Question 1

How does PDF text extraction work?

Accepted Answer

Your PDF passes through a 5-tier extraction chain. DocLing with Tesseract OCR is tried first (handles 95% of PDFs), followed by PyMuPDF, pdfplumber, LlamaParse, and GPT-4o VisionOCR as fallbacks. Scanned documents are automatically detected and processed with OCR. Large documents are processed in 30-page batches.

Question 2

Are tables preserved?

Accepted Answer

Yes. Tables are extracted into a structured row-linearized format where each row becomes a set of "Header: Value" pairs. Column semantics survive chunking, and table captions are preserved.

Question 3

Is it really free?

Accepted Answer

Yes. The free tier includes 5 conversions per day with no account required. Create a free account for 5 daily uses, or upgrade to Starter ($15/mo) for 25 daily uses.

Question 4

What about scanned or image-based PDFs?

Accepted Answer

Scanned PDFs are automatically detected and processed with Tesseract OCR. For complex scanned documents, paid tiers unlock VisionOCR powered by GPT-4o for the highest accuracy.

Question 5

How does image text extraction work?

Accepted Answer

Your image is processed by GPT-4o, which understands visual context and layout. Unlike traditional OCR that reads characters, AI vision understands the meaning and structure of the content — tables, diagrams, handwriting, and even blurry text.

Question 6

What image formats are supported?

Accepted Answer

PNG, JPEG, WebP, GIF, and TIFF. Unsupported formats like JBIG2 and JPEG2000 (common in PDF internals) are automatically converted to PNG before processing.

Question 7

Can it read handwriting?

Accepted Answer

Yes. GPT-4o vision handles handwritten text, notes, and forms with good accuracy. Results depend on handwriting legibility.

Question 8

Is it free?

Accepted Answer

Yes, 5 free extractions per day. Starter plan ($15/mo) gives 25 daily uses with image upload support across the full platform.

Question 9

How accurate is the transcription?

Accepted Answer

Whisper achieves 99%+ accuracy on clear audio with minimal background noise. Accuracy depends on audio quality, accents, and technical terminology. Medical and legal recordings may need manual review of domain-specific terms.

Question 10

What audio formats are supported?

Accepted Answer

MP3, WAV, M4A, OGG, and FLAC. Maximum file size is 25 MB. For longer recordings, consider splitting the file.

Question 11

Does it support non-English languages?

Accepted Answer

Yes. Whisper supports 99 languages including Spanish, French, German, Japanese, Chinese, Arabic, and more. Accuracy is highest for English but works well for most major languages.

Question 12

Is it free?

Accepted Answer

Yes, 5 free transcriptions per day without an account. Starter plan ($15/mo) offers 25 daily uses with full audio transcription across the platform.

Free AI Tools

PDF to Text

Image to Text

Audio to Text

Need more than conversion?

Free AI Tools

PDF to Text

Image to Text

Audio to Text

Need more than conversion?