A cross-platform desktop speech-to-text application built with Tauri v2 in Rust + React. It provides real-time audio transcription using the Deepgram API with the nova-3 model. Key features include live interim and final transcription results via WebSocket streaming, automatic WAV file saving of recordings with waveform visualization, and audio playback. Audio capture uses cpal for cross-platform support, converting to mono linear16 PCM for Deepgram.
The is a chat application that offers a streamlined, no-nonsense AI chat experience. It provides access to multiple large language models (LLMs) from various vendors. Key features include markdown rendering of AI outputs, syntax highlighting for code, and the ability to display Mermaid.js diagrams on a canvas. The app supports real-time synchronization and resumable streams using Zero (sync engine), enabling users to share links, and allows for collaborative use of LLMs with peers. Authentication is handled via Google OAuth, the database is Postgres, frontend is in NextJS, self-hosted on an Hetzner server