AI Voice Engine is our in-house voice technology. It is being built to handle everything between the microphone and the speaker: hearing what a person says, passing it to a language model and answering out loud. What to say is decided by the product built on top of it.
The engine is internal technology and is not offered as a separate product. No product uses it yet; the new version of VASS is planned to be the first.
What it does today
The engine runs in test builds for Android and Windows.
- Spoken conversation in three languages: English, Ukrainian and Russian.
- Interchangeable models. Speech recognition, the language model and speech synthesis are separate parts, so each can be replaced with a different provider. The test builds use European providers for speech recognition and the language model, and the device’s built-in voice for speech.
- Web search for questions that need fresh information.
Planned
- Interrupting the engine mid-sentence. This exists as an experimental option and is still being tested.
- iPhone and iPad support.
- More European languages.
- The new version of VASS running on the engine.