Why Rekorder?

Where browser screen recorders stop short

Recording your screen in the browser stopped being a technical problem years ago. Every serious tool captures a tab, a window or a full display and hands you a link. What separates them now is not what they can do, but how much of it you are allowed to do before a paywall interrupts the work.

That is where the familiar names show their shape. Free plans in tools like Loom or Screencastify tend to arrive with a recording length cap, a ceiling on how many videos you may keep, and the interesting parts - captions, the editor, viewer controls, team collaboration - reserved for a paid seat.

Rekorder is built on the opposite assumption. Every capability described on this page is available without paying: capture modes, camera effects, redaction, captions, chapters, summaries, the editor, noise removal, sharing controls, search, folders and workspaces. The single metered resource is cloud storage, because gigabytes have a real cost and features do not, and recording offline without an account is free permanently.

Offline recording, with no account and no upload

Most cloud recorders treat the account as the product: no sign-in, no recording. Rekorder runs offline from the moment it is installed. You capture, you stop, you keep the file - nothing uploaded, nothing captioned on a server, no account to be billed later.

This matters more than it sounds. Recording a customer environment, a staging system with real data or anything covered by a client agreement is usually a policy question rather than a technical one, and the safest answer is that the footage never left the machine. Offline mode is the entire recorder, not a trial: every capture mode, camera effect and redaction tool described below, free indefinitely.

Signing in later is additive rather than a conversion step. Log in and recordings upload with a progress indicator, get captioned, become searchable and sync across devices, while the ones you already made offline stay exactly where they are. Install the extension and decide afterwards.

Capture: tab, window, entire screen or camera only

Rekorder records the current browser tab, a specific application window, the entire desktop, or your camera alone with no screen capture at all - so a product walkthrough, a bug report against a native app, a lecture and a talking-head intro are all the same tool.

Two tab-recording interfaces: injected mode puts the camera bubble and control bar on the page itself, for demonstrating in real time; render mode keeps the tab clean and composites the camera only into the output file.
Resolution and frame rate under your control: 1080p, 720p, 480p or 320p at 15, 24, 30 or 60 fps, with high, medium and low bitrate tiers. A 60 fps capture of a UI animation and a 15 fps capture of a slide deck have very different storage costs, and you decide which you are paying.
Encoder choice with automatic fallback: auto, WebM, MP4, or MP4 through a WebRTC pipeline with device timing calibration. Hardware encoders are used when present and fall back cleanly when they are not.
Crash-resilient local storage: the recording is written in chunks to the browser file system (OPFS) while it runs, so a tab crash, a closed laptop or a browser update mid-take does not cost you the take.
Multi-monitor handling: the recorder tab stays focused after you pick a monitor and a focus-return countdown appears when you switch back, so the first seconds are not you hunting for the right display.

Camera framing and background effects, done on-device

A camera bubble is the difference between a screen capture and a person explaining something. Rekorder gives it the treatment it usually only receives in video conferencing apps, processed locally through a GPU-accelerated segmentation pipeline rather than after upload.

Shape, size and placement: original, wide 16:9, TV 4:3, square or circle, in three sizes, with adjustable border thickness and a light or dark border colour. Eight preset positions cover the corners and edge middles, and the bubble is freely draggable when none of them fit.
Background treatments: blur, virtual background replacement from a built-in image set, dim, grayscale, pixelate and a portrait lighting effect, all previewing live before you start, so you are not discovering the result after a twelve-minute take.
Performance you can steer: quality tiers run from best quality to best performance, with an automatic mode driven by a GPU and CPU benchmark. A machine that falls back to CPU gets a warning and a one-click re-benchmark rather than a silently stuttering camera.
Live control: mirror the camera, toggle it off mid-recording, or push it into Picture-in-Picture and restore it later, without stopping the take.

Background blur and virtual backgrounds are the sort of feature that competing recorders position as a premium differentiator. Here they are part of the free recorder, including in offline mode.

Redaction and cursor highlighting before you hit record

Blurring a customer name out of a finished video is editing. Blurring it before recording is prevention, and considerably cheaper. The blur bar tool lets you click any element on the page - an email address, an account number, an API key, a whole table column - and mask it for the entire recording, including while the page scrolls.

