Tambourine

Tambourine is an open-source platform designed to provide AI-powered voice dictation for use with any desktop application. It addresses the need for customizable, personal voice interfaces, allowing users to dictate text into any app that accepts text input on Windows and macOS. The platform is built to be adaptable, letting users tailor its behavior to their specific requirements.

A core feature of Tambourine is its flexibility in model selection. Users can choose their preferred speech-to-text (STT) and large language models (LLMs), with options to use cloud-based providers or run models locally, such as Whisper and Ollama. The dictation process is highly customizable through editable prompts, enabling users to define how their speech is formatted, set rules for punctuation, and adjust the tone. Tambourine also allows for the creation of custom features by modifying prompts, including backtracking corrections—where spoken corrections like "actually" or "scratch that" update the dictated text—and smart list formatting, which can automatically convert spoken lists into formatted text.

Additional personalization options include a personal dictionary for teaching the tool technical terms, names, or jargon unique to the user's work. The platform emphasizes user control, encouraging the development of new features and workflows by modifying prompts in the settings. Features such as context-awareness, app-dependent formatting, selection commands for in-place text transformation, and voice shortcuts for inserting frequently used text are noted as coming soon.

Tambourine is available as a self-hosted solution, allowing users to run the open-source code for free, with full control over their setup. A hosted service, which would handle infrastructure for users, is planned for future release. 0 license, inviting contributions and community involvement. Tambourine positions itself as a foundational tool for building personalized voice interfaces across desktop applications.