PDF-to-audio with Hermes Agent works, but voice quality is the limit
A local can be used with an API so a PDF is sent to a r by email and returned as an audio file. This is useful for turning material into audiobook-style audio for driving or doing chores.
When there is plenty of time to wait, audio quality matters more than fast generation. A larger model or with longer could likely improve the listening result.
Raising steps does not automatically improve quality and can make the output fit the too tightly. CFG and Temp changes may make the voice slightly better or slightly worse, but they do not seem to solve the main quality problem.
Key points
- can connect through an API to turn emailed PDFs into audio files.
- The fits listening during driving, commuting, or chores.
- For this use case, better sound quality is more important than fast output.
- Increasing steps can hurt the result by making it fit the too tightly.
- CFG and Temp adjustments only changed the voice slightly, not dramatically.