Keyframes without manual screenshots
FrameNotes detects meaningful visual changes, filters repeated frames, and keeps the images that make a video easier to review.
Beta: short-video MVP, local upload recommended
Turn Short Videos
into Visual Notes
Upload a video. Get screenshots, timestamps, and matching transcripts in a note you can read, edit, and share.
Watch it work
Processing preview: keyframes extracted, audio transcribed, notes packaged.
25%Extracting key images from the video
60%Transcribing audio
100%HTML and Markdown ready
Core features
Smart keyframes, no manual screenshots
Transcript with screenshots aligned to the timeline
Cloud ASR providers turn clear audio into readable transcript segments. When timestamps are available, captions are matched under the right image.
Markdown and HTML notes you can keep
Generate HTML, Markdown, and presentation-style pages for study, research, and review workflows.
How it works
Four visible stages.
No endless spinner.
The workflow separates upload, visual extraction, transcription, and export into visible stages. If something fails, you know exactly where.
Upload a local video
Choose a local MP4, MOV, AVI, or MKV file. The browser uploads the file first, then FrameNotes creates a task ID that can be used to restore progress after a refresh.
Detect visual changes and remove repeated frames
FrameNotes analyzes the video timeline for meaningful visual changes. It keeps useful slides, diagrams, browser states, and presentation moments while filtering near-duplicate frames. This produces a smaller set of images that is easier to scan than fixed-interval screenshots and avoids filling a note with repeated views of the same slide.
Extract audio and create the transcript
The processing job separates the audio and sends it to the configured speech-to-text provider. Transcript quality depends on clear speech, background noise, language support, and provider availability. When timestamp data is returned, FrameNotes matches each transcript segment to the image shown during that part of the recording.
Build portable notes and downloadable files
After image and transcript processing finishes, FrameNotes generates a reading-first result. The standalone HTML file embeds its images for convenient local viewing. Markdown can be downloaded with its image folder as a ZIP package, while Word and presentation-style HTML provide alternative formats for editing, reviewing, or sharing.
What you get
Video notes that remain useful after processing ends.
FrameNotes keeps the relationship between time, image, and spoken context, then packages the result in formats that open outside the website.
A visual timeline instead of disconnected screenshots
Each selected keyframe includes its time range and the transcript text associated with that moment. The result preserves enough sequence to understand how the speaker moved from one idea to the next. It is designed for review and reference, not as a replacement for the original video or a promise of a perfect verbatim record.
Formats for reading, editing, and archiving
Use standalone HTML when you want one file that opens in a browser with images visible. Use the Markdown ZIP when you want editable text plus a stable relative image folder. Use DOCX for a conventional document workflow. JSON remains available as structured task data for technical inspection and future integrations.
Honest failure states and recoverable task progress
FrameNotes reports the current stage, percentage, and task status instead of showing a single indefinite loading message. Recent task IDs are stored in the browser so progress can be checked after a refresh. If download, transcription, or processing fails, the public interface should show a useful product message rather than inventing transcript text.
Use cases
Built for people who
learn from video.
Turn recordings into a visual reference for learning, research, and creator review.
Course learning
Turn a short lesson into review notes that pair the instructor's slides with the explanation delivered at that moment. Students can revisit definitions, diagrams, and examples without repeatedly scrubbing through the full recording.
Lecture review
Capture the visual structure of a recorded lecture and keep the related transcript nearby. This is especially useful when a speaker builds an argument across several slides and the image alone does not preserve the spoken context.
Research clipping
Create a searchable reference from interviews, talks, webinars, or product demonstrations. Researchers can keep key visual evidence beside the speaker's words, then move the Markdown package into their preferred knowledge-management workflow.
Creator reference notes
Convert a short source video into a compact research document before outlining an article, lesson, or response. FrameNotes helps organize source material, but users should still verify quotations and facts against the original recording.
Meeting and demo recaps
For short recorded walkthroughs, extract important interface states and the explanation around them. The result can support internal review or documentation, provided participants are authorized to upload and process the recording.
Podcast and interview review
When a recording includes useful on-screen references, names, charts, or chapter cards, FrameNotes combines those visuals with transcript segments. Audio-only material can still be transcribed, although the visual-note benefit is naturally smaller.
FAQ
Quick answers.
What types of videos can FrameNotes process?
FrameNotes is designed for short lectures, software tutorials, product walkthroughs, recorded presentations, research clips, meetings, interviews, and other videos where screenshots add useful context. Clear speech and meaningful visual changes usually produce the most useful notes.
Which video file formats are supported?
The current upload interface accepts MP4, MOV, AVI, and MKV files. MP4 with a common H.264 video track and a standard audio track is usually the most portable choice.
What are the current video size and length limits?
The public beta currently supports videos under 30 minutes and local files under 500 MB. Longer videos still need a separate worker pipeline and may be gated.
Why can a local video fail to generate notes?
Common causes include a file above the beta limit, an unsupported or damaged media stream, no usable audio track, unclear speech, a transcription-provider quota or availability issue, or a video with too little visual change to extract useful keyframes. FrameNotes shows a task-level error when a processing step fails.
Why can processing take a while?
A job must upload the video, inspect the timeline, extract keyframes, extract audio, wait for transcription, and generate downloadable files. Processing time varies with file size, video length, visual complexity, audio-provider response time, and whether another task is already running.
Which languages does FrameNotes support?
Language detection and transcription support depend on the speech-to-text provider configured for the service. Clear speech in a provider-supported language works best. Mixed languages, accents, overlapping speakers, music, and poor audio can reduce accuracy, so important names, numbers, and quotations should be checked against the original video.
Which formats can I download?
Current outputs include a standalone HTML result page, Markdown files, a Markdown ZIP package with images, a DOCX document, and a presentation-style HTML page. The Markdown ZIP keeps the note and its image folder together so relative image paths continue to work after download.
How long are uploaded videos and generated files kept?
Uploads and generated files are stored in per-job server directories to complete the conversion and make downloads available. The current beta does not publish a guaranteed automatic deletion timetable, so do not treat FrameNotes as permanent storage or upload confidential, regulated, or highly sensitive material. Download important results promptly; removal requests can be sent to contact@framenotes.org with the job ID or result URL.
Will my task continue if I refresh the page?
Once the upload finishes and the API creates a task ID, processing runs as a background job. The frontend keeps recent task IDs in local browser storage and can restore their status after a refresh. Do not close the page during the initial file upload, because no server task exists until that upload completes.
How is FrameNotes different from a regular video summarizer?
Many video summarizers focus on a rewritten text summary. FrameNotes focuses on a reviewable visual note: it selects key screenshots, aligns transcript segments to the video timeline, and exports the result as HTML, Markdown, DOCX, or presentation-style files. It is designed to preserve source context, not to promise a perfect replacement for the original video.
Stop rewatching.
Start reading.
Three free conversions each month. Upload a video to create your first visual notes.
Try FrameNotes free