Voicetracking is an old radio craft. AI changes how a break gets written and voiced. It should not change who is responsible for the sound.
Start with voicetracking itself.
PlayIt Software describes voicetracking as an industry-standard technique that creates the impression of a live announcer when the audio is prerecorded and played out automatically, noting a three-hour show can be tracked in under half an hour (PlayIt Live docs). In practice the talent records only the links between elements, and each link lives at a specific spot in the log.
Every major automation system treats that link as a transition. RCS Zetta shows the previous asset, the voicetrack, and the next asset together. Rivendell uses track markers as placeholders and stores recordings in voicetrack carts. The craft is in the segue, not just the words.
What the AI part actually does.
AI voicetracking helps with two jobs: writing a break that fits its window, and voicing it. A good system reads the log first: the outgoing song, the incoming intro, the time, and the station's own vocabulary. Then it produces a read sized to the available space, so it lands before the vocal instead of on top of it.
The second job is the voice. Here there are two very different approaches. Some products use synthetic voice engines to create an announcer that never existed. RoboJock trains professional cloned voices from real radio voicetracks recorded by actual radio professionals. The difference shows up in delivery: a seasoned jock's timing, warmth, and restraint come from years of real breaks.
Where people stay in charge.
A convincing take is not the same as an approved one. Programmers and talent should be able to change wording, fix pronunciation, adjust pace and energy, and reject a read before it airs. After delivery, the station's own tools still apply: trims, segue edits, and auditioning the hour in sequence.
The direction stays with professionals.
How a break gets to air.
The handoff matters as much as the read. Finished breaks should arrive in the format the automation system expects, with identifying details like cart or cut numbers, log position, and segue marks. RoboJock delivers into the systems below, and to Dropbox, S3, or SFTP drop folders for anything else.
Questions every station should ask.
Whose performances trained the voice, and what permission covers your use? Who can change or stop that use? Who approves each break? How does a finished break enter your log? The U.S. Copyright Office's report on digital replicas found gaps in existing protections and recommended new federal law, and broadcasters have backed legislation against unauthorized voice recreations. Partnership on AI's framework centers consent and disclosure. None of this is legal advice; settle disclosure policy with your own counsel.
Frequently asked questions.
What is voicetracking?
Voicetracking is recording only the talk between songs so a prerecorded show plays out in automation as if an announcer were live. The break is placed in the log between specific elements.
What is AI voicetracking?
AI voicetracking uses software to help write and voice those breaks from the station's log and context. The quality depends on the voice source, the context the system reads, and how much control the station keeps.
Are RoboJock voices artificial?
No. RoboJock voices are professional cloned voices trained on real radio voicetracks from actual radio professionals. They are not invented announcers.
Does AI voicetracking replace radio talent?
RoboJock is built to work with talent and programmers. The voice comes from a professional, and the station team directs wording, pronunciation, pace, and approval.
Which automation systems does it work with?
RoboJock delivers finished breaks for RCS Zetta, WideOrbit, AudioVAULT, Rivendell, and ENCO DAD, plus Dropbox, S3, or SFTP drop folders for other systems.