What Are Word-Level Timestamps in a Transcript?

What are word-level timestamps? They are timestamps given for each word instead of for each sentence or paragraph. Every word carries its own start and end time in the recording. This allows exact jumps to a word, karaoke-style subtitles and precise cuts when editing audio.

Transcribe a recording

Send your voice message, audio or video to @potok_now_bot on Telegram and the text comes back in the chat.

Open @potok_now_bot in Telegram

Files up to 2 GB · export to TXT, MD, DOCX, PDF, SRT and VTT · speaker labels and timestamps · about 30 languages

Key facts

Word times are often found by alignment, a step that matches the recognised words to the sound and marks where each begins.
The output is large compared with plain text, since each word needs two numbers, so it is stored as data such as JSON.
Editors use word times to cut out a filler or a mistake by deleting the word in the text and the matching sound.
Fast speech, mumbling and overlapping voices make the edges of words blurry, so the times can be off by a fraction of a second.
Most readers do not need this level of detail, and paragraph or sentence marks are enough for notes and quotes.

What it looks like

{
  "words": [
    { "word": "send",    "start": 12.40, "end": 12.62 },
    { "word": "the",     "start": 12.62, "end": 12.70 },
    { "word": "invoice", "start": 12.70, "end": 13.18 }
  ]
}
// start and end are seconds from the beginning of the file

See the full list of languages, the tools and the glossary.

Frequently asked questions

When do I need word-level timestamps?

For editing by text, highlighting the current word during playback and building tight subtitles. For normal reading, sentence marks are enough.

Are word-level timestamps always accurate?

No. They can drift in noisy or fast speech. Spot-check a few words against the audio before relying on exact cuts.

Do SRT and VTT use word times?

Not by default. They store a time for each cue, which is a group of words. Word times need a data format or a special subtitle style.

Does Potok add timestamps?

Timestamps are optional in the bot and can be switched on before sending a file. Open an export to see how finely they are placed.

How do word times differ from sentence times?

A sentence time says when a sentence begins. Word times give a start and end for every word inside it.

Sources

Each source was opened by hand on the date shown. Pages change, so open the source if a detail matters to you.

Forced alignment (Wikipedia), Wikipedia · checked 2026-10-18
Speech recognition (Wikipedia), Wikipedia · checked 2026-10-18

Try it on your own recording