# AvatarSay Review Text-to-talking-character studio that generates fictional avatars and lip-synced MP4 clips with one-time credit packs, no subscription. ## Editor's Score: 68/100 Capability scores 17 because the core fictional-avatar-to-talking-clip pipeline works and the MCP integration is genuinely novel, but the 30-second clip ceiling, English-only voices, and 8 preset voice limit are hard constraints that HeyGen and Runway do not share, and there is no team, collaboration, or export feature beyond a single MP4 download. Ease of use scores 20 because the credit cost table is fully transparent, onboarding requires no card, and the MCP setup is zero-install, though the async polling model adds friction for casual users. Value scores 20 because the pay-once, never-expire credit model at $9.90 entry is genuinely cheaper than any monthly subscription competitor for low-volume use, and the 14-day refund policy reduces purchase risk. Delivery scores 11 because no independent user reviews exist to corroborate output quality claims, render times of 3 to 12 minutes per short clip are slow, the only available benchmark data comes from what appears to be the vendor's own DEV Community post rather than third-party testing, and the tool is newly launched with no track record to evaluate. ## Pros - Genuinely subscription-free: one-time credit packs never expire and unused packs are refundable within 14 days, a billing model almost no competitor in this category offers. - Fictional-only generation with no photo upload requirement removes consent and deepfake risk entirely, and the pricing page lists every credit cost with no hidden multipliers. - MCP server at avatarsay.com/mcp connects to Claude and Cursor via Streamable HTTP with zero local install, making it one of the few talking-avatar tools with a working agent integration. ## Cons - Render times of 3 to 12 minutes per talking clip are slow even by AI video standards, and the async polling model means you cannot watch progress in real time. - Only 8 preset English voices with no voice cloning, no custom voice upload, and no language support beyond English, which is a hard ceiling for any multilingual or branded-voice use case. - The free tier's 30 credits cover roughly 7 standard avatar generations or a single 2-second talking clip at minimum length, which is thin for evaluating the video quality before committing to a paid pack. ## Pricing Freemium. A free tier is available. - Free: Free - Starter: $9.90 - Creator: $24.90 - Studio: $59.00 Pricing reflects what we saw at the time of review (2026-11). Confirm current pricing on the official site. Category: AI Video Tools (https://aitoolseekers.com/category/ai-video-tools) Official site: https://avatarsay.com/ Last verified by a human: 2026-11-16 Reviewed by a human at AI Tool Seekers, a hand-reviewed AI tools directory. Every listed tool is verified by a person before publishing. Scores are editorial opinions from a fixed rubric (capability, ease of use, value, delivery on promise) and cannot be bought; there is no pay-to-rank. Directory: https://aitoolseekers.com This review: https://aitoolseekers.com/tools/avatarsay Methodology: https://aitoolseekers.com/how-verification-works ## Verdict AvatarSay is the clearest option for creators who need fictional talking characters without uploading a real face, and the pay-once credit model is genuinely rare in this category. The tradeoff is real: video renders take 3 to 12 minutes, output tops out at 30 seconds per clip, and only 8 preset English voices are available, which puts it behind HeyGen for professional or multilingual production work. If you need human-likeness, real-face lip-sync, or anything beyond short English clips, look elsewhere. Best for: Indie creators, tabletop RPG designers, and faceless-channel operators who need fictional talking characters without a subscription. Not for: Teams needing real-face avatars, multilingual voice output, or clips longer than 30 seconds. ## Overview AvatarSay is a browser-based tool that takes a text prompt and produces two 1024x1024 fictional character portraits, then optionally animates one of them into a lip-synced MP4 up to 30 seconds long. The whole pipeline runs without a photo upload or face swap: every face is generated from scratch using FLUX.1 models from Black Forest Labs, and voice is handled by MiniMax Speech 2.8 Turbo with Kling AI Avatar for the animation layer. That fictional-only constraint is the product's defining choice, and it shapes everything from the trust model to the use cases it fits. The credit system is pay-once and genuinely subscription-free. Standard avatar sets cost 4 credits, HD sets cost 6, and talking video costs 12 credits per second with a 5-second minimum. At the Creator pack rate of $0.0166 per credit, a 10-second talking clip runs about $2. Failed jobs refund credits automatically, and the pricing page lists every action that spends credits with no hidden multipliers. That transparency is a real differentiator against tools that bury per-minute or per-seat costs in fine print. The MCP server at avatarsay.com/mcp is a notable addition for developer workflows. It connects directly via Streamable HTTP to Claude and Cursor with no local runtime install, letting agents generate avatars and queue talking clips programmatically. The async job model returns a jobId immediately and expects polling every 20 to 30 seconds, which fits agent workflows but means render waits of 3 to 12 minutes are baked into the experience. What AvatarSay does not do is equally important to understand. Eight preset English voices only, no voice cloning, no custom voice upload, no clips beyond 30 seconds, and no team or collaboration features. The tool is narrow by design, and buyers who need breadth will hit those walls fast. ## Who should use it AvatarSay fits a specific kind of creator: someone building a faceless YouTube channel who wants a consistent cartoon mascot, a tabletop RPG designer who needs voiced NPC portraits for a campaign, or an indie game developer prototyping character dialogue without a budget for voice actors or 3D artists. The pay-once model also suits anyone who produces characters in bursts rather than on a monthly schedule, since credits sit unused without penalty. Creators who need real-face lip-sync, multilingual output, or clips longer than 30 seconds should look at HeyGen instead. HeyGen supports dozens of languages, custom AI avatars built from real video, and much longer output, making it the stronger choice for professional marketing or corporate video production. ## FAQ Q: Do AvatarSay credits expire? A: No. Credits from any paid pack never expire. Unused packs are also refundable within 14 days of purchase. Q: Can I use my own photo or voice with AvatarSay? A: No. AvatarSay generates fictional characters from text only. There is no photo upload, face swap, or voice cloning at any tier. All faces are AI-generated from scratch. Q: How long does a talking video take to render? A: Based on available benchmarks, render times range from roughly 3 minutes 32 seconds for a 3-second clip to over 12 minutes for a 6-second clip. The system returns a jobId immediately and requires polling to retrieve the finished MP4. Q: What is the maximum talking video length? A: 30 seconds per clip, with a minimum of 5 seconds. At 12 credits per second, a 30-second clip costs 360 credits. Q: Can I use AvatarSay inside Claude or Cursor? A: Yes. AvatarSay provides a remote MCP server at avatarsay.com/mcp that connects via Streamable HTTP with no local install. API keys are managed from the dashboard. Q: What languages does AvatarSay support for talking video? A: Currently only English, via 8 preset voices. There is no multilingual support, no voice cloning, and no custom voice upload option.