Feature
VidNotes turns any video into a searchable, timestamped transcript using advanced speech-to-text. Lecture recording, Zoom meeting, YouTube tutorial, local file. You'll get accurate results in seconds. The engine handles different accents, technical jargon, fast talkers, and background noise, adjusting to whatever audio quality you throw at it. Take a 90-minute university lecture recorded on a phone in a big hall with echo and ambient noise. You'll still get a reliable transcript that picks up technical terms correctly. The audio runs through OpenAI's Whisper model via AIProxy, trained on 680,000 hours of multilingual audio. That's broad coverage of accents, dialects, and domain vocabulary across fields like medicine, law, and engineering.
Paste a YouTube, TikTok, Instagram, or Vimeo link straight into VidNotes on iOS, the web app at app.vidnotes.app, or use the Chrome extension for instant YouTube transcription. You can also upload video files from your device, iCloud Drive, Google Drive, or Dropbox. All the common formats work. MP4, MOV, M4V, AVI. On iOS, the import system uses native file browsing with full cloud storage integration. On the web, drag and drop or paste any video URL.
VidNotes pulls the audio track using AVFoundation and converts it to an optimized M4A file stored in the app's Documents/Audio directory, then sends it through the Whisper transcription engine via AIProxy. You get back a segmented transcript with precise timestamps for each passage. If a YouTube video already has captions, VidNotes pulls them straight through the VidNavigator API for near-instant results, typically under 3 seconds no matter how long the video is.
Browse the transcript in segmented or full-text mode. Tap any timestamp to jump to that exact moment with time-synced playback. Search inside the transcript to find specific words or phrases instantly. The segmented view shows each passage as its own block with its own timestamp, so you can scan a long video and find the section you need without scrubbing through a timeline by hand.
Students
Record your lectures and get a full, searchable transcript instead of scrambling to take notes by hand. Before exams, search for key terms and jump straight to the right timestamps, turning a two-hour lecture into a focused five-minute review.
Journalists
Transcribe interviews accurately and search for exact quotes without replaying hours of audio. Export timestamped transcripts to point to precise moments when fact-checking or writing, and handle sources in any of 30+ languages without reaching for a separate translation tool.
Meeting organizers
Turn recorded meetings into written records so no decision or action item slips through. Share searchable transcripts with anyone who missed the call, and pair with the action items feature to pull commitments and deadlines straight from the conversation.
For local videos, VidNotes pulls the audio track using AVFoundation, converts it to an optimized M4A, and sends it to the Whisper transcription engine through AIProxy. The extraction pipeline handles videos with multiple audio tracks, picks the primary track automatically, and compresses it for upload without sacrificing speech clarity. You get clean, accurate results even from videos with messy audio, background music, or uneven recording quality.
For YouTube and social media videos, VidNotes first checks for existing captions through its VidNavigator API integration. When captions are there, you get results almost instantly, usually in 2-3 seconds no matter the length. When they aren't, VidNotes falls back to a RapidAPI integration that pulls captions from a different source. As a final fallback, the video runs through the full Whisper transcription pipeline with the same accuracy you'd get from a local file. Three tiers, maximum speed and reliability.
Every transcript lives locally on your device via SwiftData, instantly searchable and available offline. You own your data and can export or delete it whenever you want. The SwiftData schema connects each transcript to its source video through a VideoProject entity, holding onto metadata like duration, source type, and thumbnail. Transcript segments are stored as individual entities with their own timestamps, which is what powers tap-to-jump navigation and granular search.
No account required. Paste a video link and see it in action.