Meetily Offers Free Meeting Transcription and Summaries Without a Subscription
An open-source app called Meetily lets users transcribe and summarize video calls for free using locally run AI models, avoiding the monthly fees and cloud privacy concerns tied to commercial meeting assistants. It works with any videoconferencing tool and is available for Windows and macOS.
Turning conversations into text has become one of the more practical uses of artificial intelligence, and now this capability is available at no cost. Tech outlet Wired has highlighted Meetily, an application that transcribes and summarizes meetings without the subscription fees typically charged by commercial services, which usually run between $10 and $20 per month.
Unlike most paid meeting assistants, which require uploading recordings to the cloud and raise confidentiality concerns, Meetily runs open-source AI models locally on the user's own computer. The app is available for Windows and macOS, while Linux users must build it from source. A paid Pro version exists for $10 a month, but the community-built open-source version is fully functional and free.
How it works
After installation, the app offers to download the AI models it needs to function. Users must grant access to both their microphone and system audio so Meetily can capture every side of a conversation. Because it accesses audio directly rather than integrating with a specific platform, it works with Zoom, Google Meet, Microsoft Teams and other videoconferencing services, and can theoretically be used for in-person meetings as well.
During a meeting, users can watch the transcription happen in near real time or minimize the window and focus on the discussion. Once recording stops, users can request an AI-generated summary and specify which points it should focus on. A beta feature also allows processing of previously recorded audio or video files.
Limitations
The free version has drawbacks — it cannot distinguish between speakers or label who said what, which may be a dealbreaker for some users. Other free alternatives mentioned include noScribe, which can identify speakers but only works on existing files rather than in real time, along with Anarlog and Quill, each with their own tradeoffs.
