How It Works
How It Works Plans ← Back to app

How It Works

Draw your mouth shapes, upload your audio, and get a frame-perfect lip-sync video — in any language, including gibberish and singing.

Quick cheatsheet

Premium is coming

No ads, longer exports, transparent MOV, session restore, saved characters, and more — Premium plans coming soon.

🔔 Notify me when it's live
↓ Jump to Common problems & fixes
🌍 EN HE AR ES
First — before anything else
Choose your language
  • Select the language your character speaks from the dropdown in Step 01 of the app.
  • This determines which phoneme set Allosaurus uses — and which filenames the tool expects.
  • Don't have your language? Choose Universal (IPA) — it works for any language, gibberish, and singing.
🌍 Universal 🇺🇸 English 🇮🇱 Hebrew 🇸🇦 Arabic 🇪🇸 Spanish
Step 01
Name & place your mouth PNGs
  • PNG or WebP with transparent background (alpha channel).
  • Name each file after the sound it represents — the app reads the filename to auto-assign phonemes.
  • Place the mouth at its actual position on the character canvas — not centered. It drops straight into your composite.
Universal filenames:
ahoheeoommppffsschnnddrrwwrest
⚠ Always include rest.png — shown during silence and between words.
Step 02
Upload clean audio
  • Use WAV for best phoneme accuracy, or MP3.
  • Free plan: 3 exports/day, 30 sec max. Longer exports coming soon with Premium.
  • Verify the server shows ✓ online before analyzing.
⚠ No audio effects whatsoever. Reverb, echo, EQ, compression, and pitch shifting confuse the phoneme detector and cause stuck or wrong mouth shapes. Record clean and dry.
Step 03
Review & edit the mapping
  • Every uploaded PNG appears as a card. Click any card to reassign which sounds it represents.
  • Multiple sounds can share one shape — e.g. map nn and dd to the same PNG.
  • Your mapping auto-saves and restores on your next visit — coming soon with Premium.
Step 04
Analyze & preview
  • Pick your FPS (24 is standard for animation).
  • Click ⚡ Analyze audio — Allosaurus detects phonemes and maps them frame by frame.
  • Watch the colored timeline and play the preview to check sync.
  • If you change language or FPS, re-analyze before exporting.
Step 05
Export your animation
  • Green screen MP4 — solid #00FF00 background. Use Ultra Key in Resolve / Premiere, or Keylight in After Effects.
  • Resolution matches your PNG canvas exactly. Audio is baked in.
  • Transparent MOV export — coming soon with Premium.
⚠ Common problems & fixes
Mouth stuck on one shapeReverb or echo on audio. Export a clean dry recording with zero effects.
All frames show "rest"Recording volume is too low. Use a louder, cleaner take.
Wrong mouth for a soundClick the mouth card in Step 03 and reassign manually.
Missing shapes warningAdd the missing PNG or reassign that sound to an existing shape.
Server shows ✗ offlineRailway free tier may be sleeping. Wait 30 s and click "check" again.
Export fails or hangsTry a shorter clip or lower FPS (12 or 24).
Preview fine, export wrongAlways re-analyze after changing FPS or language.
Mouth misaligned in compositePNG must be positioned at the mouth's location on the full character canvas, not centered.