Pointer highlighting works the other way around, drawing attention rather than hiding it. Highlight the cursor on movement, on click or both, and viewers follow exactly what you pointed at and pressed - which removes most of the "wait, where did you click?" replies that otherwise arrive a day later.

Recording controls that keep takes usable

Most wasted recordings are lost to the mechanics of recording rather than to what was being recorded:

Pause and resume mid-take instead of restarting, with restart and cancel-and-delete behind confirmation dialogs so a misclick does not discard twenty minutes of work. A live timer sits on a control bar that snaps to any screen edge and remembers where you left it.
Countdowns before the first frame: none, three or five seconds, with an audible beep, a skip option, and a countdown mirror in the original tab when recording a window or a full screen.
Audio you can verify: pick the microphone, watch a live level waveform to confirm it is picking you up, and mute mid-recording. Tab and system audio are captured and mixed with the microphone at a bitrate that follows the quality tier.
Keyboard shortcuts for everything time-critical: start and stop, pause, mic mute, restart, camera toggle, cancel, and moving the camera bubble between corners when it covers the control you need to show. If a camera or microphone drops out, the recording continues without it.

Captions in around 99 languages

Cloud recordings are captioned on demand with a Whisper large-v3 model, stored at both word and segment level. That granularity is what makes everything downstream possible: captions on the exact word, an editor that can mute a single filler word, and search that jumps to the second a phrase was spoken.

Automatic WebVTT captions served with the video, so recordings are watchable on mute, in an open-plan office, or by viewers who read faster than you speak. Accessibility for viewers who are deaf or hard of hearing becomes a default rather than a task someone has to remember.
Editable transcripts: correct or delete segments where a product name, an acronym or a surname came out wrong, and download the transcript as plain text for documentation, tickets or release notes.
Around 99 supported languages, independent of the interface language. Edits that change timing automatically re-queue the transcript, so captions never drift out of sync with the cut.

Captions are metered or reserved for paid tiers across most of the category. Here they are part of the free plan, constrained by storage rather than by minutes captioned.

Video is a bad format for finding things and a good one for showing them. The timeline closes that gap by binding transcript and playhead together: scroll the transcript and the video follows, scrub the video and the matching text highlights.

Chapters: generated automatically from the transcript, with manual add and delete when you want the structure to follow your agenda rather than the model. A forty-minute recording becomes a list of six navigable topics.
Comments anchored in time: feedback lands on the moment it refers to instead of arriving as "around the middle somewhere", which is what makes asynchronous review faster than scheduling a call. Emoji reactions cover quick acknowledgement, threaded replies cover discussion, and owner-only tasks capture the follow-ups.

Summaries for the people who will not watch it

Every recording has an audience that will watch it and a larger audience that will not. Automatic summaries serve the second group: a condensed account of what the recording covered, produced from the transcript without anyone writing it up afterwards.

In practice this is what makes a recording forwardable. A stakeholder reads the summary, checks the chapter list and jumps into the two minutes that concern them; a new team member reads six months of summaries in an afternoon instead of watching six months of video. Summaries are indexed too, so a recording nobody titled carefully is still findable through what it was about.

Search across everything you have ever recorded

A recording library without search is an archive nobody opens. Rekorder searches every text surface a recording produces, not just its title:

Weighted multi-source matching across titles, transcripts, summaries, chapters and comments, so a phrase mentioned once in passing eight months ago is still reachable.
Full-text indexing with a keyword fallback for the short, partial or unusual terms that strict full-text matching tends to miss, surfaced through debounced autosuggest with the matched terms highlighted, so you can see why a result matched before opening it.
Workspace-scoped and permission-aware results: search covers your recordings and those genuinely shared with you, and never leaks titles or transcript fragments from anything else.

This is the compounding benefit of recording routinely: each capture adds to a searchable body of institutional knowledge - decisions, walkthroughs, bug reproductions, onboarding explanations - that would otherwise survive only in whoever happened to be on the call.

A non-destructive editor that does the tedious passes for you

Re-recording because of one stumble is the most common waste in screen recording, and exporting to a desktop editor costs more time than the mistake did. The built-in editor is non-destructive: the original is kept untouched and you can discard your changes back to it at any point, however many edits have accumulated.

