Aegisub is a free editor for timing and typesetting subtitles by hand, and it has no speech recognition: the text has to come from somewhere else. suubbs is a $9.99 Mac app that listens to the video, writes every line, times every word against the audio and exports SRT, WebVTT or styled ASS. If you are typing dialogue into Aegisub one line at a time, that is the part suubbs replaces. If you are typesetting signs or building karaoke, keep Aegisub.
They overlap on editing and on the ASS format. Where the subtitles come from is the whole difference.
Aegisub has been the reference tool for Advanced SubStation subtitles for well over a decade. It is free and open source, it runs on Windows, macOS and Linux, and it is unmatched at the fine work: positioning a sign so it tracks a moving wall, writing override tags that rotate and fade a line, timing karaoke syllable by syllable, and automating repetitive edits with Lua scripts. Fansub groups built their workflows around it, and nothing here is trying to talk anyone out of that.
What Aegisub does not do is listen. Every line starts empty. You type the dialogue, or paste it from a script, and then you time it, line by line, against the waveform or the spectrogram. That is the right method when you are translating from a script you already have, and a slow one when all you have is a video of someone talking and you need captions on it by tonight.
suubbs starts from the audio. Drop the video in, press Generate, and Whisper runs on your own Mac and produces the transcript with a timestamp on every word. In the Mac app the word timing is measured at about 3 ms of error against the audio. Lines are then built from those words to reading-speed rules — 42 characters a line, two lines, about 20 characters a second by default — so the first draft is already segmented the way a subtitler would segment it. Your job becomes correcting the handful of words the model misheard, not typing the other two thousand.
Aegisub wins on price, platforms and typesetting. suubbs wins on everything that happens before the first word exists.
| Aegisub | suubbs | |
|---|---|---|
| Price | Free, open source | $9.99 once; free browser trial |
| Platforms | Windows, macOS, Linux | macOS 11 or later only |
| Writes the transcript | No — you type or paste it | Yes, Whisper on your Mac |
| Timing | By hand, against waveform or spectrogram | Automatic, per word, then editable |
| Typesetting (signs, moves, transforms) | The best there is | No — one style, with colour and weight per line or word |
| Karaoke and Lua automation | Yes | No |
| Word-by-word highlight | Hand-built with karaoke tags | One switch, from the real word timings |
| Formats | ASS natively; opens and saves SRT and others | Exports SRT, WebVTT, ASS, word-level JSON |
| Uploads your video | No | No |
The same job, done the way each tool is designed to be used.
The steps you skip in suubbs are the second and third, which in practice are most of the evening. What you give up is the control Aegisub offers from step five onwards: suubbs has one caption style for the document that a line or a single word can override — a colour, a weight, italics — but no signs, moves or transforms, a short list of five typefaces chosen because they resolve when libass renders the file, and no override tags you can write by hand.
The two are not mutually exclusive, because they speak the same file format.
If you want Aegisub's typesetting but not its blank page, let suubbs write and time the dialogue and open the result in Aegisub for the finishing work. suubbs' ASS export is a standard file: one style named Default written from your settings, inline overrides only where you changed a line or a word, and the play resolution set to the video's own frame size, so positions and sizes mean the same thing in Aegisub as they did in the preview.
Three things are worth knowing before you do:
BorderStyle 3 box takes its
padding from the Outline column and its colour from OutlineColour.
suubbs currently writes Outline 0 with the box colour in BackColour,
so the box does not show when rendered. In Aegisub, set the style's outline to a few pixels and
its outline colour to the box colour. Either way it is a rectangle: ASS has no corner radius.Going the other way works too: open a video in suubbs and import an existing SRT or VTT, and you can edit, restyle and re-export it. Word timings in an imported file are interpolated across each line, since SRT never stored them, and suubbs says so when it imports. For burning the finished ASS into the picture, the ffmpeg command is in burning subtitles into a video.
…you typeset signs, build karaoke, script your edits, work from an existing translation, or need Windows or Linux. It is free, and for that work nothing else comes close — suubbs included.
Buy suubbs if your subtitles start as speech and you are on a Mac: the transcript, the word timing, sensible line breaks and a styled export are what it does, it does them without uploading the video, and it is $9.99. The free browser version transcribes the first 60 seconds of your own footage, so you can judge the accuracy before paying.
Mostly about where one tool stops and the other starts.
No. Aegisub has no speech recognition. You type or paste the text yourself and time each line by hand against the audio waveform or spectrogram.
Yes. suubbs exports standard Advanced SubStation (.ass) with the play resolution set to the video's own frame size, and SRT, both of which Aegisub opens. A file with word highlighting arrives as one line per word, because that is how the highlight is written.
The original project's last stable release is from 2014. Development has continued in community forks, which is where current builds and fixes come from.
Aegisub. Signs, typesetting, karaoke effects and automation scripts are what it was built for, and suubbs does none of them.
To download the speech model on first launch, and after that only for small things you can do without: an update check when the app opens, and a translation model if you translate into a new language. Transcribing, editing and exporting need no network, and the video is never uploaded.
Written in November 2026. Open-source projects and their forks move, and we are not always the first to notice. If something here is out of date, tell us at hello@suubbs.com and we will fix it — an inaccurate comparison is not worth a sale.