Voice Tracks

Voice tracking is how a station gets the sound of a live DJ without actually having someone in the studio around the clock. A DJ records their breaks in advance — intro, outro, talk-up, whatever — and AutoCast plays those recordings at the right moments in the broadcast, synchronized with music underneath. Done well, it sounds completely natural and keeps the station sounding staffed even during hours when nobody's physically there.

What AutoCast Does (and Doesn't Do)

AutoCast's job is playback. It takes the voice track recordings you've created and plays them at exactly the right moment, with exactly the right music underneath, with professional-sounding fades in and out.

AutoCast does not record voice tracks. Recording is done in a standard wave editor — any audio software you're comfortable with works fine. The DJ records their break, saves it as an audio file, and drops it into the appropriate folder in the Station Folder. AutoCast picks it up from there.

How Voice Track Playback Works

A voice track in the program log is written as a three-line block: the # VT command on one line, the voice track audio file on the next, and the song it introduces on the one after that. The song that follows the voice track is the one the DJ is talking up — the song that rises under the last few words of their break.

When AutoCast encounters this block, here's what happens:

1. The voice track recording begins playing on its own audio channel.
2. If a talkbed is configured on the # VT command, AutoCast selects one from the music library and starts it playing at the Talkbed Ducking Level (default 50%) — a musical bed underneath the DJ's voice.
3. Looking ahead at the introduced song's Ramp point (the moment the song's vocals begin), AutoCast starts that song quietly underneath the voice track — at the Ramp Ducking Level (default 50%) — timed so the song's vocals arrive just as the DJ wraps up.
4. During a short crossfade (the Talkbed Crossfade period, default 3 seconds), the talkbed fades out while the song fades in at ducked volume.
5. As the voice track nears its end, the song begins fading up to full volume — timed so the song is at full volume just before its vocals start.
6. The voice track recording ends, the song is at full volume, and broadcast continues normally.

The result: the DJ's voice rides over the talkbed, the next song rises under the last few words of their break, and everything flows without dead air or awkward gaps.

The # VT Command

# VT Comment TalkBed

The # VT command's two parts — the attribute name (Comment in this example) and the value (TalkBed) — tell AutoCast which music to use as the talkbed. AutoCast searches the music library for files whose specified attribute matches the specified value, then selects one using the Rotation Tracker.

The Rotation Tracker cycles through all matching files in sequence, ensuring you don't hear the same talkbed twice in a row. It remembers where it left off between sessions — even if you restart AutoCast, it picks up the rotation where it stopped.

If you use # VT with no attribute/value — just the command alone — the voice track plays without a talkbed. The DJ's voice goes out over the song coming in underneath, with no separate music bed under the break itself. This works fine, but a talkbed usually sounds warmer.

The voice track audio file itself is always specified by the line immediately following the # VT command in the program log, and the song being introduced is the line after that. AutoCast doesn't "look up" the voice track file — it plays exactly the file the program log names.

The Song's Ramp Point — The Key to Good Voice Track Timing

For voice tracks to sound their best, AutoCast needs to know when the introduced song's vocals begin — so the DJ finishes talking just before the singer starts. This information comes from the Ramp point set in the music library for the song being introduced (not the talkbed).

The Ramp point is the moment in the song where its vocals begin, measured in seconds from the start of the file. AutoCast uses it to time the voice track against the song: the voice track's end is placed so the song's full-volume vocals land exactly where you'd want them to — right after the DJ's last word.

If the song being introduced doesn't have a Ramp point set, AutoCast falls back to a simpler behavior: the voice track plays, and when it ends, the song starts at full volume. It works, but you lose the lovely crossfade-to-vocal timing — so it's worth spending a few minutes in Librarian to set Ramp points on songs you expect to be voice-tracked over.

Ramp points (and the other transition values AutoCast uses — EOM and Cue) are set from Librarian's Transitions tab. See the Librarian documentation for details on how to set and fine-tune them.

Voice Track Timing Scenarios

AutoCast adapts to whatever timing combination the program log throws at it:

Voice track fits within the song's ramp — The common case. The voice track plays; the song starts quietly partway through the voice track and builds up; the song's vocals arrive right on cue as the DJ wraps up. This is how professional voice-tracked radio sounds.

