How the voiceover pipeline works
Voiceover generation in TubeForge is per-scene and per-channel. Each scene has its own narration text (from the AI script or manually edited); clicking 'Generate voice' on a scene synthesises the MP3 for that scene. 'Generate all pending' runs the full script in one batch. The channel settings store the voice provider, voice ID, and speaking rate as a per-channel default — so every video on a channel uses the same voice automatically. Voice files are stored locally alongside your project; the render pipeline picks them up when composing the final video.
Voice options in TubeForge
- Grok Voice TTS (Eve, Ara, Rex, Sal, Leo) — 5 character voices via OpenRouter on your plan credits, pay per character on your plan credits.
- Edge-TTS — 400+ Microsoft voices, completely free, no key required, works offline.
- Rate control — speaking rate adjustable per channel (percentage faster or slower than default).
- Voice style override — per-scene voice style override for emphasis or character variation.
Animated captions synced offline to your voice
TubeForge's animated captions feature runs word-level timing alignment offline using a bundled faster-whisper model — no cloud upload, no waiting for a transcription API. Word-synced captions are then burned into the video in your chosen style (Karaoke, Word Pop, Typewriter, Fade-Up, or Bounce) at render time, staying perfectly aligned with the narration throughout.
How this compares to dedicated voice tools like Fliki
Dedicated voice-to-video tools like Fliki have an advantage in raw multilingual voice breadth — a large built-in catalogue of voices across many languages and accents in a browser. TubeForge's voice roster (Grok + Edge-TTS) is narrower in breadth but different in structure: generation runs at your plan's credit rate, with no voice access tied to a subscription tier. If multilingual reach across dozens of languages is your primary need, Fliki or a similar dedicated tool may serve you better. If you produce in English and want voice narration that's cost-transparent, integrated with the full production pipeline, and synchronized offline to animated captions, TubeForge is the stronger base.
Try TubeForge on your own machine
TubeForge is a local-first desktop app for Windows 10/11 and macOS 11+ (Apple Silicon & Intel). Bring no API keys, render on your own GPU with bundled FFmpeg, set a per-video spend cap with Profit Mode, and keep every project on your disk. Grab the installer below.
Free tier + plans from $9/mo · no API keys · install guide
Frequently asked questions
- Is TubeForge's AI voiceover generator free?
- Edge-TTS (400+ Microsoft voices) is completely free — no key, no plan needed, and it works offline. Grok Voice TTS (Eve, Ara, Rex, Sal, Leo) runs on your TubeForge credits. There is no per-minute subscription in the app — you pay only for what you generate.
- What voices are available?
- Free: Edge-TTS — 400+ Microsoft voices across many languages, no key required. Premium: Grok Voice TTS via OpenRouter — five character voices (Eve, Ara, Rex, Sal, Leo), billed per character. You can set the voice provider and voice ID per channel in Settings → Channel → Voice.
- Does TubeForge sync the voiceover to the video automatically?
- Yes. TubeForge generates a separate voice file per scene and then assembles them into a continuous audio track during the FFmpeg render. The narration track is always the primary audio; any background music is mixed in at a configurable volume (-18 dB by default) with optional ducking. Animated captions are aligned word-for-word to the voice track offline.
- Can I use my own pre-recorded voiceover instead of AI?
- Currently TubeForge's narration pipeline is AI-generated voice. If you have a pre-recorded narration track, you can bring it in by placing it in your project and selecting it as the voice file for a scene — the render pipeline will use it as the scene's audio instead of generating a new one.