How to Export SRT Subtitles From Any Video or Audio

By whispertranscribe.ai Editorial Team · Last updated: November 2026 · 12 min read

To export SRT subtitles, start from wherever your captions already are. Generate a new file from the audio with an online subtitle generator, download it from YouTube Studio for your own uploads, export it from Premiere Pro, DaVinci Resolve, or Final Cut Pro, or create it locally with Whisper. Then check the file before you upload it.

Below are the steps for each route, the export settings behind most broken files, and a quick check for the finished file. For what each line inside an SRT means, see our guide to what an SRT file is; this one covers getting a clean file out.

Quick answer: 4 ways to export SRT subtitles

Your starting point Best way to export SRT What you need
A video or audio file with no subtitles yet Method 1: generate the SRT from the audio online The file or a link, and a web browser
A video already on your YouTube channel Method 2: download the track in YouTube Studio Access to the channel
An edit with a caption or subtitle track Method 3: export from your video editor Premiere Pro, DaVinci Resolve, or Final Cut Pro
Private recordings that should stay on your computer Method 4: Whisper and FFmpeg Python, FFmpeg, and some setup time

Each route gives you a sidecar text file to upload next to the video. Burning subtitles into the picture is a separate job; see hard vs soft subtitles.

Before you export: 4 settings that decide whether the file works

Most broken SRT files come from export settings, not the words. Check these four in any tool.

  1. Characters per line. Netflix's English (USA) Timed Text Style Guide caps a line at 42 characters, a sensible default for web video too.
  2. Lines per cue. The same guide allows two lines at most and prefers one when the text fits.
  3. Start time. Professional timelines often begin at 01:00:00:00. If the export keeps that timecode, every subtitle shows up an hour late, which is why Final Cut Pro asks you to choose "Relative to Timeline" or "Absolute".
  4. Formatting. SRT has no formatting standard, so Apple recommends deselecting Include Formatting when exporting SRT from Final Cut Pro.

Reading speed matters too: Netflix allows up to 20 characters per second for adult programs. Few export dialogs set it, but the check script below flags fast cues.

Method 1: Export SRT subtitles with whispertranscribe.ai

Use this route when you have a recording but no subtitles. whispertranscribe.ai is a browser-based transcription tool powered by OpenAI's Whisper model that times each subtitle and exports SRT or VTT. Start from the page that matches your file:

Starting page Use it for
AI subtitle generator Any video or audio file, a microphone recording, or a link
MP4 to SRT converter MP4 video files
MP3 to SRT Podcasts, interviews, and other audio-only recordings
Transcribe from URL YouTube, Vimeo, Google Drive, or Dropbox links

The steps use the labels shown on those pages in November 2026:

  1. Step 1: Choose the spoken language from 130+ languages. Many also offer a Noise / Low-Quality Audio/Video variant for hard-to-hear files, and Multilingual Mix covers speakers who switch languages.
  2. Step 2: Upload or drop your file. Use the From files tab for an upload (up to 5GB per file; plans support recordings up to 10 hours), From URL for a link, or Record to capture your microphone.
  3. Set the subtitle options. Turn on Enable Speaker Detection when more than one person talks. The Maximum Words per Subtitle slider runs from 1 to 50 and starts at 8. Lower it if your lines run past about 42 characters.
  4. Click Create Subtitles, then check names, figures, and technical terms. Accuracy is up to 99% on clear audio; noise, strong accents, and crosstalk lower it.
  5. Export SRT or VTT. The page also lists DOCX, PDF, TXT, and CSV if you want a plain transcript too.

Limits, stated plainly: you can start without creating an account, but the page doesn't say whether free use includes the full SRT download, so check the current terms there. Paid plans start from $9.99/month billed annually (pricing). whispertranscribe.ai exports subtitle files only and doesn't burn subtitles into video. Need words without timing? Use the video to text converter.

