How to Create SRT and VTT Subtitles from Audio or Video

Author : whisper webdev | Published On : 27 Jul 2026

Subtitles make video easier to understand in noisy environments, more useful to viewers who prefer reading, and more accessible to people who cannot hear the audio clearly. They also help creators repurpose spoken material into searchable text, quotations, and written content.

The difficult part is not creating an empty subtitle file. It is turning speech into accurate, readable, correctly timed segments.

Browser-based speech recognition can produce a strong first draft. With Whisper Web, creators can transcribe common audio and video files and export the result as SRT or VTT without installing a desktop editing suite.

What Is an SRT File?

SRT stands for SubRip Subtitle. It is one of the most widely supported subtitle formats and works with many video editors, social platforms, and media players.

Each SRT caption contains:

a sequence number;
a start and end time;
one or more lines of subtitle text.

An entry looks like this:

1
00:00:02,000 --> 00:00:05,200
Welcome to this transcription guide.

SRT uses commas in its millisecond timestamps. The format is simple, portable, and a good default when a video platform accepts uploaded caption files.

What Is a VTT File?

VTT stands for Web Video Text Tracks, often called WebVTT. It is commonly used with HTML5 video players and web applications.

A VTT file begins with a WEBVTT header and normally uses periods in timestamps:

WEBVTT

00:00:02.000 --> 00:00:05.200
Welcome to this transcription guide.

WebVTT can support additional positioning and styling information, making it a natural choice for browser-based video.

SRT Versus VTT

Both formats connect text to time ranges, but they are used in slightly different environments.

Format	Best fit	Timestamp style	Header
SRT	Editors, social platforms, desktop players	`00:00:02,000`	None
VTT	HTML5 video and web players	`00:00:02.000`	`WEBVTT`

If the publishing platform gives a specific requirement, follow it. If both are accepted, SRT is often the easiest general-purpose option, while VTT is convenient for websites.

How to Generate Subtitle Files

Step 1: Use the Final Audio or Video

Generate subtitles from the version that will actually be published. If the creator changes the edit after transcription, every later cut can shift the timing.

Common supported sources include MP3, WAV, M4A, MP4, WebM, FLAC, OGG, and AAC. A clean final mix with clear dialogue will produce better captions than audio with loud music or overlapping speakers.

Step 2: Open the Transcription Tool

Open Whisper Web in a modern browser and add the authorized file. Local mode performs speech recognition on the device through WebGPU or WebAssembly and does not require an account.

The first use may take a moment because the browser downloads the selected model. Later sessions can reuse the cached model.

Step 3: Choose the Language and Model

Whisper supports more than 100 languages and can detect the spoken language automatically. Manually select the language when the recording is short, begins with music, or contains only a few words.

A smaller model is faster on limited hardware. A larger model may improve difficult audio but requires more memory and processing time.

Step 4: Generate the Transcript

Start transcription and allow the tool to process the complete file. The result should include text segments and timestamps that can be converted into a subtitle file.

Step 5: Export SRT or VTT

Choose the format required by the target player. Save the file with a clear name, such as:

interview-en.srt
product-demo-es.vtt

A language code in the filename helps when several caption tracks belong to the same video.

Why Automatic Subtitles Still Need Review

Speech recognition can save time, but subtitles have requirements beyond word accuracy.

Check Names and Specialist Terms

Product names, people, places, acronyms, and technical vocabulary are frequent error points. Compare them with an approved glossary or the source material.

Check Numbers and Negations

A wrong date, price, dosage, percentage, or missing “not” can change the meaning. Review these passages against the audio.

Improve Line Breaks

A subtitle should be readable at a glance. Avoid leaving a short article or preposition alone on a line. Break at natural phrase boundaries when possible.

Keep Segments Short Enough to Read

Do not place a full paragraph on screen for two seconds. Split dense speech into smaller units while keeping complete ideas together.

Check Timing

Captions should appear when speech begins and disappear after the phrase ends. Correct segments that arrive too early, remain too long, or overlap the next speaker.

Add Meaningful Non-Speech Audio

Accessible captions may need information such as [music], [laughter], or [alarm sounds] when the sound is important to understanding the scene.

A Subtitle Review Checklist

Before publishing, confirm that:

the caption file matches the final edit;
the selected language is correct;
names, numbers, and quotations have been checked;
segments do not cover too much text at once;
line breaks follow natural phrases;
timestamps do not overlap incorrectly;
important non-speech audio is represented;
the SRT or VTT file loads in the target player;
the character encoding preserves punctuation and non-English text.

The most reliable test is to watch the entire video with captions enabled. Skimming only the first minute can miss timing drift or an export problem later in the file.

Common Subtitle Use Cases

YouTube and Social Video

Uploaded subtitle files give creators more control than relying entirely on platform-generated captions. A reviewed file can improve names, punctuation, and specialized vocabulary.

Courses and Tutorials

Captions help learners follow technical explanations and search for a specific step. The same transcript can also become written lesson notes.

Podcasts with Video

Podcast creators can reuse the transcript for captions, show notes, articles, and quotation clips. One reviewed source reduces duplicated editing work.

Interviews and Research Media

Timed text makes it easier to return to an exact answer. Sensitive or identifying details should be reviewed before the caption track is shared.

Website Video

VTT files integrate naturally with HTML5 video. They can provide a caption track without permanently burning text into the video image.

One Transcript, Several Outputs

Subtitle generation is more useful when it becomes part of a broader content workflow. The same reviewed transcription can produce:

an SRT file for a video platform;
a VTT track for a website;
a TXT transcript for readers;
a JSON file for an application;
show notes or a blog outline;
translated caption drafts for other languages.

The guide to generating SRT and VTT subtitles with AI covers additional workflow details. Podcasters can also use the guide to free podcast transcription to turn the same recording into searchable content.

Start with a Reviewable Draft

Good subtitles combine automation with human judgment. Speech recognition handles the first pass, while the creator verifies wording, timing, line breaks, and context.

If you have an audio or video file ready, open Whisper Web to generate a transcript and export SRT or VTT captions directly from the browser. Review the final file in the target player before publication, and keep the source recording available until every important detail is confirmed.