Audio or video to SRT

Upload the audio or video and download the .srt with the actual timings

Word-by-word timestamped transcription: the subtitle appears when it is spoken, not when a formula calculates it should be.

Typing the subtitles while listening to the audio three times is the slowest part of editing. Here you upload the file and receive the .srt with the timings measured against the voice.

You choose the audio or video from your mobile or computer —MP3, M4A, AAC, WAV, WebM, OGG, FLAC, MP4, MOV, MKV, AVI or 3GP— and upload it. Before transcribing anything, the server measures the actual duration of the file with ffprobe and shows it to you along with the minutes it will consume: if it exceeds the limit of 15 minutes or 20 MB per document, it stops there and informs you, without spending a minute of transcription.

The transcription is done by Deepgram Nova-3 in pre-recorded mode, requesting the timestamp for each word. With those timestamps, the same deterministic engine that segments the scripts distributes the text into titles: each subtitle starts at the moment its first word begins and ends when the last one finishes. There is no estimation based on reading speed or invented roundings.

You control the segmentation: choose one of the five presets —Balanced, Vertical, Comfortable, One line, or Karaoke— or manually adjust the characters per line (12-60), lines per title (1-3), characters per second (6-30), and the minimum and maximum duration of each title and the gap between them. If the distribution doesn't convince you, you can re-segment the already saved words as many times as you want, and that costs nothing: the audio is transcribed only once.

You get the result in .srt (SubRip, validated to the byte before download: CRLF, numbering from 1, comma in milliseconds, and non-overlapping cues), in .vtt for web or in .txt for review. You can transcribe in six languages —Spanish, English, French, Italian, Portuguese, and German— and the work remains in your history to open, rewind, or delete it. Transcribing consumes a pay-per-minute audio service, so this function requires an account: creating it is free.

No sign-up

Try Audio or video to SRT right here

Loading the tool…

Use it free without an account. If you want to save your result, you can sign up afterwards.

Save / Sign up

Benefits of Audio or video to SRT

Measured timings, not calculated

Deepgram Nova-3 returns the timestamp of each word and each title starts and ends at the exact moment it is spoken.

Audio and video, almost any format

MP3, M4A, AAC, WAV, WebM, OGG, FLAC, MP4, MOV, MKV, AVI, and 3GP, up to 20 MB and 15 minutes per document.

You know the cost before you spend it

when uploading the file, its actual duration is measured with ffprobe and the minutes to be transcribed are shown; if it exceeds the limit, it is cut off before calling the engine.

Rewinding is free

change the preset or parameters and the distribution is recalculated based on the already saved words, without having to transcribe again or pay for the audio again.

Segmenting to your liking

five presets and six parameters —characters per line (12-60), lines per subtitle (1-3), characters per second (6-30), minimum duration, maximum duration, and gap between subtitles.

Six audio languages

Spanish, English, French, Italian, Portuguese, and German, choosing the language before transcribing so the model does not have to guess.

Clean import file

validated .srt to the byte —CRLF, numbering from 1, comma in milliseconds, cues without overlaps— plus .vtt and .txt from the same work.

Honest when it can't be exact

if the transcription does not have word marks, the text is segmented by reading speed and the work is marked as estimated time instead of disguising it.

Your audio is now a subtitle file with real timestamps

Upload the file, check the minutes, transcribe, and download the .srt. The account is free and you can repeat the splitting at no cost.

In the features that use AI, responses are generated by artificial intelligence and are not a substitute for professional advice.

Transcribe audio or video to SRT with real timestamps