Media resources
Two resources that pull text out of a media file. Each has its own reference page. Both can run fully offline with no API key.
Applies to both workflow mode and agent mode.
| Resource | Extracts text from | Reference |
|---|---|---|
transcribe: | Audio (speech to text via Whisper) | Transcribe |
ocr: | Images (printed or handwritten text via tesseract) | OCR |
See also
- LLM resource - vision models can read images directly in a prompt
- Image analysis tutorial -
files:on a chat prompt instead of OCR - RAG resources - index the extracted text for search
- Resources overview - all resource types