Stenograf

Overview

Stenograf is an on-device conversation recording and transcription application developed by Soniqo for macOS, Windows, and Android. It captures system audio and microphone inputs locally to generate real-time transcripts, speaker identifications, and summaries without sending conversation data to cloud servers. Users can query past meetings, review action items, and translate speech offline directly on their own hardware.

Key Features

• Dual-source recording — Captures both computer audio and microphone input separately, supporting in-room meetings as well as third-party virtual call applications.
• On-device transcription and diarization — Generates local transcripts, labels speakers, and recognizes saved voice profiles across future sessions entirely on the local machine.
• Offline mobile translation — Allows Android users to record speech and translate conversations across seven languages without an internet connection after initial model downloads.
• Local conversational search — Enables users to chat with single meetings or search across historical archives with cited transcript passages and attached reference documents.
• External AI integration via MCP — Provides read-only access to meeting recordings and notes for developer tools such as Claude Code, Codex, and Claude Desktop using the Model Context Protocol.
• Calendar and task management — Syncs with Google Calendar, Outlook, and Apple Calendar for automated joining reminders, alongside export workflows for platforms like Slack and Jira.

Who It's For

Stenograf is designed for privacy-conscious professionals, executives, and technical teams who need automated meeting documentation but operate under strict confidentiality or data-governance policies. It is well-suited for legal, financial, healthcare, and engineering environments where cloud-based recording bots are restricted or banned. The tool also serves field professionals requiring offline voice transcription and translation on mobile devices.

Notable Strengths

Stenograf stands out for its complete on-device execution, ensuring that audio files, transcripts, calendar entries, and voice profiles remain exclusively on the user's computer or phone. Its implementation of the Model Context Protocol allows local AI assistants to safely query meeting records without exposing internal data to external SaaS vendors. Furthermore, the desktop platform offers predictable flat pricing at $12 per month, avoiding the metered per-minute billing models standard among cloud transcription services.