AI Voice Tracking for RCS Zetta

Zetta runs the most demanding stations in the world, and it sets the standard for what a professional air chain should be. RadioAIVT adds the one thing no automation platform is meant to provide on its own: presentation for the hours nobody is presenting. It reads your scheduled hour, generates the breaks and deposits finished voice tracks into Zetta, placed against the music. Running today on a live Zetta system, with early access open to stations and groups.

Why a Zetta station expects more

Zetta is the platform serious broadcasters choose, and the stations that run it hold their air chain to a high standard. That standard is exactly why a generic solution will not do here. A text-to-speech box bolted on the side has no idea what GSelector scheduled, ignores your intros, and sounds nothing like your air staff. It would be a step down from everything else in the chain.

The other option, hiring voice trackers for every daypart in every market, is excellent where it is affordable and simply does not scale across a group.

What a Zetta operation deserves is presentation that meets the platform's own standard: generated from the actual log, processed like the rest of your air chain, and delivered into Zetta without anyone touching a workstation.

Voice tracks on Zetta: recording them, or generating them

Zetta already has a good voice track tool. Its Voice Tracker lets a presenter hear the end of one song and the start of the next, record over the ramp, and drop the voice track straight into the log — and Zetta2GO puts a Voice Track Editor in a browser so it can be done from anywhere. If you are learning how to do voice tracks on Zetta, that is the right place to start, and RCS's own tutorials cover it well.

What that tool cannot change is the arithmetic. Every voice track on Zetta still has to be recorded by a person, for every break, for every hour you want hosted. A four-hour show at four links an hour is 80 voice tracks a week. Cover mornings and evenings across a group and you are into the thousands per month — and a presenter still has to sit down and record each one.

That is where generating them instead of recording them starts to make sense. RadioAIVT reads the same hour your log already contains, writes and voices each break, and deposits the finished voice track into Zetta in the slot where it would have been recorded. Same object, same place in the log — produced without anyone in the chair. Your live presenters keep the voice tracks that matter; the system takes the hours nobody was going to record anyway.

How the integration works

1. The bridge runs as a Windows service. On the Zetta machine, started automatically, with no window to keep open and no operator involved. It survives reboots.
2. Your scheduled hour is read. The music your scheduler has placed for that hour — artists, titles, order, intro lengths — is exported and analysed before anything is written.
3. Each break is written for that hour. Real presentation about the songs actually on the log, with editorial content behind it, not a liner from a pool.
4. The voice is rendered and processed. Your chosen presenter voice, then SOUND4 Big Voice.CL broadcast processing on every track — so it sits in the mix like your other voices.
5. The track is deposited into Zetta. Placed in the right slot, fitted to the available space, ready to air. Nothing to import by hand.

Fitted to the ramp, not dropped on top

On a professional air chain, placement is everything. A break that spills into the vocal or leaves a hole is noticed immediately by people who have spent their careers listening to this.

RadioAIVT measures the space it has — the intro of the song, the slot available between elements — and fits the talk-up to it. The script length is adjusted to the time available rather than the other way round, so the break finishes where it should.

Your air staff, around the clock

Twenty-plus built-in presenter personalities are available, assignable per show and per daypart. You can also clone your own authorised talent so your station voices cover the hours they cannot physically be there. More on how that is built: AI on-air personalities.

What gets said

• The music — what just ended, what is next, how they connect.
• Artist stories — facts and anecdotes on the artists in your rotation.
• Concert dates — real upcoming shows for artists you are playing.
• Today in history — the date's anniversaries, used as transitions.
• Weather and local news — per market, depending on plan.

On Premium plans, every script then goes through an editorial layer that rewrites it the way a host speaks, while leaving every fact from the log untouched.

Where the Zetta integration stands

The Zetta chain is running in production on a live system: service, scheduling, generation, deposit and monitoring. We are opening it to a small number of stations and groups first, so each installation gets direct attention rather than a download link.

If you run Zetta and want to hear what your own hours would sound like, get in touch and we will set it up on your log.

Common questions

Does it change anything in my Zetta configuration?

Your clocks, categories and scheduling rules stay as they are. The integration reads the hour and delivers audio into it.

Does someone have to be at the machine?

No. The bridge runs as a Windows service with automatic start, and the chain is monitored from a dedicated status panel.

Can each market have its own voices?

Yes. Each station runs in its own isolated pipeline, with its own voices, vocabulary and editorial configuration.

What about pricing for a group?

Multi-station and FM group deployments are quoted case by case, around the integration and support you need. Talk to us.

Which other systems are supported?

StationPlaylist and RadioDJ are live. WinMedia and mAirList are in progress. Background on the technology: AI voice tracking for radio.