Skip to main content

Is it possible to segment by sentences?

Feature Request

When I use generated transcripts, the quality and segmentation isn’t quite what I’d like it to be.

I have a script that uses Elevenlabs to transcribe by characters and pieces it back together as sentences with a language model so I can get logical sentence and word level segmentations in a video transcript. The only problem is I don’t have the youtube URL → audio component… Any suggestions?

Posted by isaac ·
1 comment

Log in to comment and vote

Comments1

  • tcytcy111

    •

    Jan 22

    00:00:02 I need a new wallet. When I was 10 years

    00:00:07 old, I got my tonsils removed.

    00:00:13 Bedside in the hospital, my Nana gave me a

    00:00:18 wallet. "You're a man now," she said. After 14

    00:00:23 years, that wallet started to fall apart.

    00:00:28 Everything kept falling out. For a while, I

    00:00:33 used a Ziploc bag. I liked the zipper. But

    00:00:38 it didn't fit so well in my pocket, and as

    00:00:43 Nana would say, it wasn't very gentleman-like.

    00:00:49 Then I treated myself to a fancy new wallet. Just like the plastic bag, it

    00:00:55 too had a zipper. It was really great, but... So, last week I ran out

    00:01:00 of my apartment and like I do every day, carried my bike down four

    00:01:05 flights of steps and out the door. I was in a rush to meet Candace

    00:01:10 for coffee, where we were supposed to discuss our upcoming RV trip.

    00:01:18 So I ordered two iced lattes which were $10.13 only to reach for my

    00:01:23 wallet and discover I didn't have it. Now Candace is pissed off because she

    00:01:28 doesn't have her purse. So I have to get back on my bike, ride 10

    00:01:32 blocks home, leave the bike unlocked, back up the four flights, I'm searching my apartment

    00:01:38 looking for the wallet, and I eventually find it right on my kitchen table.

    00:01:46 Then it's back out the door, down the steps, on the bike,

    00:01:51 back to the coffee shop where Candace is waiting. I

    00:01:57 pay the man and our once delicious iced lattes are now

    00:02:02 all watered down. All this because I didn't see my

    00:02:08 dark wallet on my dark kitchen table. I need a new wallet. The

    00:02:14 problem isn't the wallet. It's the color of the wallet.

    00:02:19 My new wallet, I need a bright wallet. And that's exactly

    00:02:25 what I got.

    00:02:38 It's nearly

    00:02:41 perfect. It

    00:02:44 needed only a

    00:02:47 few

    00:02:50 adjustments. No

    00:02:53 self-respecting

    00:02:57 man uses a

    00:03:00 change purse.

    00:03:03 And of course,

    00:03:06 no possession

    00:03:09 is complete

    00:03:12 without a

    00:03:16 return notice.


    The segmentation of the transcript is just so bad that i can’t use it with my app. If it could return word-level timestamps, i can piece them together into coherent sentences.