How to Transcribe Audio With TurboScribe in 4 Steps

November 18, 2026

TurboScribe turns audio and video files into text transcripts with Whisper large-v3, and the free tier gives you 3 files a day with a 30-minute cap per file. Upload a recording, pick a language, and export SRT, VTT, TXT, or DOCX once the job finishes.

What is TurboScribe?

TurboScribe is a browser-based transcription service built on OpenAI’s Whisper models. It launched as a solo project by Loïc Magnien in 2023, shortly after Whisper weights went public. There is no desktop app, no CLI, and no API on the standard plan — everything runs through the upload page.

That matters. Most competitors push you toward a paid developer API, while TurboScribe is aimed at people who just have a folder of MP3s and a deadline.

One thing to know up front: the free tier is real but narrow. Three uploads per day, 30 minutes each, 1080 minutes total on the paid unlimited plan per month if you go that route. Files above 30 minutes on free require splitting first, or you get an error before the job starts.

How do you transcribe a file with TurboScribe?

The workflow is four steps, and none of them involve a blog post’s worth of setup.

  1. Open the TurboScribe upload page and drag a file in. Supported formats include MP3, MP4, M4A, WAV, WMA, FLAC, OGG, WEBM, and video containers like MOV.
  2. Set the language. Auto-detect works for most European and East Asian languages but drifts on code-switched audio; if you know the language, pick it manually.
  3. Choose transcription quality. The default is Whisper large-v3, which is the accurate one. Lowering it speeds things up and costs you punctuation quality.
  4. Pick an output format — plain text for reading, SRT or VTT for subtitles, DOCX for editing in Word — and download when the progress bar clears.

A 20-minute interview usually finishes in 60 to 90 seconds at peak times, longer when the queue is deep. You can leave the tab; jobs keep running server-side and show up in your file list.

What does TurboScribe cost in 2026?

The pricing is simple, and it is the main reason people pick it over a $20/month subscription elsewhere.

Plan Price Limits
Free $0 3 files/day, 30 min each, 1080 min/month cap
Unlimited About $10/month billed yearly 10 files/day upload, unlimited total minutes
Unlimited (monthly) About $20/month Same, cancel anytime

The yearly Unlimited tier is the one most people buy. Ten files a day is generous for a podcaster, tight for someone processing a 40-episode archive in a weekend.

How accurate is it, really?

Accuracy depends more on your audio than on the tool. Clean two-person interviews with decent microphones land around 95-98% on English, which matches Whisper large-v3 benchmarks. Phone recordings with background music, overlapping speakers, or heavy accents drop to 85-90% and need cleanup.

Numbers and proper nouns are where Whisper models fail most. “Thirty percent” becomes “30%” or “thirty person,” and product names get mangled. Plan on a read-through pass no matter what any vendor claims.

Tip that saves time: enable speaker labels in settings before uploading if you have a two-person conversation. It costs no extra minutes and saves an hour of manual attribution later.

What are TurboScribe’s weak spots?

  • No official API. If you need to pipe transcriptions into n8n, Make.com, or a Python pipeline, this is the wrong tool.
  • Web-only. There is no offline mode and no mobile app; the site works in mobile Safari but uploading large files from a phone is painful.
  • No built-in editor for real transcript tidying. You can fix a word inline, but there is no timeline scrubbing to jump to a timestamp the way Descript or Riverside offer.
  • Speaker labels are decent but not speaker-diarization grade. Three or more speakers in a noisy room get merged.

Who should use it — and who should not?

Use TurboScribe if you are a podcaster, journalist, researcher, or student who needs transcripts of files you already have and does not want to touch an API key. The free tier genuinely covers a few interviews a week.

Skip it if you need programmatic access, real-time transcription, or you are processing hundreds of hours a month — at that volume, Whisper via a cloud provider usually costs less and gives you the raw JSON you need.

Quick Summary

  • TurboScribe is a web app for transcribing audio and video with Whisper large-v3, launched in 2023.
  • Free plan: 3 files per day, 30 minutes each, 1080 minutes per month total.
  • Paid Unlimited runs about $10/month yearly or $20/month month-to-month.
  • English accuracy sits around 95-98% on clean audio; numbers and names need manual fixing.
  • No API and no offline mode — it is built for people, not pipelines.