Update guide • Tested Feb 2025

HeyGen 2025: translator v3 lip-sync, multi-avatar scenes, safer voice import

We ran the 2025 HeyGen build on real scripts (product demos, onboarding, multilingual explainers). Here’s what changed, what to toggle, and how to pair HeyGen with external assets without breaking compliance.

· ≈12 min read

What’s new

  • Translator v3: tighter lip-sync on plosive sounds, better timing on short hooks (15–30s).
  • Multi-avatar scenes: two avatars per canvas with split layouts (A/B frames) and over-the-shoulder variants.
  • Voice import checks: auto-rejects low-SNR uploads; watermark toggle remains on by default.
  • Template refresh: more corporate scenes (meeting, product table, app UI) with matching camera angles.
  • Faster renders: short clips (<90s) render 15–20% faster in our tests vs Q4 2024.

Test results (real scripts)

  • Product demo 2 min: lip-sync accuracy 9/10 on EN/FR/ES; needed 1 manual retime on DE plosives.
  • Onboarding 60s: multi-avatar split-screen worked; keep text under 38 words/scene to avoid subtitle overflow.
  • Localization: translator kept brand names intact; captions regenerated per language with correct line breaks.
  • Audio: music ducking improved; if using external VO, normalize to -16 LUFS before upload.

Recommended workflow

  1. Write a 3–5 beat outline; choose Translator if you need multilingual, otherwise Script to video.
  2. Select layout: split (2 avatars) for Q&A, picture-in-picture for demos, single for storytelling.
  3. Keep subtitles to 2 lines max; enable “auto resize captions”.
  4. Export draft at 1080p; final at 4K once visuals are locked.
  5. For higher realism, source hero shots from a photoreal generator (Akool) then import into HeyGen as cutaways.

Pricing & limits (Feb 2025)

  • Free: watermark, limited credits, translator available with caps.
  • Creator: higher minute pack, most avatars, 4K export.
  • Business: team workspaces, custom avatars, priority rendering, enterprise consent controls.

Notable change vs 2024: translator minutes are pooled per workspace, and custom avatar approvals are reviewed faster.

Compliance & rights

  • Only upload voices you own the rights to; watermark is on by default for transparency.
  • For EU audiences, add a short disclosure below the player when using synthetic voices.
  • Keep logo safe-zone when using multi-avatar frames to avoid overlap.

Pairing with other tools

For product explainers, we generate 2–3 photoreal stills on a dedicated visual generator, then weave them into HeyGen timelines. It keeps shots consistent and reduces the number of retakes.

FAQ

  • Is 4K available on all plans? No, Creator and above.
  • Can I use external TTS? Yes, upload WAV/MP3; normalize to -16 LUFS and leave -3 dB headroom.
  • Does multi-avatar double the cost? No, minutes are counted on final video length, not per avatar.