Method 2: Download SRT subtitles from YouTube Studio

If the captions already sit on your own YouTube video, download them instead of starting over. YouTube's help page on editing captions lists these steps:

  1. Sign in to YouTube Studio.
  2. Select Subtitles in the left menu.
  3. Click the video.
  4. Under Subtitles, click Edit next to the language you want.
  5. Click Options, then Download subtitles. Your browser saves the caption track as a file.
  • Only channels you manage. Studio lists your own uploads, not other creators' videos.
  • Check the extension. YouTube's help page doesn't name the download format. If you get a .vtt file, Method 4 shows how to convert it.
  • Fix automatic captions first. Correct names and punctuation in Studio, publish, then download.

On Vimeo, the ⋮ menu next to each caption file lets you download it, according to the Vimeo Help Center.

Method 3: Export SRT from Premiere Pro, DaVinci Resolve, or Final Cut Pro

Editors export SRT only from a caption or subtitle track. Text added as titles or graphics is part of the picture and won't appear in the file. These steps come from each vendor's documentation.

DaVinci Resolve

The DaVinci Resolve 21 Reference Manual (Chapter 59, Subtitles and Closed Captioning) describes three routes. Each one needs at least one subtitle track:

  • File menu: choose File > Export Subtitle, then pick a location and file type. SRT and VTT are available.
  • Track header: right-click the subtitle track header and choose Export Subtitle.
  • Deliver page: at the bottom of the Video panel in Render Settings, tick Export Subtitle, set Format to As a separate file, set Export As to SRT, and tick the tracks to include. Each selected track becomes its own file.

The first two routes give you the file without rendering the video.

Adobe Premiere Pro

Adobe's Export caption tracks page (updated January 2026) says to select Export in the header bar to open the Export workspace, then choose burned-in, embedded, or sidecar captions. For SRT, pick the sidecar export with the .srt extension. Two details are easy to miss:

  • An .srt file holds one caption stream, so export each language separately.
  • Premiere picks the active track in the timeline by default. With several tracks enabled, it takes the one with the lowest track number, so confirm the selection.

Final Cut Pro

From Apple's guide to exporting captions in Final Cut Pro:

  1. Open the project in the timeline and choose File > Export Captions. The current version of the guide calls it File > Export Closed Captions.
  2. Name the file and choose a folder.
  3. Under Roles, tick each language subrole you need. Final Cut Pro writes a separate file for each one.
  4. Under Start Time, choose Relative to Timeline to subtract the project's start time (usually right for web uploads), or Absolute to keep the original timecode.
  5. In the SRT section, deselect Include Formatting unless you know your player handles it.
  6. Click Export.

Method 4: Export SRT locally with Whisper and FFmpeg

Choose this when recordings shouldn't leave your computer or you want to script a batch. Whisper is OpenAI's open-source speech recognition model, and FFmpeg is a free command-line media tool. Neither is part of whispertranscribe.ai.

The Whisper README gives this basic command, which uses the default turbo model. It comes straight from the documentation; we didn't run it with turbo:

whisper audio.flac audio.mp3 audio.wav --model turbo

By default Whisper writes every output format; add --output_format srt for the subtitle file only. We did run the commands below on November 4, 2026 (openai-whisper 20250625, tiny model, FFmpeg 7.1.5, Linux) on the 11-second sample clip from the Whisper repository, checking file structure and line length, not accuracy.

whisper jfk.flac --model tiny --language en --output_format srt

That produced a valid SRT, but the whole sentence landed in one cue with a 107-character line. Whisper's --help lists line-wrapping options that need word timestamps:

whisper jfk.flac --model tiny --language en --output_format srt \
  --word_timestamps True --max_line_width 42 --max_line_count 2

The second run split the speech into two cues with no line over 41 characters:

1
00:00:00,000 --> 00:00:09,220
And so my fellow Americans, ask not what
your country can do for you, ask what you

2
00:00:09,220 --> 00:00:10,500
can do for your country.

