Yggruzin/stenoai
↗ GitHubPrivacy focused AI powered meeting intelligence using locally hosted Small Language Models. StenoAI Scribe for structured clinical notes in development.
568
Stars
82
Forks
4
Watchers
20
Open Issues
Safety Rating A
No hardcoded secrets, malicious code patterns, suspicious dependencies, or prompt injection attempts were detected. The repository is a straightforward Electron + Python desktop application that bundles Ollama and ffmpeg for local AI inference. All processing is described as on-device. The README contains no embedded instructions attempting to manipulate AI analysis. The project appears to be a legitimate open-source productivity tool.
ℹAI-assisted review, not a professional security audit.
AI Analysis
StenoAI is a privacy-focused, AI-powered meeting intelligence desktop application for macOS that runs entirely on-device. It records, transcribes, summarizes, and enables natural language querying of meetings using local AI models (whisper.cpp for transcription, Ollama-served LLMs for summarization). It supports 99 languages, speaker diarisation, system audio capture, and macOS Shortcuts integration. A clinical note-taking variant (StenoAI Scribe/Med) targeting healthcare workflows is in development.
Use Cases
- Automated meeting transcription and summarization for privacy-sensitive professionals (healthcare, legal, finance)
- Generating structured clinical notes from medical consultations
- Natural language querying of past meeting transcripts
- Offline/local AI inference on Apple Silicon Macs with no data leaving the device
- Calendar-triggered automatic recording of scheduled meetings
Tags
Project Connections
Murmure
→Both are privacy-first, local speech-to-text desktop applications using neural network models. Murmure uses NVIDIA Parakeet TDT targeting 25 European languages with optional Ollama post-processing; StenoAI uses whisper.cpp targeting meeting transcription with speaker diarization and a clinical note-taking variant.
Voicebox
→StenoAI transcribes and summarizes spoken audio entirely on-device; Voicebox synthesizes speech from text entirely on-device. They address opposite ends of the voice pipeline and can be combined for a fully local dictation, summarization, and voice output workflow.
local-talking-llm
→Both are fully local voice plus LLM pipelines with no cloud dependency. local-talking-llm targets real-time conversational voice assistants using Whisper and Ollama; StenoAI targets meeting intelligence with whisper.cpp and Ollama summarization — overlapping technology stack, different end-user workflows.