Six ways to get something out of a video

Two of these tools touch nothing

This is the part most converters are vague about, so it is worth saying plainly. Re-encoding a video or a piece of audio always costs quality, because it decodes what is there and compresses it again. Two of the tools here avoid that step entirely whenever they can, and Compress Video is the one that cannot: reducing a file size is re-encoding, which is why its result screen reports what the trade actually cost rather than leaving you to guess.

MOV and MP4 are containers: different boxes around the same encoded video. A MOV from an iPhone, a Mac screen recording or most cameras is already carrying video that MP4 holds natively, so MOV to MP4 moves the streams across without decoding them and the picture that comes back is identical. Only footage ordinary players cannot open, ProRes from an editing program being the usual case, has to be encoded again. The result screen tells you which of the two happened rather than letting you assume the good one.

Video to Audio offers the same deal on sound. Its default is an MP3, which is a second lossy pass over audio that a video had already compressed once, and no bitrate recovers detail the recording never held. The alternative copies the existing track into its own file without decoding a sample, so the sound is exactly what the video carried. It is also the fastest option and usually the smallest file. When a track has no container it can sit in on its own, the job is refused with a message instead of quietly re-encoding it and calling that the same thing.

When one of these is the wrong answer

A 50 MB ceiling rules out most footage straight off a camera, and no option here raises it. A feature-length recording, a folder of clips, or anything you will do again next week wants a local copy of ffmpeg rather than a browser tab: it is free, it has no size limit, and it will be faster than uploading.

Two more cases belong elsewhere. A transcript that has to be right rather than roughly right still needs a human pass, because automatic speech recognition guesses at names, crosstalk and anything competing with background noise, and it guesses confidently. And a recording that is confidential in the way a legal or medical one is confidential deserves the tool your workplace has approved: our deleting it an hour later is easy for us to state and hard for you to verify from outside, which is exactly why the mechanism is written down above rather than asserted.

Questions people actually ask

Are these video and audio tools free?

Yes. No account, no watermark on anything you download, and nothing held back behind a paid tier. The site is funded by ads rather than by charging for the work. Speech is the only rationed part, because it is the only one that costs real processing time: five jobs a day per connection, shared between the transcript tool and the subtitle generator, because both run the same pass.

Do these tools upload my video?

The video and audio tools do, and that is the honest difference between this category and the quick utilities on this site. A video is far too large to process inside a browser tab, so each of them sends the file to our server over HTTPS, does the work there, and sends a result back. The one exception is Subtitle Converter: a subtitle file is a small piece of text, so it is converted inside your browser and never uploaded.

What happens to my video afterwards?

It is used for your one job and then deleted automatically, along with whatever the job produced. The retention window is an hour from when the job was created, and a sweep runs every ten minutes, so nothing lives longer than about seventy minutes whatever its status ended up being. A video still waiting in the transcription queue is kept until it runs, with a twelve-hour limit for work that was abandoned.

How large and how long can a video be?

Every tool here takes one file of up to 50 MB. The length limit is not the same on all of them: the two speech tools and the audio extractor accept up to 60 minutes of running time, while MOV to MP4 and Compress Video stop at 20 minutes, because a job that has to re-encode the picture is far slower than one that reads the sound. Go over and the job is refused with a message that names the limit, rather than failing partway through.

Is my recording sent to a third-party AI service?

No. Speech is transcribed by a faster-whisper worker running on our own machine, so the audio does not leave our infrastructure for an outside model. That covers both the transcript tool and the subtitle generator, which are the same pass with a different file at the end. The conversion tools involve no AI at all: they are ffmpeg doing container and codec work.

Why did the transcription say the queue is full?

Because ten transcriptions were already running or waiting. Speech recognition holds a worker for the length of the audio, so the admission cap keeps the queue honest rather than accepting a job that would sit for an hour. It is a plain refusal with a message, not a silent failure, and trying again a few minutes later usually works.