The break after "ask what you" is awkward, which is why a human pass still matters. When subtitles already exist, two FFmpeg commands help:

# Extract the first subtitle track as SRT
ffmpeg -i input.mkv -map 0:s:0 subs.srt

# Convert SRT to WebVTT (or the reverse)
ffmpeg -i subs.srt subs.vtt

Both worked in our test, including extraction from an MP4. One catch: extracted times came out 23 milliseconds late because our test video's container started at −0.023 seconds, so compare a few cues after extracting. This works for text tracks only; image-based subtitles from DVD or Blu-ray rips need text recognition software.

After you export: SRT quality checklist

Run through this for every exported file, whatever tool made it.

  • Encoding is UTF-8. YouTube says SRT files "must be in plain UTF-8", and Vimeo requires UTF-8 so special characters display correctly.
  • Timecodes look like 00:00:04,200, with a comma before the milliseconds, and the first cue starts near the first spoken word, not at one hour.
  • Line length: no line over 42 characters, and no cue with more than two lines.
  • Reading speed: 20 characters per second or less for adult viewers.
  • Order: each cue ends after it starts and doesn't overlap the next.
  • Coverage: the last cue matches the end of the speech. If it stops early, the export or your plan cut it short.
  • Player test: load the file with the video in a media player; watch the first, a middle, and the last minute.

Check an SRT file with a short script

We wrote this Python script for this article and ran it on the files above. It stops with an error if the file isn't UTF-8, then flags bad timing lines, overlaps, long or extra lines, and fast cues. On the first Whisper export it reported "cue 1: 107 characters in one line"; the wrapped export passed.

import re, sys

MAX_CHARS, MAX_LINES, MAX_CPS = 42, 2, 20
TIME = re.compile(r"(\d\d):(\d\d):(\d\d),(\d{3}) --> (\d\d):(\d\d):(\d\d),(\d{3})$")

def ms(h, m, s, f):
    return ((int(h) * 60 + int(m)) * 60 + int(s)) * 1000 + int(f)

text = open(sys.argv[1], encoding="utf-8").read()  # errors if not UTF-8
prev_end = 0
for block in re.split(r"\n\s*\n", text.replace("\r\n", "\n").strip()):
    num, timing, *body = block.split("\n") + [""]
    body = [line for line in body if line]
    t = TIME.match(timing)
    if not t:
        print(f"cue {num}: bad timing line {timing!r}")
        continue
    start, end = ms(*t.groups()[:4]), ms(*t.groups()[4:])
    if end <= start or start < prev_end:
        print(f"cue {num}: ends before it starts, or overlaps")
    prev_end = end
    if len(body) > MAX_LINES:
        print(f"cue {num}: {len(body)} lines")
    for line in body:
        if len(line) > MAX_CHARS:
            print(f"cue {num}: {len(line)} characters in one line")
    if len(" ".join(body)) / max((end - start) / 1000, 0.001) > MAX_CPS:
        print(f"cue {num}: faster than {MAX_CPS} characters per second")
print("check finished")

Run it with python3 srt_check.py subs.srt and adjust the three limits at the top for your platform.

Common SRT export errors and how to fix them

What you see Likely cause Fix
Every subtitle is about an hour late Timeline starts at 01:00:00:00 and the export kept it Re-export relative to the timeline, or shift all cues back
No subtitle export option, or an empty file Text was made as titles, not on a subtitle track Move it to a subtitle track, or generate a new SRT from the audio
<i> or <b> tags show on screen Formatting was exported Re-export without formatting
Accented letters turn into symbols File isn't UTF-8 Re-save as UTF-8 in a plain text editor
One long line covers the picture No line-length limit at export Lower Maximum Words per Subtitle, or use --max_line_width 42 in Whisper
The file ends before the speech does Export cut short, or a plan limit Compare the last cue with the video length and check your plan
Subtitles drift further off over time SRT made from an earlier cut Regenerate it from the final export

SRT or VTT: which should you export?

