Read the Video Instead of Watching It

A recorded talk is a strange artifact: an hour of someone's thinking, locked into a format you can only consume at their pace. You cannot skim it. You cannot search it. Finding the one sentence you half-remember means dragging the scrubber back and forth until you stumble onto it. The transcript is the same content in a form that answers to you instead.

YouTube knows this, which is why the transcript panel exists — under the video on desktop, and on mobile behind the video title, inside the expanded description, down past everything else. It is genuinely useful for jumping to a moment. It is not built for reading — and on phones it is not built for copying either, since selection is blocked in that panel (a separate problem with its own workarounds).

The other thing the panel cannot do is play. That sounds odd for a video app, until you notice what people actually want: not the original audio — YouTube can already keep that going — but the words, read at their own speed, with their place kept. A lecture at 2× with the current line highlighted is a different activity from watching at 2×. You can look away, come back, and know where you are.

CastReader takes the caption track and does three things to it. The cues get merged into paragraphs, because a wall of one-line timestamps is not prose. The text stays selectable, so quoting a source takes seconds rather than the twenty to forty minutes people report spending on manual mobile copying. And every paragraph keeps its timestamp, so hearing something that needs checking means one tap back to the video at that exact second.

The honest limitation is the caption track itself. Where a creator uploaded real captions, the text has punctuation and speaker turns and reads well. Where only auto-generated captions exist, there is no punctuation and no speaker labels — a single-presenter lecture survives that, a fast four-person podcast does not. CastReader always prefers a human-written track when the video has one, in your reading language first, then English, and only falls back to auto-generated captions when there is no other choice.

Where there are no captions at all, nothing can be done, and it is better to say so than to spin. The text comes from the caption track, not from listening to the audio — this is a caption reader, not a transcription service. Before you try, you can check in ten seconds: open the video, tap the title to expand the description, and look for "Show transcript". If it is not there, no caption-based tool will help you.

Two more boundaries worth stating plainly, because the neighbourhood is full of tools that blur them. This does not download anything: no video file, no audio file, no MP3. Speech is generated from caption text, and the video and audio streams are never played or retained. And it only reaches public videos — anything behind sign-in, age verification, membership or a regional lock stays out of reach, by design rather than by oversight.

Who this ends up mattering for: students working through a semester of recorded lectures, researchers pulling quotes out of conference talks and interviews, anyone learning a language who follows written text more easily than speech, and people who simply have more videos saved than hours to watch them. For all of them the video was never the point — the content was.