Hyper-realistic realtime avatars for scalable human presence, made by @HeyGen
San Francisco
Joined October 2025
- Tweets219
- Following18
- Followers1.2K
- Likes202
Pinned Tweet
30 new LiveAvatars just dropped.
6 emotions each. Full body. More natural expressions.
Free to use, live now 👇
LiveAvatar speech recognition just got a lot more multilingual 🌍
• 32 languages recognized, up from 6 (AssemblyAI STT)
• Switch to Japanese, Korean or Chinese mid-conversation
• Our sales agent understands you in more languages
Try it in yours → liveavatar.com/ai-sales-agen…
Embeds built on a LiveAvatar voice agent now use the agent's exact voice setup 🎙️
• Same TTS provider, model & speed you picked
• Same speech-to-text provider
• Clear errors for unsupported language picks
Try it → app.liveavatar.com
Cartesia is now the default voice engine on LiveAvatar 🎙️
• Picked automatically for any voice that supports it
• Prefer ElevenLabs? Pin it in voice_settings
• SSO sign-in now opens your team workspace
Try it → app.liveavatar.com
Custom avatar training failed? Now you'll always know why 🛠️
• Failure reasons are saved even when details arrive late
• Retries start clean, no leftover errors
• API returns the reason in error_message
Fix your footage, try again → app.liveavatar.com
OpenAI GPT Live and Realtime are now voice agents in LiveAvatar - API and web app, on your own OpenAI key.
• Pick model, voice and context
• Works in sessions and embeds
• Steadier GPT-Live playback
Try it → app.liveavatar.com
New in LiveAvatar:
• voice_engines per voice, selectable in the UI
• GPT Live (gpt-live-1) for realtime
• ElevenLabs v4 (eleven_v4_turbo)
Heads up: openai_realtime_config.model is now a fixed list - gpt-4o-realtime-preview no longer validates.
→ app.liveavatar.com
Ask anyone what the hardest part of learning a language was. Most will say speaking, because it always needed a person on the other side.
That just changed. Here's what it means for learners, teachers, schools and the teams building for them.
New voices just landed on LiveAvatar 🎙️
• Expanded voice vendor lineup
• New models incl. Sonic 3.6 + S2 Pro family
• More accents, tones + styles out of the box
Pick your avatar's voice → app.liveavatar.com
ElevenLabs V3 just landed on LiveAvatar 🎙️
• eleven_v3 + eleven_v3_conversational in FULL mode
• EU/US residency auto-detected for your ElevenLabs key
• Space invite links fixed
Build an interactive avatar in a few minutes → app.liveavatar.com
Bring your own ElevenLabs key? It just got smoother.
• Auto-detects your key's region (US/EU/IN/SG) and routes there
• Residency keys work with no extra setup
• Voice imports now consistent across ElevenLabs, Fish Audio & Cartesia
Try it → app.liveavatar.com
Two backend improvements shipped for LiveAvatar:
• Voice data caching - faster voice loading during avatar setup
• Fixed a race condition causing intermittent "bad handshake" WebSocket errors on session start
Smoother sessions, less waiting → app.liveavatar.com
LiveAvatar retweeted
Built a Gemini Live adaptation of this @HeyGen liveavatar demo.
I replaced the OpenAI realtime layer with @GeminiApp Live, while keeping the @TryLiveAvatar + LiveKit experience intact.
Live, tested end to end, and open source:
github.com/mahdijnt/liveavat…
Real-time AI experiences are now solved AND open sourced.
We worked closely with @OpenAI to build the best framework for it, combining GPT-Live-1, @TryLiveAvatar, and @HyperFrames_
Check out the demos below and show us what you build with it.
github.com/heygen-com/liveav…
Excited to be a launch partner for GPT-Live-1 with @OpenAI. Building this demo together was a blast.
We also open sourced the whole demo, see link below.
Why it matters:
GPT-Live-1 is one model that listens and speaks at the same time. No STT to LLM to TTS chain, no brittle handoffs. It handles interruptions, background noise and mid-sentence changes of mind, and hands off deeper reasoning to a backend model while the conversation keeps flowing.
Pair that with LiveAvatar and the voice gets a face that reacts in real time. Full duplex audio plus a live avatar is as close to talking to a person as AI gets today.
openai.com/index/introducing…
Repo here. GPT-Live-1 for the conversation, LiveAvatar API for the face. Fork it, swap the language, ship your own:
github.com/heygen-com/liveav…
Want to feel it first? You can try it yourself at liveavatar.com.
Pick Tokyo, Madrid, Seoul, Hong Kong, London or Singapore, start ordering, then interrupt yourself and change your mind halfway through. See how it keeps up.
That's what happens when the best real-time voice model meets the best real-time avatar.
LiveAvatar retweeted
we were legit amazed by GPT-live-1
and when we combined it with @HyperFrames_ and @TryLiveAvatar - and it's magical
here's a demo of a real-time poker coach:
- all cards & ui fully generated on the fly w hyperframes
- GPT-live is the brain and voice
- live avatar makes it all come to life
and we open sourced everything, so anyone can build on top (and yes, the repo is agent-friendly)
LiveAvatar retweeted
GPT-Live-1's voice + a real-time avatar = the language tutor I wish I had.
Full Japanese demo here. French, Korean, Spanish and more in the link below. Pick where you want to travel and start there 👇
Real-time AI experiences are now solved AND open sourced.
We worked closely with @OpenAI to build the best framework for it, combining GPT-Live-1, @TryLiveAvatar, and @HyperFrames_
Check out the demos below and show us what you build with it.
github.com/heygen-com/liveav…