Questions, answered —
no customer service dance.

A video-to-text toolkit for YouTube, TikTok, Instagram, and Facebook. You paste a URL, we return a timestamped transcript plus AI summary, translation into 40+ languages, and eight repurposing formats — all from the same paste.

YouTube (including Shorts and live replays), TikTok (long-form and Stories-free video posts), Instagram (Reels, IGTV, video posts), and Facebook (Reels, Watch, Live replays). Four platforms on every plan, including free.

Our internal benchmark suite sits at 98.7% word accuracy. In practice, expect near-perfect results on clear speech and ~92–95% on heavy music, overlapping dialogue, or strong accents.

Yes — word-level timestamps on every line. You can copy text with or without them, or export to SRT, VTT, TXT, or PDF with the timing intact.

Yes — transcription is free to try with no signup. Free accounts get 10 transcripts a month. Paid plans unlock higher volume, HD downloads, batch processing, and the full AI toolkit.

No. Unused credits roll into next month, up to 2× your monthly allowance. If you're on Plus (1,000/mo) you can accumulate up to 2,000 before new ones stop adding.

All major credit cards, Apple Pay, Google Pay, and PayPal — processed through Polar. Invoicing is available on higher-volume plans.

Yes. One click in your dashboard — no retention call. Your credits remain usable through the end of the billing period.

Plain text, text with timestamps, SRT, VTT, TXT, and PDF. SRT and VTT drop directly into Premiere Pro, Final Cut Pro, DaVinci Resolve, and CapCut without re-timing.

Transcripts are read-only in the tool, but you can copy them into any text editor or Notion/Google Docs and edit there. Formatting survives the paste.

The transcription model handles every major language creators post in — 100+ covered. Translation is a separate step into 40+ target languages after you have the source transcript.

Two-sentence AI summaries, chapter stamps for long-form, translation into 40+ languages, and repurposing to eight formats (Twitter threads, LinkedIn posts, blog posts, Reddit posts, IG captions, YouTube descriptions, newsletters, TikTok scripts).

A two-sentence TL;DR plus the key claim. For long-form videos we also generate chapter stamps you can paste into show notes or a YouTube description. One credit covers the transcript and the summary — no double-billing.

Spanish, French, German, Portuguese, Italian, Dutch, Polish, Japanese, Korean, Mandarin, Cantonese, Arabic, Hindi, Hebrew, Turkish, Vietnamese, Thai, Indonesian, Swedish, Norwegian, Danish, Finnish, Greek, Czech, Hungarian, Romanian, and more — 40+ total.

Yes — Talk to Transcript is an AI chat grounded in the transcript. It cites the exact timestamp of any quote it uses, so you can verify the claim and jump straight to that moment.

Yes, on every paid plan. SD and HD, no watermark, no platform UI baked in — alongside the transcript, in the same tab.

They drop a Transcript button right into YouTube, TikTok, and Instagram alongside Like and Share. One click to get the transcript inline — no tab switching, no URL copying.

Yes — install costs nothing and it uses your existing free-tier credits. Sign in to sync history and upgrade to higher-volume plans.

Every Chromium-based browser: Chrome, Edge, Brave, Arc, Opera, and Vivaldi. Install from the Chrome Web Store and it lands in all of them.

Not yet. For Facebook videos, use the web tool at /facebook — same transcript, same toolkit, one extra paste.

Transcription for personal use, accessibility, commentary, and content analysis is generally fair use. Respect copyright when publishing — attribute creators, and don't republish the whole transcript verbatim if the video is paywalled or gated.

Only if you save them in your dashboard history (paid plans). Anonymous sessions process on-demand and are cleared after the response. See the privacy policy for specifics.

Account data (email, name, profile picture) if you sign in; video URL and transcript metadata for usage tracking; payment info through Polar (we never see your card). That's it. No ad networks, no retargeting pixels. See /privacy.

Accuracy drops on heavy music, overlapping speakers, fast speech, or strong regional accents. If you need human-grade accuracy, download the SRT and fix the handful of lines that matter — usually takes a minute or two.

Usually means the video has no speech (music only), the audio is too quiet, or the video is restricted. Try a different video to verify the tool is working, then retry.

Private, age-gated, or region-locked videos return a 400 because the audio simply isn't reachable. If the video plays publicly in a private tab, we can transcribe it.

Timestamps are synced to the audio track. If they look off, the player may be including an intro frame or ad slate that we skip. Relative timing between segments is always accurate.

Pro plans include batch processing — drop in a list of URLs and get all transcripts back as a zip. For programmatic use, see the REST API: one credit per call, 120 req/min, JSON response.

Download the SRT and import it into your NLE (Premiere, Final Cut, DaVinci, CapCut). The cue windows are editor-grade, so it drops in without re-timing.

Currently one transcript at a time. For multi-video research, transcribe each video separately and combine the plain-text exports in a single doc or Notion page.

Use plain text (no timestamps) in the of the page, ideally with H2 section breaks. Search engines index the text as-is and the video gets indexed as a rich snippet alongside.