialogue isolation, music removal, multi-speaker separation, dubbing and localization, copyright compliance, and more.
Talk to us
If you're working on live broadcast, sports media, captioning, localization, or any workflow where clean audio is the bottleneck — let's keep the NAB conversation going.
Or read the full Dialogue RT announcement and ESPN partnership story.