Transparent clip package
Alpha-enabled idle, speech, and action videos are bundled with a manifest in one .nva file.
NVA STUDIO
Beyond 3D quality: a Naia video avatar that speaks in real time without a GPU
The NVA (Naia Video Avatar) format is an open-source avatar format that combines the high visual quality of video with the real-time responsiveness of VRM and Live2D. It can be created quickly and affordably without modeling or rigging, then speaks and responds in real time in browsers and desktop apps without a GPU. The file format and player are released as open source. nva-player GitHub
How does NVA work?
GPU-free playback
Heavy models are used during creation; the result plays like ordinary video in a browser.
Voice model combination
The speech clips below combine the Naia voice model with NVA lip movement and expression as one finished character experience.
File structure
avatar.nva (ZIP-compatible) ├── manifest.json ├── clips/idle.webm ├── clips/talking-ko.webm └── clips/<action-id>.webm (optional)
manifest.json
The current NVA v0.2 examples store clips directly under clips/. manifest.animations defines each state or event's clip path, looping, and return behavior; action and localized speech files are registered only when the package includes them.
- animations
- Defines IDs for idle, speech, and optional actions together with their actual paths under clips/, looping, and speech capability.
- speech_clips
- Records language and voice metadata for finished speech assets included in the package. Every locale is not required.
- expressions
- Maps states and user events to animation IDs so one-shot actions can return smoothly to idle.
- Lightweight delivery
- Transparent video, audio, and playback rules travel together—without generation models or production prompts.
COMPLETE EXAMPLES
Try complete NVA examples
Play transparent idle, Naia voice speech, and action clips, then change the background.
Playing: Naia · 2D cel anime
See how the Naia voice model and NVA lip movement and expression come together as one character experience. A finished example of the speech quality and character expression produced when the Naia voice model is paired with NVA.
Instant browser voice preview
This approximate preview uses the browser's built-in voice and is separate from the Naia voice-model clip above.
Available voices and gender classification vary by OS and browser.
Selected voice: No system voice is available.