# Paper Edit from a YouTube Transcript (SRT)

Source: https://b-roll-6hk.pages.dev/guides/youtube-transcript-paper-edit/
Published: 2026-09-25
Updated: 2026-09-25
Author: Ema Zamboni, video editor (https://b-roll-6hk.pages.dev/about/)
Section: Guides (https://b-roll-6hk.pages.dev/guides/)

**In short:** A YouTube video's subtitle file is already a timecoded transcript for a paper edit. Download it as an .srt with B-ROLL, read it as a document, mark the lines you want, then use their cue timings as in and out points to download only those selects as named clips.

## What is a paper edit, and why start from a YouTube transcript?

A paper edit is a written plan of a cut: the quotes you want, in story order, each with its source timecode. A YouTube subtitle file is already a timecoded transcript for a paper edit, so you can plan a cut from a two-hour interview by reading instead of scrubbing.

Reading finds a quote faster than playback, and a transcript lets you reorder quotes on paper before any clip hits the timeline. The SRT gives you that document for free whenever the video has subtitles or auto-generated captions. The workflow has four stages: download the SRT, clean it, mark selects, and download only those ranges as clips.

The table shows how three selects from an interview travel from SRT cue to clip file, with a second of handle added on each side.

| Select | SRT cue timing | In point | Out point | Clip file name |
| --- | --- | --- | --- | --- |
| "We shot the whole thing on one lens." | `00:02:10,480 --> 00:02:14,920` | 2:09 | 2:16 | `Interview [02-09 to 02-16].mov` |
| "Nobody told us the permit had expired." | `00:14:52,100 --> 00:15:03,640` | 14:51 | 15:05 | `Interview [14-51 to 15-05].mov` |
| "That was the last day we filmed." | `01:05:30,200 --> 01:05:41,700` | 1:05:29 | 1:05:43 | `Interview [05-29 to 05-43].mov` (the hour is dropped) |

> **Careful:** Download only footage you have the right to use: your own uploads, Creative Commons, licensed or permitted material.

[Download B-ROLL for Windows](https://b-roll-6hk.pages.dev/download/): free, Windows 10 and 11 today, macOS and Linux planned. No account, no ads.

## How do you get a timecoded transcript from a YouTube video?

B-ROLL saves a YouTube video's subtitles as a separate `.srt` file when you set the **Subtitles** menu, next to Format, to **SRT**. The file lands next to the video as `Title.lang.srt`, for example `Interview.en.srt`.

Download a light reference copy for this step, an MP4 at 720p or 480p, because the paper edit only needs the text and a picture to check it against. Before the download starts, the **Subtitles** dialog lists the languages the video really has, split into **OFFICIAL, uploaded by the creator** and **AUTO-GENERATED**. Every option of the menu and the language picker is covered on the [YouTube subtitle downloader for SRT, ASS or VTT](https://b-roll-6hk.pages.dev/features/youtube-subtitle-downloader/).

Prefer the official track when there is one. As of September 2026, [YouTube Help on automatic captioning](https://support.google.com/youtube/answer/6373554) warns that automatic captions can misrepresent speech because of accents, dialects or background noise. What to expect from those tracks is in [download YouTube auto-generated captions as SRT](https://b-roll-6hk.pages.dev/guides/download-auto-generated-captions/).

## How do you clean an SRT file into a readable transcript?

An SRT file is plain text built from numbered cues: a cue number, a timing line with a start and an end joined by `-->`, one or two lines of text, and a blank line. Any text editor opens it, and a few passes turn it into a transcript.

```text
41
00:02:10,480 --> 00:02:14,920
We shot the whole thing
on one lens.
```

Delete the cue numbers and the end times, then join the text lines into paragraphs at each change of speaker or topic. Keep the start time of every paragraph in square brackets, for example `[00:02:10]`, so each block of text still points to a place in the video. Save the result as a new file and keep the original `.srt` untouched.

Auto-generated captions need an extra pass. Their cues often repeat words from the previous cue and carry no punctuation, so read the text against the video wherever a quote matters. The differences between the three subtitle formats B-ROLL writes are in [SRT vs VTT vs ASS: which subtitle format to keep](https://b-roll-6hk.pages.dev/formats/srt-vs-vtt-vs-ass/). If you picked VTT, the timings use a period before the milliseconds instead of a comma, as the [W3C WebVTT specification](https://www.w3.org/TR/webvtt1/) defines.

## How do you mark selects in a transcript?

A select is a quote you intend to use, marked in the transcript with its in and out timecode. Highlight each select, number it in the order the story needs, and write its in and out points in the margin.

Take the in point from the start timing of the first cue and the out point from the end timing of the last cue. Round the in point down and the out point up to the whole second, then add a second on each side as handle, so a breath or a cut on a word has room. That is how `00:02:10,480 --> 00:02:14,920` becomes 2:09 to 2:16 in the table of selects.

Build the order on paper before downloading anything. A numbered list of selects reads like a radio edit, and weak joins show up while changing them costs a line of text. Collecting from many videos at once, with a link list and a credit for each moment, is covered in [how to collect and track clips for a video essay](https://b-roll-6hk.pages.dev/guides/video-essay-footage-workflow/).

