lingotion.com

What It Is

Lingotion is an on-device AI voice acting engine launched by Lingotion AB for the games industry, positioning itself around being “legal, ethical, and real-time.” It is not a general-purpose TTS system or script generator. Instead, it clones an actor’s character performance capabilities using properly licensed actor data, enabling NPCs, character dialogue, and narration to generate emotional speech dynamically in-game based on state.

Core Capabilities

Its Lingotion Thespeon engine runs on the player’s device and does not rely on cloud inference, so there are no per-line or usage-based API costs, and it is better suited to real-time interaction. Technically, Lingotion says it is not an LLM, but a proprietary non-autoregressive architecture. Character models come in five tiers from 3.4M to 33M parameters, can run on CPU/GPU, and typically have 100ms–500ms latency. Each character currently supports 33 emotions, while 59 emotions, emotion blending, and non-verbal sound control are still on the roadmap.

Integration, Privacy, and Compliance

Lingotion supports Unreal and Unity, and can also be integrated with custom engines. The SDK is public on GitHub. Input is JSON containing language, emotion, and text, and the system supports synchronization from word-level text positions to audio sample positions. Compliance is its most distinctive differentiator: the company says its training data is licensed byte by byte, models are trained from scratch, it contracts directly with actors, and 30% of revenue is paid to actors as royalties. Inference does not go through the cloud; actor data is stored on Google Cloud and is accessible only by the company.

Pricing and Trial

The business model is a one-time purchase of an actor clone license per game title. Smaller teams can choose a lower upfront fee plus revenue share. There are no per-line, per-character, or usage-based fees. The website offers free starter actors for testing in Unreal/Unity, with costs only incurred when publishing a game, and also provides technical demos for Windows/macOS. Lingotion Studio additionally offers a token-based mode for purchasing generated audio, but specific pricing has not been disclosed.

Pros, Cons, and Who It’s For

The main strengths are a clear copyright chain, low-latency on-device inference, and well-defined game engine support, making it suitable for commercial game teams that need large volumes of dynamic dialogue. The downsides are that it currently supports English only, with no mention of Chinese; key expressive features such as non-verbal sounds and emotion blending are not yet available; licensing prices are not transparent; and mobile deployment still needs evaluation in terms of model package size and performance. Access from China and payment methods are not explained. Teams in China may also want to evaluate alternatives such as iFlytek, Tencent Cloud voice services, ElevenLabs, or Replica Studios.