Whisper (OpenAI)
Leading#2 in Open sourcehigh confidence
The open-source transcription foundation (105k★) behind most local meeting-notes stacks — accurate, multilingual, and fully offline. The engine, not a full notetaker.Our read
Why it ranks #2
The category's foundational open-source speech model, with about 105.5k GitHub stars and extensive downstream adoption across transcription tools.Here is the catch
not a complete meeting-notes application
large models are slow and memory-intensive without a GPU
no built-in diarization or polished real-time workflow
Does this well
widely tested model with broad language support
large ecosystem of ports, wrappers and tutorials
runs locally without sending audio to a service
Pricing
Checked by hand on 2026-07-23. Prices in this category change often — if this looks wrong, it probably is.
Key features
Multilingual speech recognitionSpeech translation into EnglishMultiple model sizesCommand-line and Python interfaces
Sources we read
Quick facts
More in this area
The rest of the Open source column.- 1MeetilyThe recognized #1 self-hosted, open-source meeting notetaker (26.4k★) — local Parakeet/Whisper live transcription, speaker diarization, and Ollama summarization, 100% on-device with no cloud.
- 3whisper.cppThe high-performance C/C++ port of Whisper (52.2k★) — fast, dependency-light on-device transcription that powers Meetily and many local meeting-notes tools.
- 4WhisperXWhisper plus accurate word-level timestamps and speaker diarization (23.2k★) — the who-said-what layer that makes raw transcripts usable as meeting minutes.
- 5Natively-AI-assistant/natively-cluely-ai-assistantDesktop overlay that transcribes calls locally, feeds you suggested answers mid-conversation without joining as a participant, and writes notes afterwards.
- 6Hankanman/Parley
- 7MikeSchirtzinger/silent-notetakerRuns the whole transcription and note-extraction pipeline inside a WebGPU browser tab, so there is no app to install and no audio upload.