Trim, cut and reorder segments, then either save a draft without rendering or apply the edit and render it, so an unfinished cut costs nothing.
Silence-based auto-cut suggestions with adjustable sensitivity and minimum duration, so the pauses where you were reading the next step or waiting for a page to load are proposed as cuts and removed in one pass.
Filler-word detection and auto-mute driven by the word-level transcript, silencing the "um" and "you know" without cutting frames, alongside audio-only mute ranges for when the video is fine but the audio is not: a phone ringing, a colleague talking, a name that should not be in the recording.
Overlays from an asset library: composite images, video or audio onto the timeline, pushing the timeline apart when you need to make room rather than cover something. Overlay audio can mix in, duck the main track, or stay muted.
Audio enhancement: noise removal at low, medium or high strength for fans, air conditioning and keyboard clatter, plus two-pass loudness normalization so a quiet recording is not the one nobody finishes. The lossless original audio is kept for a one-click revert, and rendering is hardware-accelerated.

An editor of this scope is normally a paid tier of its own, when a browser recorder has one at all. It is included here, along with the asset library that feeds it - reusable images, video and audio with automatic thumbnails, renaming and batch deletion.

Most recorders offer two states: private, or anyone with the link. Rekorder adds the level real work needs, then makes it granular.

Three access levels per recording: private, specific people, or a public link - decided per recording rather than per account.
An invite list with autocomplete drawn from your prior sharing history, and notification emails sent only to newly added recipients, so re-sharing does not re-spam the original audience.
Per-viewer controls: comments, downloads, transcript visibility, view-analytics visibility and a suggested playback speed. A recording sent to a customer and the same recording sent to your team need not behave the same way.
Protected streaming with seek support and an owner-controlled download permission, so a shared recording is not silently a downloadable file for everyone who opens it.
View tracking with per-session deduplication, so the count reflects people rather than page refreshes, and a "shared with me" view where recipients can remove a share from their own list without affecting anyone else.

Workspaces, folders and teams that do not punish collaboration

Recording libraries grow quickly, and the organisational model decides whether that growth is an asset or a mess. Personal and team workspaces stay separate, each with nested folders, and recordings move between folders and workspaces in batch alongside archive, unarchive and delete. My Recordings, Shared with Me, Archive and Search views keep mixed contexts legible with status tags.

Team workspaces carry owner, admin, member and viewer roles, with billing and seat management restricted to owners and admins. Entitlements are per workspace rather than global, so a personal plan does not leak into a team account and a team plan does not silently upgrade a private one. Invitations go out in batches with time-limited tokens, resend and revoke, and acceptance runs through the existing sign-in flow rather than a second account system.

The part worth reading twice: free collaborators always keep view, comment and download rights, and seats are only needed for people who create, upload, edit or caption. Under per-seat pricing, ten people who only ever watch are ten billable seats. Here they are not billable at all, which is usually the largest single difference against tools priced strictly per user.

What actually costs money

Everything described above is available on the free plan. There is no recording length cap, no monthly quota of captions, no editor upsell and no premium tier hiding the camera effects. The one metered resource is cloud storage, because storing video has a genuine ongoing cost while unlocking a feature you already have does not.

That constraint is also the one you control directly. Choosing 720p instead of 1080p, 30 fps instead of 60, trimming dead air out of a take, or archiving what you no longer need all buy back space without costing you a capability. Outgrow the free storage and you are paying for gigabytes, not for permission.

And if none of it should touch a server, offline mode remains free and unlimited: the full recorder, camera effects and redaction included, with your recordings staying on your machine. Compare that with the usual free tier in this category - a five-minute ceiling, a video count limit, captions and editing behind a subscription - and the difference is not a discount, it is a different arrangement. See the plans or read how recording works.

Conclusions

A screen recorder is worth what its output is worth six months later. A file on a drive named recording-14 is worth very little. A recording that is captioned, chaptered, summarised, searchable across every word spoken in it, editable without a desktop tool and shareable with exactly the permissions the situation calls for is a durable piece of the record.

Rekorder makes that second outcome the default rather than the upgrade path: capture in the mode that fits, redact before recording instead of after, clean up silences and noise in the editor, and let captions and search do the archival work - offline when it should stay local, synced when it should be shared. Record something today and check what you had to pay to finish it.