Video

Which tool for which job

• Too big to send: Compress video. Cut first with Trim video and drop useless sound with Mute video; both shrink the file with no quality cost.
• Only one part matters: Trim video, accurate to a tenth of a second.
• Several clips should become one: Merge videos.
• Recorded sideways: Rotate video.
• Faster tutorial or slow-motion replay: Change video speed, from 0.5× to 2×.
• The file will not play or import: MOV to MP4, MKV to MP4, AVI to MP4 or WebM to MP4.
• You only need the sound: MP4 to MP3.
• A short silent loop for a ticket or a README: Video to GIF.
• Speech as text: Transcribe for TXT, SRT or VTT, Translate video for English text from another language, Add subtitles to burn them into the picture.

Containers, codecs and why MP4 wins

A video file is a container holding separately compressed streams. The extension names the container; what decides whether a device can play it is the codec inside.

• MP4 with H.264 video and AAC audio is the combination nearly every phone, browser, TV and editor from the last fifteen years decodes. Every converter here produces it.
• MOV is Apple's container. Recent iPhones fill it with HEVC, Mac editors with ProRes, which most non-Apple software rejects.
• MKV can hold almost any codec plus several audio and subtitle tracks, which is why films and OBS recordings use it.
• WebM holds VP8, VP9 or AV1, never H.264, so converting it always means re-encoding.
• AVI dates from 1992 and usually contains DivX, Xvid, Motion JPEG or DV. An hour of DV is about 13 GB.

When the video stream is already H.264, the MOV, MKV and AVI converters copy it into MP4 without touching it, which takes seconds and loses nothing. Everything else is re-encoded at CRF 23, a constant-quality setting where the difference from the source is hard to see. HEVC, VP9 and AV1 need roughly 30 to 50% fewer bits than H.264 for the same picture, so a compatible MP4 is often larger than the file it came from.

For scale: a 1080p phone clip is recorded at 15 to 20 Mbit/s, about 120 to 150 MB per minute. The same minute as a 192 kbps MP3 is about 1.4 MB.

In the browser or on the server

By default every video tool except the speech tools runs FFmpeg compiled to WebAssembly inside the page. The engine is about 30 MB, downloaded once and cached. Nothing is uploaded, but the work happens on one CPU core and the whole file sits in the tab's memory, so 1080p converts at roughly real time on a laptop and much slower on a phone. Compress, trim, rotate, mute, MP4 to MP3 and the four converters accept up to 2 GB; merge, speed and GIF accept up to 500 MB.

Compress, rotate, mute, MP4 to MP3 and the four converters also offer a server path with native FFmpeg. It uses the slower medium encoder preset instead of the browser's fast one, which gives a noticeably smaller file at the same quality, writes the index at the start so playback begins before the download ends, and avoids the memory limit. Compression maps Light, Medium and Heavy to CRF 28, 32 and 38 in the browser and CRF 20, 25 and 30 on the server.

Transcription, translation and burned-in subtitles run only on the server, because the Whisper speech model is far too large for a browser tab. Recognition takes about as long as the recording. Free runs accept recordings up to 10 minutes; Premium accepts up to 90 minutes per file.

Server use is free once per rolling 24 hours per device, for files up to 500 MB, without an account; failed runs are not counted. Premium removes the daily limit and raises the cap to 4 GB. The upload is deleted as soon as the result is ready, and the result expires about an hour after the job was created.

Workflows that chain well

1. A clip for a messenger: trim the dead ends, mute wind noise, then compress at Medium.
2. iPhone footage for a Windows user: MOV to MP4 re-encodes HEVC to H.264. If the result is larger, run it through Compress video at Light.
3. Transcribing a long screen recording: extract the sound with MP4 to MP3, turning 300 MB into 10 to 15 MB, then upload the MP3 to Transcribe. Split anything over your limit at a pause.
4. Captioned social clip: trim to the part you will post, then add subtitles. For full control over wording, transcribe to SRT, correct it, and fix timing with the subtitle converter.
5. A GIF from the middle of a video: the GIF length always counts from 0:00, so trim first, convert with Video to GIF, and shrink with Compress GIF if needed.
6. Merging phone and webcam clips: the first clip sets the frame, 1280×720 for landscape or 720×1280 for vertical, so rotate sideways clips and put the right one first.

When something goes wrong

• The MP4 is bigger than the original. Usually an HEVC or VP9 source plus the fast browser preset. Use the server path or compress afterwards.
• The trim sliders never appear. The browser cannot play the codec, typically iPhone HEVC. Convert with MOV to MP4 first.
• The tab reloads or the job stops. The browser ran out of memory. Close heavy tabs, use a desktop, or switch to the server path.
• A WebM screen recording shows 0:00 and will not seek. The recorder never wrote a duration. Converting to MP4 writes a proper index.
• An MKV lost its second language or subtitles. One video and one audio track are kept. Keep the subtitle file and convert it with the subtitle converter.
• Old camcorder footage shows comb lines. DV is interlaced, and the converter does not deinterlace.