Slide sharing service 'Docswell' now supports full-text transcription for image-based PDFs, including those from NotebookLM

Applucid Inc. (Headquarters: Chiyoda-ku, Tokyo; hereinafter "the Company"), operator of the slide sharing service "Docswell," has announced the launch of a full-text transcription feature for image-based PDFs.

With this feature, text within slides will be automatically transcribed even when uploading image-based slides generated by tools like Google NotebookLM, or PDFs where text has been outlined in software such as Adobe Illustrator. Previously, image-based PDFs resulted in blank transcription fields, making them difficult for search engines to index properly. This update ensures that simply uploading these files will trigger a full-text transcription, significantly improving their discoverability through search engines.

Background of Development

Docswell extracts text information from uploaded slides to facilitate discovery via search engines and internal site searches. However, in the following cases, PDF files do not contain embedded text information, leaving the transcription field blank and making it difficult for high-quality materials to be found via search engines:

• Slides generated by AI tools like NotebookLM (which are output as images and lack text data).
• PDFs created with Adobe Illustrator where fonts have been outlined (converting text into path data).
• Scanned paper documents converted to PDF (containing only image data without a text layer).

In response to the recent increase in image-based slides due to the widespread adoption of AI tools like NotebookLM, we developed this feature to correctly recognize and transcribe text from these types of PDFs.

Feature Overview

When an image-based PDF is uploaded to Docswell, the service automatically recognizes the text information within the slides and transcribes the full content. The transcribed text is reflected on the slide's detail page, allowing it to be indexed by search engines such as Google and Yahoo!.

Target	PDF files consisting only of images (PDFs without a text layer)
Process	Automatic full-text transcription performed during slide conversion

Back to Newsroom (3)