kazunori279 / pdf2audiobook

pdf2audiobook
Apache License 2.0
303 stars 120 forks source link

Idea: Improve pronunciation using GPT / Bard #9

Open kaieberl opened 1 year ago

kaieberl commented 1 year ago

Dear Kanzunori san,

I was thinking that maybe you could use the gpt-3.5 or bard API to understand the content of the text and add appropriate ssml tags to it, e.g. like this:

Mathematical reasoning is not about mathematics <emphasis level="moderate">per se</emphasis>, it is about reasoning <emphasis level="moderate">in general</emphasis>.

This would make it easier to follow the content of the text. Maybe you could also make mathematical formulas and tables readable using this method.