Upload your own avatar
Choose a preset actor or upload your own clear avatar image (JPG, PNG, WebP, or AVIF up to 10 MB).
Generate speech directly from text. Token cost is based on estimated speech duration.
Choose a Voice
Higher resolution can take longer to generate.
Sign in to spend tokens
Choose an Avatar is required
Your generated video will appear here
HeyGen Avatar 4 turns any avatar image plus a script or voice recording into a talking AI avatar video — billed by duration at 1 token per 5 seconds of speech. On Prizmad it's built in: upload a face or pick a preset actor, add audio, generate.
HeyGen Avatar 4 is an image-to-talking-video model: give it one clear avatar image and something to say, and it renders a lip-synced presenter video. Unlike avatar tools locked to a fixed cast, Avatar 4 works with any face — choose a preset actor from the picker or upload your own image (JPG, PNG, WebP, or AVIF up to 10 MB) to turn a founder photo, a brand mascot shot, or an AI-generated character into a spokesperson.
Audio comes from either of two modes. Generate from Text sends your script with one of the built-in HeyGen voices and creates the speech for you. Upload Audio takes a clean voice recording of up to 60 seconds and lip-syncs the avatar to it precisely. Either way you control the output shape — six aspect ratio options including 9:16, 4:5, and 1:1, resolutions from 360p to 1080p — plus a Talking Style setting: Stable keeps motion subtle, Expressive adds more avatar movement.
On Prizmad, HeyGen Avatar 4 uses the same token balance as , video, voiceover, , and music. No HeyGen account, no API key — you pay in the same tokens as every other tool, your videos land in the same asset library, and everything you generate is yours to use commercially.
Choose an Avatar
9:16 presenter clips with expressive delivery — the talking-head format that carries most winning dropshipping and DTC ads.
Upload a photo of yourself or your brand's presenter once, then generate new spokesperson videos any time without re-filming.
At 1 token per 5 seconds, you can generate ten different 5-second hooks for 10 tokens and let the ad account decide the winner.
Animate a preset actor or your own uploaded image — JPG, PNG, WebP, or AVIF up to 10 MB. A clear, front-facing photo gives the best result.
Type a script and pick a HeyGen voice, or upload a clean voice recording up to 60 seconds for exact lip-sync to your own audio.
16:9, 9:16, 4:5, 5:4, 1:1, or Auto (keeps the source format when available) — cover Reels, feed, and YouTube from one tool.
Choose an avatar. Pick a preset actor from the gallery or upload your own clear avatar image (JPG, PNG, WebP, or AVIF up to 10 MB).
Choose your audio source: Generate from Text or Upload Audio.
In text mode, write the script and select a voice — token cost is estimated from the speech duration of your script. In audio mode, upload a clean voice recording up to 60 seconds; cost is based on the verified audio length.
Set the aspect ratio (16:9, 9:16, 4:5, 5:4, 1:1, or Auto) and resolution (360p up to 1080p).
Optionally open advanced settings to switch the Talking Style between Stable and Expressive, or set a solid hex background color.
Click Generate. The video renders in roughly 3–8 minutes and lands in your asset library, ready to download or use in a Prizmad project.
Avatar 4 bills purely by speech duration: 5 seconds = 1 token, 10 seconds = 2, 30 seconds = 6, and the 60-second maximum = 12 tokens. Text mode estimates duration from your script; audio mode uses the verified length of your upload.
HeyGen Avatar 4 is unlocked on the same token balance as image generation, video, voiceover, and music — no separate HeyGen account and no API key setup.
If your tokens run out mid-campaign, buy a one-off top-up directly from the top-up page. Every avatar video you generate stays yours with full commercial rights.
It runs on your Prizmad token balance — there's no separate charge for the tool and no HeyGen account required. Pricing starts at 1 token for a clip of up to 5 seconds.
1 token per 5 seconds of speech, rounded up: a 5-second clip is 1 token, 10 seconds is 2, 30 seconds is 6, and the 60-second maximum is 12 tokens. The tool shows the estimated cost before you generate.
Yes. Upload any clear avatar image — JPG, PNG, WebP, or AVIF up to 10 MB — or pick one of the preset actors. Front-facing, well-lit photos produce the cleanest lip-sync.
Published 2026-07-16 · Last updated 2026-07-16
30–60 second scripted walkthroughs of the product, offer, or guarantee for landing pages, retargeting, and email embeds.
Keep the same face, swap the script or uploaded voiceover, and produce market-specific versions of a proven creative.
Five resolution tiers from 360p to 1080p. Render low for quick script tests, full HD for the creatives you ship.
Stable keeps the avatar's motion subtle and newsreader-calm; Expressive adds more movement for energetic UGC-style delivery.
Billed at 1 token per 5 seconds of speech — a 5-second hook costs 1 token, a 30-second read 6 tokens, a full 60-second video 12 tokens.
Up to 60 seconds. Uploaded audio is accepted up to 60 seconds, and duration-based billing is clamped to that range — a full-length video costs 12 tokens.
Yes. Switch the audio source to Upload Audio and attach a clean voice recording up to 60 seconds — MP3, WAV, AAC, OGG, and several other formats are accepted. Cost is calculated from the verified audio duration.
Yes. You keep full commercial rights to every video generated on Prizmad — run them on TikTok, Meta, and YouTube, embed them on product pages, or use them in email campaigns.
Stable keeps the avatar's motion subtle — good for calm, authoritative reads. Expressive adds more avatar movement, which suits energetic UGC-style ad delivery. It's an advanced setting; Stable is the default.
Avatar 4 animates any image you upload, offers six aspect ratios up to 1080p, and bills from 1 token per 5 seconds — best for custom faces and cheap short-clip testing. Avatar V uses HeyGen's library of 1,200+ pre-built digital-twin avatars with output up to 4K, transparent WebM, and background removal, starting at 8 tokens. Both are on the same token balance, so test both.