How to edit video by editing the text transcript
In Video by Zaweba you transcribe a clip in the Transcript tab, select the words you don't want and press Delete to cut them from the edit.
Text-based editing is built into Video by Zaweba. Open the Transcript tab and let it recognise the speech, then select any words you want gone and press Delete; the matching stretch of picture and sound is cut from the edit. It all happens on your own PC, with no upload and no account.
For talking-head videos, interviews, lectures and podcasts filmed on camera, this is often the fastest way to make a first cut. Reading is quicker than scrubbing, and you can see a whole ten-minute answer on one screen.
How the transcript is made
The Transcript tab recognises speech in pieces of about 30 seconds, working through the clip on your computer. When it's done you get the words laid out as text, tied to their moments on the timeline. Long recordings take longer, so start the transcript and get on with something else while it runs.
Recognition uses the same local engine as the captions, so nothing leaves the machine. That matters for interviews with private details, for client work under agreement, and simply for working on a train without a connection.
Cutting with the keyboard
Read through the transcript and select the words you don't want, the same way you'd select text in a document. Press Delete and they're cut from the edit. A false start, a repeated sentence, a tangent that went nowhere: each one is a drag and a key press.
Because the edit is ordinary timeline editing underneath, everything else still works. Undo (Ctrl+Z) brings a cut back, and you can switch to the timeline at any point to tidy an edit by hand, add a cutaway on the track above or drop a transition over a jump. Many people rough out with text and finish on the timeline, which plays to the strengths of both.
Find helps on long recordings. Search for a word or a name to jump to every place it's said, which makes it easy to pull out the best version of an answer that came up more than once.
Tidying the gaps
Two buttons do a lot of the routine work for you. Remove pauses takes out the silences between phrases, and Remove filler words strips the ums and ahs. Run them after your own content cuts, not before, so you're judging the material rather than the rhythm. Then play it back; a talk with every pause gone can feel breathless, and undoing a few is quick.
Hiding the joins
Text editing produces jump cuts, since the speaker's head moves between kept sentences. You have choices. Put B-roll on a track above so the cut is covered. Use a gentle punch in on alternate sections with the scale controls. Or try the Smooth jump cut transition, which blends two takes of the same shot with an optical-flow morph to hide the jump.
If more than one person is talking, Transcript > Find speakers groups the words by who says them, still on your PC and without a script. Click a speaker's heading to rename them everywhere, which makes a two-person interview much easier to read and cut.
Keeping the words
Save text writes the transcript out, useful for show notes, a blog post or checking quotes. And since the words are already recognised, captions are a short step away in the Captions tab once the cut is locked.
Treat the transcript as a map of the recording. Cut from it, then let the timeline handle the fine finishing.
