Paste a meeting transcript, interview, speech or script to get the word count, how long it takes to say out loud, how long it takes to read, an estimate of AI tokens, and words per speaker when lines start with a name. It runs 100% in your browser, and nothing is uploaded.
Sources: reading rate from Brysbaert (2019), Journal of Memory and Language; token rule of thumb from OpenAI's help center.
See how the conversation was split with the meeting talk-time calculator, or tidy the text with the transcript cleaner.
Need every transcript, not just one? MinuteHaul exports all your tl;dv meetings into one ZIP (Markdown per meeting + CSV index + one combined file for ChatGPT). Try it free on your 10 most recent meetings or pre-order unlimited: $29 one-time.
Word count divided by the speaking rate you set. The default is 150 words per minute, a common planning figure for conversational speech. Change it to match your own pace; slow presenters may be nearer 120 and fast talkers 170 or more.
From Marc Brysbaert's 2019 meta-analysis of reading rate in the Journal of Memory and Language, which estimated about 238 words per minute for silent reading of English non-fiction by adults. You can change it.
OpenAI's guidance is that one token is roughly three quarters of an English word, so the tool divides the word count by 0.75. Real counts vary by model and language, so treat it as a ballpark for whether a transcript fits a model's context window.
No. Lines with "-->", cue numbers, the WEBVTT header and tags like <v Dana> are ignored, so only spoken words are counted.
No. Counting happens in your browser and the text never leaves your device.