Voice track is longer than the song's ramp — AutoCast detects this and adjusts: the song starts as soon as the voice track begins, ducked under the talkbed or under the DJ's voice, and stays ducked until the voice track ends. The timing still works; you just lose a little of the "music rises into vocals" polish.

Song has no ramp point — Simplest case. The voice track plays to its end, then the song starts at full volume. If a talkbed was configured, it fades out as the song comes in.

No talkbed configured — The voice track plays on its own. The song still comes in underneath the voice track's end for the crossfade-to-vocals effect; only the musical bed under the DJ's voice is absent.

Voice Track Timing Settings

All voice track timing is configured in Settings → Voice Tracks:

Setting	Default	What It Controls
Ramp Ducking Level	50%	How loud the introduced song plays underneath the voice track during its ramp — before the song fades up to full volume. 50% = half volume underneath the DJ's voice.
Talkbed Ducking Level	50%	How loud the talkbed bed plays while the voice track is speaking. Independent of Ramp Ducking Level — you can have the talkbed quieter than the ducked song, or vice versa, depending on how you like the mix to sound.
Talkbed Crossfade	3.0 sec	How long the talkbed takes to fade out while the introduced song simultaneously fades in. Both fades run concurrently during this window.
Fade Up Duration	1.0 sec	How long the introduced song takes to rise from its ducked level to full volume after the voice track ends.
Fade Up Early	0.75 sec	How many seconds before the voice track ends to begin fading the song up. Starting early creates the natural "music rises under the last words" effect.
No-Ramp Talkbed Fade	2.0 sec	Fallback talkbed fade duration used when the introduced song has no Ramp point — applied as the talkbed fades out at the voice track's end.

# VTR — Precision Voice Tracks from VoiceTracker

There is a second kind of voice track with a different command and a more sophisticated playback path: # VTR. VTR stands for "remote voice track" — voice tracks recorded in the VoiceTracker browser application, where the DJ actually hears the transition between the outgoing and incoming songs and positions their voice visually over the waveforms.

Because the DJ records while listening to both songs, the exact timing of voice against music is decided at record time. VoiceTracker passes that timing along to AutoCast behind the scenes, and AutoCast uses it to overlap the voice track smoothly with the outgoing and incoming songs exactly as the DJ positioned them. No library lookups or station settings are involved in the overlap timing itself — it's a closed loop between VoiceTracker and AutoCast.

# VTR
Station/Music/SongA.mp3
Station/VoiceTracks/VTA_042326_18.mp3
Station/Music/SongB.mp3

A # VTR block is four lines: the command, the outgoing song, the voice track file, and the incoming song. The extra line versus # VT is what lets AutoCast duck the outgoing song early — a level of control # VT doesn't need, because a plain # VT simply lets the outgoing song end at its normal EOM.

How # VT and # VTR Differ

	# VT	# VTR
Recorded in	Any audio editor	VoiceTracker browser application
DJ hears the real transition while recording?	No	Yes
Program log lines	3 (command + voice track + incoming song)	4 (command + outgoing + voice track + incoming)
Timing source	Incoming song's Ramp point + station ducking settings	Pre-calculated in VoiceTracker at record time
Outgoing song handling	Ends normally at its own EOM	Ducks early under the voice track, per the DJ's positioning
Talkbed support	Yes, via the # VT command's attribute/value	Not applicable — the DJ already positioned their voice directly over the real songs
Best for	Quick breaks, airstaff drops, voiced announcements	Polished song-to-song transitions with precise voice placement

The two formats coexist freely. A single program log can mix # VT and # VTR blocks in any order; AutoCast handles each according to its own rules without any configuration on your part.

Organizing Voice Track Files

Voice track recordings are typically stored in a folder within your Station Folder — many stations use a subfolder of Programs or create a dedicated VoiceTracks folder. The exact location depends on how your program logs are set up. The key requirement is that AutoCast can find the file at the path specified by the line following the # VT or # VTR command.

If you're using ClockWork and TuneStacker to generate your logs, the voice track file paths will be written into the log automatically based on your scheduling configuration. Files recorded in VoiceTracker land in a shared Dropbox folder and are synced to AutoCast automatically — no manual file movement required.

Next up: Commercial Breaks & Traffic →