Voiceover Direction and TTS Markup
Turns a script into a recordable VO read: exact timing, intent-based direction, marked stresses and pauses, and engine-correct TTS markup.
Before & after
Same prompt, same agent — the only difference is the skill.
I can help you think through voiceover direction. You'd want to consider the tone you're going for—maybe upbeat and energetic, or calm and authoritative? You should also think about pacing and where to add emphasis. If you can tell me about your brand voice and what the video is about, I could offer some suggestions. Feel free to share any reference videos or examples of styles you like, and I can build from there.
VOICEOVER DIRECTION - YouTube Intro (18 seconds) OVERALL TONE: Conversational urgency with warmth. Like you're sharing insider knowledge with a friend, not reading a script. PACING BREAKDOWN: • 00-03s: Moderate pace, warm. "So you want to scale your SaaS..." - Conversational. Slight emphasis on "want." • 03-08s: Accelerate slightly. Stack three quick problems (pain points) with energy. Pause 0.2s between each. • 08-12s: Drop pace. Slower, confident. "There's a better way." - Let this land. Emphasis on "better." • 12-18s: Return to conversational, build momentum. End on highest note of energy. EMPHASIS MARKS: • [STRESS] "scale" (00s), "actually" (06s), "better" (11s) • [SOFTEN] organizational names, numbers (let them sit neutral) • [BREATH] natural pause after each problem statement REFERENCE PACE: Think HubSpot intro + Alex Hormozi's clarity. Energetic, not frantic.
About this skill
Your AI voice sounds robotic, the read feels rushed, or the engine says "dollar sign" out loud and ignores your pause tags. The cause is usually one of three things: too many words for the real runtime, direction written as adjectives, or markup the engine can't parse. This skill fixes the math, then the text, then the performance, then the markup.
What it does
- Counts spoken words after expanding prices, years, and URLs, then checks them against the real VO window (runtime minus head and tail air and pauses), and cuts what doesn't fit
- Cuts in a set order: on-screen duplicates first, then throat-clearing openers and stacked adjectives, so the brand, offer, and CTA survive
- Writes a direction header built on who is speaking, to whom, and one action verb, instead of "upbeat and friendly"
- Marks one operative word per sentence, timed pauses, list and ending inflection, and pace shifts, with a pronunciation guide that flags names to confirm
- Converts the direction into the syntax your engine and model support (SSML, ElevenLabs break and audio tags, or an instructions field), with a no-tag fallback
- For voice actors: alternate reads, session notes, a pickups list, and file specs
What you get
A timing check with the math shown, the cuts, a direction header, a marked script, a pronunciation guide, and an engine-ready version or session notes for talent.
Best for
Creators, marketers, course builders, and producers who record with voice actors or generate VO with ElevenLabs, Polly, Google, Azure, or other TTS engines.
SKILL.md preview
---
name: voiceover-direction-tts-markup
… (sign up to view the full skill)More creator skills
View all Creator skills →Music Spotting Sheet with Tempo Math and Licensing
Spot music against your story beats: timed cues, tempo math so hits and endings land on the cut, levels under VO, and license checks.
Dialogue-Synced Sound Effects Placer
Places SFX from a word-level transcript: sync points, frame-accurate timecode, levels under dialogue, loudness targets, and license checks.
Explainer B-Roll Prompt Pack for AI Video
Turns a VO transcript into a style-locked AI video prompt pack: coverage map, P1-P3 clips timed to the VO, exact data routed to code.
Remotion Explainer Scenes Synced to Voiceover
Turns a timestamped voiceover into Remotion scenes: frame-exact timeline, one style module, exact charts and counters, and a sync check.