A Hands-On HeyGen Walkthrough
Create an AI avatar video from a script, clone your own voice, and produce multilingual content — full production walkthrough.
A Hands-On HeyGen Walkthrough
HeyGen creates AI-powered avatar videos — type a script, pick an avatar, get a video of a realistic human presenter speaking your words. This walkthrough covers creating your first video, voice cloning, and producing multilingual versions for global audiences.
Step 1: Create Your First Avatar Video
- Go to heygen.com and sign in (free trial includes 1 credit)
- Click Create Video → choose “Template” or “Start from Scratch”
- Pick an aspect ratio: 16:9 for YouTube/presentations, 9:16 for social media, 1:1 for posts
- Select an avatar from the library (100+ options: different genders, ethnicities, attire, settings)
- Type or paste your script into the text box (left panel)
- Select a voice for the avatar (50+ AI voices across languages)
- Click Submit — processing takes 2-5 minutes depending on script length
Script tip: Write like you speak, not like you write. Short sentences, contractions, conversational tone. A 150-word script produces ~60 seconds of video. Aim for 60-90 seconds for social, 3-5 minutes for presentations.
Step 2: Customize Avatar Appearance and Scene
HeyGen lets you control more than just the avatar:
- Background: Choose from stock backgrounds, solid colors, or upload your own image/video as a backdrop
- Avatar position: Drag the avatar to left, center, or right third of the frame
- Avatar size: Scale the avatar up (fills more of the frame) or down (more background visible)
- Outfit: Some avatars allow outfit changes (business casual, formal, casual)
- Gestures: Add natural gestures (hand movements, head nods) at specific points in the script. Click a word → add a gesture marker.
For professional presentations, a neutral background (solid color or office setting) with the avatar in the center-bottom third works best. For social media, tighter framing and dynamic backgrounds perform better.
Step 3: Clone Your Own Voice
Voice cloning makes avatar videos sound like you:
- Go to Voices → Voice Cloning
- Record a 2-minute sample: read the provided script in your natural speaking voice
- Upload the recording (MP3 or WAV, clean audio with no background noise)
- Wait for processing (~10 minutes)
- Your cloned voice appears in the voice selection dropdown
Quality tips for the sample:
- Use a decent microphone (USB mic minimum, not laptop mic)
- Record in a quiet room
- Read naturally — don’t over-enunciate
- The 2-minute sample covers enough phonemes for HeyGen to model your voice
Once cloned, select your voice for any avatar. The avatar speaks with your voice and your lip-sync matches. This is powerful for personal branding — your audience sees a consistent presenter even when you’re not recording.
Step 4: Add Multi-Language Translation
HeyGen’s standout feature — translate your video into 40+ languages with lip-sync:
- Create your video in English first
- After generating, click Translate
- Select target languages (Spanish, French, German, Japanese, Hindi, etc.)
- HeyGen translates the script and regenerates the avatar with lip-synced speech in each language
- Each language generates as a separate video file
Use cases:
- YouTube creators: produce the same video in English + Spanish to double addressable audience
- E-learning: one script → 5 language versions without re-recording
- Marketing: localized product demos for regional markets
Quality note: Translation accuracy is good for conversational content (~90%+). Technical jargon and brand names may need manual script correction before generating. Always review the translated script before generating the video.
Step 5: Use Your Own Photo as an Avatar (Photo Avatar)
Beyond stock avatars, HeyGen can animate your photo:
- Go to Avatars → Photo Avatar
- Upload a front-facing photo (high resolution, neutral expression, looking at camera)
- Crop to the avatar frame
- Select a voice (your cloned voice or an AI voice)
- Type your script and generate
The photo animates — eyes blink, mouth moves in sync with speech, subtle head movements. It’s not as fluid as a full video avatar, but it’s more personal and faster to set up.
Photo requirements:
- Front-facing, looking directly at camera
- Neutral or slight smile expression
- Even lighting, no harsh shadows
- Resolution: minimum 512×512px, recommended 1080p+
- No glasses (causes reflection artifacts in animation)
Step 6: Add Visual Elements and Captions
Enhance your video with overlays:
- In the editor, click Elements to add:
- Text overlays: Bullet points, titles, callouts (appear and disappear at specific timestamps)
- Images: Product screenshots, logos, charts (position anywhere on screen)
- Stock video: Background video clips from HeyGen’s library
- Click Captions → enable auto-subtitles
- Choose a caption style (font, size, color, position)
- Captions auto-generate from the script and sync to the avatar’s speech
For social media: Always enable captions. 80%+ of social video is watched muted. Captions increase watch time by 40%+ on average.
Step 7: Export and Distribute
- Click Export → choose resolution:
- 720p: faster export, smaller file (fine for social media)
- 1080p: standard quality (recommended for YouTube and presentations)
- 4K: highest quality (Enterprise plan, for premium content)
- Download the MP4 file
- Upload to your platform (YouTube, TikTok, LinkedIn, LMS, website)
Distribution strategy:
- YouTube: 1080p, 16:9, with custom thumbnail (generate one with Canva/Midjourney)
- TikTok/Reels: 9:16, 60-90 seconds, captions enabled
- LinkedIn: 1:1 or 16:9, professional avatar, B2B script tone
- Website: embed directly using HeyGen’s share link or self-host the MP4
What This Workflow Produces
| Output | Time |
|---|---|
| First avatar video | ~5 min (2 min processing) |
| Voice clone setup | ~15 min (10 min processing) |
| Multilingual versions | ~3 min per language |
| Captioned final video | ~2 min |
| Total for 1 video + 3 languages | ~30 min |
Same output with traditional video production: 2-4 hours (filming, editing, translation, captioning).
Common Mistakes to Avoid
- Scripts that are too long — AI avatar videos lose viewer attention faster than human recordings. Keep social videos under 90 seconds. Break longer content into a series.
- Ignoring gesture markers — an avatar standing motionless for 2 minutes looks uncanny. Add a gesture every 15-20 seconds.
- Poor voice clone sample — if your 2-minute sample has background noise or echo, every future video inherits that quality. Record it properly once.
- Not reviewing translated scripts — auto-translation can produce awkward phrasing. Read through the translated script before generating, especially for technical or cultural content.
- Using default backgrounds — the stock office background screams “AI video.” Customize with your brand colors or a relevant image.
Pro Tips
- Batch script writing — write 5-10 scripts in one sitting, then generate all videos in a batch. Context switching between writing and generating wastes time.
- Use custom avatars for brand consistency — if you produce a video series, use the same avatar + background + voice for every episode. Consistency builds recognition.
- A/B test avatars — for marketing videos, generate two versions with different avatars and see which performs better. Audience response to avatars varies by niche.
- Use HeyGen’s templates — pre-built templates for product demos, explainers, and social posts have optimized layouts and element timing. Start from a template, then customize.
- Embed on your website — HeyGen provides embed codes. An AI avatar video on your landing page can increase conversion rates by 20-40% compared to static text.