Sources & Content Library · Add Sources

Supported file types and size limits

Full table of what you can upload to Ritsu, plus per-file size caps for Free and Pro plans.

3 min read

Ritsu accepts most common study materials. Here's the full list, what gets extracted, and the per-file limits.

Supported types

Type Extension What we extract OCR?
PDF .pdf Text, headings, tables, figure captions Yes (for scanned)
Word .docx Text, headings, comments
PowerPoint .pptx Slide text, speaker notes
Plain text .txt, .md Verbatim
HTML .html, web URL Article body (boilerplate stripped)
YouTube URL Transcript (auto-captions or Whisper fallback)
Audio .mp3, .m4a, .wav Whisper transcription
Video .mp4, .webm Audio extracted + transcribed
EPUB .epub Text + chapter structure
Pasted text (none) Verbatim

Size + length limits

Limit Free Pro
File size 50 MB 200 MB
PDF pages 500 2000
Audio / video duration 60 min 8 hours
YouTube duration 60 min 8 hours
Pasted text length 100 KB 500 KB
Total storage 1 GB 50 GB
Sources in library 50 Unlimited
Imports per day 10 Unlimited

Free plan limits reset at midnight UTC. Hit a daily limit and the next upload is queued until tomorrow — no work lost, just waiting.

What we DON'T support yet

  • Live streams (use the recording after the stream ends)
  • Encrypted / DRM-protected files (Audible audiobooks, ePubs from Apple Books, etc.)
  • Protected video platforms (Coursera, Udemy DRM streams) — paste the transcript text directly instead
  • Spreadsheet-style content (.xlsx, .csv) — paste as text or convert to PDF first
  • Image-only content (.png, .jpg) — for OCR-only flows, use the OCR utility in your OS first then paste text

File quality matters

A clean, well-structured source produces better PoKs than a messy one. Specifically:

  • Searchable PDFs beat scanned PDFs (no OCR errors)
  • Original transcripts beat auto-captions (creator-uploaded subtitles are usually 95%+ accurate)
  • Single-language sources beat mixed-language (Ritsu handles mixed but auto-detection sometimes flips mid-source)
  • Headings + chapter structure in the source produces sharper PoK extraction

Tips

  • Convert obscure formats to PDF first (most apps export to PDF) — that's our most-tested path.
  • For lecture audio, the speaker quality matters more than the file format. Clean mic audio = clean transcript.
  • For long sources, consider breaking them up. A 1500-page textbook works on Pro, but you'll get better PoKs uploading two chapters at a time.

Trouble

  • "Unsupported format" — verify your file extension matches the table. Some platforms add hidden extensions (.pdf.gz, .docx.zip).
  • Upload starts then fails — usually a size issue. See Upload failed.

Was this article helpful?