Launched this week
Scriber — YouTube transcripts, instantly
YouTube transcripts that work even without captions
15 followers
YouTube transcripts that work even without captions
15 followers
Most transcript tools just copy captions, so they fail on the videos that don't have any. Scriber uses captions when they exist and transcribes the audio when they don't. Paste a link, get clean, copy-ready text in seconds. No sign-up to try.
A speaker label toggle would be great for podcast clips with multiple people talking. Could also be nice to export directly to a doc or markdown instead of just plain text. Otherwise, super handy for pulling quotes from interviews that never got captions uploaded.
@batsemra22286 Thanks Semra, this is genuinely useful.
Markdown export is the easy one, and honestly it should have been there already. Right now you get plain text with optional timestamps, and there's no reason it can't hand you clean markdown you paste straight into Notion or Obsidian. That one I'll look at first.
Speaker labels are a bigger job. The model I use is very good at words but doesn't separate who is speaking, so that needs a different engine underneath, not a toggle. Not a no, but I won't pretend it's quick. Noting it properly, because interviews and podcasts are exactly where it matters most.
Really glad the no-captions part is landing. That was the whole reason I built this, those interview uploads are the ones every other tool chokes on.
One question if you have a second: when you pull a quote, do you mostly need to know who said it, or is a timestamp enough to find it again in the video? That would tell me which of the two actually matters more.
@batsemra22286 Update: markdown export is live, under the Export button next to Copy.
It gives you the title, a source link, and every timestamp as a clickable link straight to that second of the video, so a quote you paste into Notion stays verifiable in one click. That last part came from another commenter today who wanted to jump back to specific moments, and it turned out to be the thing that makes markdown genuinely useful rather than just tidier text.
Speaker labels are still the harder one, and still on the list. :)
A download button for .txt or .srt files would make this way more useful for me, since I often pull transcripts into Notion or Subtitle Edit. Right now I'd have to copy and reformat everything by hand.
@betls8fo Thanks Betül, this one is on me.
The .txt download already exists, but the button was just an icon next to Copy with no label, so it's no surprise you didn't spot it. That's a design failure, not a missing feature. I've labelled it properly and it's live now, so you should see a "Download" button next to Copy.
.srt is a genuine gap though. Right now you get plain text, with timestamps if you toggle them on, but not proper subtitle formatting with numbered cues and start/end timecodes. Subtitle Edit is exactly the case for it, so I'll add it. Someone else asked for markdown, so I'll probably do a proper export menu rather than keep bolting on buttons.
One question, since it changes how I build it: when you take a transcript into Subtitle Edit, are you mostly correcting the text or re-timing it? That tells me how tight the timecodes need to be.
@betls8fo Update: .srt export is live. You'll find it under the Export button next to Copy, alongside plain text and markdown. Cues are numbered with proper start and end timecodes, so it should drop straight into Subtitle Edit.
Still curious about the earlier question, since it affects what I do next: are you mostly correcting text in Subtitle Edit, or re-timing? If it's re-timing, cue lengths matter and I'd want to add a max-characters-per-cue option rather than inheriting whatever YouTube or the model produced.
Congrats on the launch, @marcos_nahr ! Love that it transcribes audio when captions aren't available. that must open up a lot of content. Can this be extended to other non YT formats as well?
@xterra1111 Thanks Harish. That's the part I find most useful too, since a lot of the good stuff was never captioned in the first place.
Right now it's YouTube links only. Direct file upload is what I'm building at the moment, so you'd be able to drop in an audio or video file instead of pasting a link. Other platforms are realistic later on, because once there's audio the rest of the pipeline doesn't care where it came from.
What are you working with? If it's podcast files or recorded calls I'd rather build for that specifically than guess.
@xterra1111 That's useful, thanks. Meeting recordings are well covered by what I'm building, because Zoom and Teams both let you download the recording as a file, and that file is exactly what upload will take.
Speaker labels went live today, which matters more for meetings than for anything else I've built. You get who said what instead of one block of text.
A Teams or Zoom integration is a bigger piece of work than it looks, mostly permissions and app review rather than transcription, so I'm not planning one for now. If downloading the recording turns out to be the annoying part, rather than the transcribing, tell me and I'll rethink that.
Finally tried it on a podcast clip with no captions and it pulled the audio just fine, which is usually where other tools give up. Clean output too.
@kaltal73447 Thanks Gönül, that's exactly the case I built this for.
Podcast clips are the worst offenders. Long, conversational, and very often uploaded with no captions at all, so tools that only copy existing captions have nothing to work with. Good to hear the audio path handled
it cleanly, because that's the part that took the most work to get right.
If you ever hit one it struggles with, send me the link. Those are the most useful bug reports I get.
The no sign-up trial is a nice touch, got my transcript in seconds on a captioned video. One thing that would seal it for me: timestamps alongside the text so I can jump back to a specific moment when I'm quoting a video for a newsletter.
@levent39344 Thanks Levent, and you found a real gap.
Timestamps were already there behind a toggle next to Copy, but two things were wrong. The toggle didn't look like a button, so it was easy to miss, and more importantly the timestamps were just text. You could see the moment but not actually get to it, which is the whole point when you're checking a quote.
Both are fixed and live now. Turn on Timestamps and every one is a link that opens the video at that exact second, so you can verify a quote in one click before it goes in the newsletter.
You're the second person today to ask for something that technically already existed, which tells me the output controls needed to be more obvious rather than more numerous. Genuinely useful, thanks.
pulled a transcript from a video i've been stuck on for weeks, no captions at all, and it came back surprisingly clean with the actual speaker names intact. honestly kind of wild how fast it ran.
@marypricerraf Thanks Mary, really glad it got you unstuck on that one.
One honest correction though, because I don't want to take credit for something the tool doesn't actually do. Scriber doesn't identify speakers. If names showed up in your transcript, it's because they were said out loud in the audio and the transcription picked them up as ordinary words, not because it worked out who was talking.
Proper speaker labels are a genuinely different feature and need a different transcription engine underneath. Someone asked for it earlier in this thread, so it's firmly on the list, but I'd rather tell you now than have you hit a video where nobody says a name and think something broke.
The no-captions part and the speed are real though, and that's the bit that took the most work. Glad that landed.