[TIKA-3384] Convert new transcribe package to a Parser

• Type: Task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Assignee: Unassigned
• Reporter: Tim Allison
• Created: 04/May/21 21:25
• Updated: 18/May/21 14:47

Description

This is a proposal to convert lewismc et al's awesome new transcribe code into a parser along the lines of Tesseract.

In 2.x, I inverted the call order from 1.x. The image parsers now look to see if there's a parser that supports a pseudo mime, like image/ocr-jpeg, if there is, then they apply that parser to the stream. We could do the same thing with media files that the new transcription package supports.

For those who want only ocr/transcription, they can turn off the image parsers and then decorate the OCR parser, for example, with supports "image/jpeg" and that parser will be called directly.

What do you think?

Issue Links

• is related to: TIKA-94 Speech-to-text transcription