## How do you turn transcript selects into downloaded clips?

B-ROLL downloads each select as its own clip once you set **Advanced options** to **Only a part, ask me which each time** and paste the video link again. Each range becomes its own file named `Title [mm-ss to mm-ss].ext`.

First set **Subtitles** back to **None**. Subtitles fetched with a clip still cover the whole video, according to the app's code, so leaving SRT on only repeats the transcript you already have. Then pick **ProRes** or **DNx** as Format. Ema Zamboni built that conversion because 4K60 AV1 downloads were heavy to play back in DaVinci Resolve, and AV1 or VP9 files crashed it at times.

With VLC media player installed, the preview and trim player opens. Click the timeline near a select's in point, fine-tune with `J`, `K` and `L`, press `I` and `O`, then `Enter` to add the part. Repeat for every select in that video, then click **Download (n)**. The player sets points to the whole second, so the rounded times from your paper edit fit it exactly.

Without VLC, the **Download which part?** prompt takes one Start and End per paste, so you paste the link once per select. Type `2:09`, never a bare `2`, which B-ROLL reads as two minutes.

Clip names keep minutes and seconds only, so a select past the first hour loses its hour, as in the last row of the table of selects. Two selects with the same minute and second in different hours get the same name, and the second counts as already downloaded; shift one in point by a second. The player keys and cut modes are on the [YouTube clip downloader with preview and in and out points](https://b-roll-6hk.pages.dev/features/youtube-clip-downloader/).

Every clip also gets its own credits file, for example `Interview [02-09 to 02-16].credits.txt`, with a copy-paste credit line. Check the license on the video page yourself: the file reports what yt-dlp returns and is a convenience, not proof of rights. More workflows like this one sit in the [guides for YouTube footage from download to timeline](https://b-roll-6hk.pages.dev/guides/).

## What goes wrong when a paper edit meets the clips?

A select that starts a moment early comes from B-ROLL's default fast cut. The fast cut snaps to the nearest keyframe, so a clip can begin a little before your in point; tick **Exact frame cut** in Advanced options when the first word must land on the cut.

A clip off by minutes comes from a typed time without a colon: `2` means two minutes, so write `0:02` for two seconds. Both cases are covered in [YouTube clip starts at the wrong time](https://b-roll-6hk.pages.dev/troubleshooting/clip-starts-at-wrong-time/). An empty language list or a missing `.srt` means the video has no track in that language, or only auto-generated captions that you left unticked; the fixes are in [YouTube subtitles not downloading](https://b-roll-6hk.pages.dev/troubleshooting/youtube-subtitles-not-downloading/).

## How to build a paper edit from a YouTube transcript with B-ROLL

1. **Set Subtitles to SRT.** In B-ROLL, open the Subtitles menu next to Format and choose SRT. Set Quality to 720p or 480p for a light reference copy.
2. **Download the reference copy.** Paste the video link and press Enter. In the Subtitles dialog, tick the language you want and prefer the OFFICIAL group when it exists. The file Title.lang.srt lands next to the video.
3. **Clean the SRT into a transcript.** Open the .srt in a text editor, delete the cue numbers, merge the text into paragraphs and keep the start timing of each paragraph in square brackets.
4. **Mark your selects.** Highlight the lines you want, number them in story order and write an in point and an out point for each, with a second of handle on both sides.
5. **Set Subtitles back to None and pick a clip mode.** Set Subtitles to None, choose ProRes or DNx as Format, and in Advanced options choose Only a part, ask me which each time.
6. **Download the selects as clips.** Paste the link again. In the preview player, click the timeline near each in point, press I and O, then Add part. Click Download (n) to get one file per select.

## Frequently asked questions

Q: Can I get a transcript in another language?
A: Only when the video has a subtitle track in that language. B-ROLL's Subtitles dialog lists the languages a video really has, with official subtitles and auto-generated captions in separate groups and a search box. Tick several and each arrives as its own Title.lang.srt file, grouped in a folder named Title - subtitles.

Q: Is a transcript the same as subtitles?
A: No. A transcript is the spoken text laid out as a document, while subtitles split the same words into short timed cues for the screen. An SRT file carries both the words and the timings, so removing the cue numbers and line breaks turns it into a timecoded transcript you can read like a script.

Q: Do the SRT timecodes match a clip I already cut?
A: No. SRT timings count from the start of the full video, and B-ROLL does not cut subtitles to clips, according to the app's code. For a clip with its in point at 2:09, every cue must move 2:09 earlier: slide the subtitle track in your editor by the clip's in point, and tick Exact frame cut so the clip really starts there.

## Sources

1. [YouTube Help: Use automatic captioning](https://support.google.com/youtube/answer/6373554)
2. [W3C: WebVTT, the Web Video Text Tracks format](https://www.w3.org/TR/webvtt1/)

---

B-ROLL is a free, donation-supported desktop YouTube downloader for video editors: it downloads videos, clips, playlists and channels, converts them automatically into edit-ready ProRes or DNxHR (or saves MP4, MKV, WEBM or audio), and includes built-in trimming, subtitles and automatic creator credit files. Part of [Guides: YouTube Footage from Download to Timeline](https://b-roll-6hk.pages.dev/guides/) on the B-ROLL website. Download B-ROLL, free for Windows 10 and 11 (macOS and Linux planned): https://b-roll-6hk.pages.dev/download/