Export SRT for platform uploads and editors, and VTT for an HTML5 player on your own site. Unsure? Export both.

Where the file goes Export Why
YouTube SRT Suggested for people new to caption files; basic SRT only, no markup (YouTube Help)
Facebook SRT Its help center sets a naming rule for .srt files
Vimeo Either Vimeo supports SRT and WebVTT and recommends WebVTT
A player on your own site VTT WebVTT is the W3C format for timed text in HTML video (W3C)
Premiere Pro, DaVinci Resolve, or Final Cut Pro SRT All three list SRT among their caption formats

Upload the exported SRT: official instructions

Platform Where to upload Do this first Official help page
YouTube YouTube Studio > Subtitles > your video > Add language, then Upload file and choose "With timing" Plain UTF-8, no markup Add subtitles & captions
Facebook Edit the video post and add captions Name it like video.en_US.srt SubRip (.srt) naming convention
Vimeo Open the video > Languages > + > upload, then set Language and Type UTF-8; switch the track on after upload How to add captions or subtitles

Our what is an SRT file guide has click-by-click steps for all three. When Vimeo asks for a Type, subtitles vs captions explains the choice.

FAQs

How do I export SRT subtitles for free?

Use a tool you already have, or run Whisper yourself. For your own YouTube videos, download the caption track in YouTube Studio. If you already edit in DaVinci Resolve, Premiere Pro, or Final Cut Pro, exporting your caption track costs nothing extra. Whisper is open source under the MIT license, so running it locally is free apart from setup time. whispertranscribe.ai lets you start without an account, but its page doesn't state whether free use includes the full SRT download, so check there first.

Can I download SRT subtitles from someone else's YouTube video?

Not through YouTube Studio. Studio only lists videos on channels you manage, so YouTube's download steps apply to your own uploads. If you have permission to reuse another creator's video, you can paste its link into whispertranscribe.ai's YouTube transcription tool and export SRT from the new transcript. Having a subtitle file doesn't give you the right to republish someone else's words or footage, so ask the creator first.

Why is my exported SRT file exactly one hour out of sync?

Your timeline most likely starts at 01:00:00:00, and the export kept that timecode instead of counting from zero. The first subtitle then sits at 01:00:00 in the file while the uploaded video starts at 00:00:00. In Final Cut Pro, export again with Start Time set to Relative to Timeline. In other editors, set the timeline to start at zero before exporting, or shift every cue back by one hour in a subtitle editor.

Should I export SRT or VTT?

Export SRT for platform uploads and editing software, and VTT for video players on your own website. YouTube and Facebook both work with SRT, and every editor in this guide can export it. VTT, short for WebVTT, is the W3C format for HTML video and supports positioning and styling that SRT lacks. Vimeo accepts both and recommends WebVTT. When in doubt, export both; they're small text files.

How many characters per line should an SRT subtitle have?

Aim for 42 characters or fewer per line and no more than two lines per subtitle. Those limits come from Netflix's English timed text style guide, which also caps reading speed at 20 characters per second for adult programs. Other platforms publish their own rules, so check the guide for wherever you deliver. To enforce the limit, lower Maximum Words per Subtitle in whispertranscribe.ai, use --max_line_width in Whisper, or set a line limit in your editor.

Why does my exported SRT show tags or strange symbols?

Tags such as <i> mean formatting was exported, and strange symbols point to the wrong text encoding. SRT has no formatting standard, so YouTube ignores markup and some players print the tags as plain text. Export again with formatting turned off, for example by deselecting Include Formatting in Final Cut Pro. If accented letters show up as symbols like é, open the file in a plain text editor and save it as UTF-8, which YouTube and Vimeo both require.


Need an SRT file from a recording? Upload it to the whispertranscribe.ai subtitle generator, choose from 130+ languages, and export SRT or VTT. See every tool on the features page.

References

Ready to turn your recordings into text?

Transcribe audio and video with whispertranscribe.ai.

Start transcribing ↗