feat: option to pipe text-to-speech result to GPT for post processing

[peritus]

New feature:

• Take the text result from whisper
• feed it to GPT4-o with a user defined prompt
• use that to create the final markdown document

so that one does not only get the verbatim text but a full fledged note that is usable for copy-pasting, integration in other notes or to keep as is in another place in the vault.

Example for a post processing prompt:

Please respond in english. The following text is a transcript of a voice message. Format it in markdown, giving it a clear structure while keeping it as accurate as possible. Add a summary section at the beginning. If you come across to-do lists, format them as Markdown task lists with "[ ]". Avoid more than one level of nesting wherever possible. Finally, render the transcript exactly as spoken, with one sentence per line.

Settings:

[drkpxl]

I would love this merged in. I'm new to Obsidian plugins so is there a way to add this PR as a plugin directly so I don't need to wait and hope this get merged into mainline?

[heimoshuiyu]

Love this PR

But I think it should allow users to customize the API URL and the model. These are the two most basic parameters.

In addition to that, allowing users to set the temperature and other parameters (I believe the temperature is quite important)
This would make it better 👍

[ezuk]

@nikdanilov hey Nik, just figured I'd follow up on this PR since it's been open for a little while — this is my "one crucial feature" for the plugin so I wanted to ask. :) Would you be able to review? Thank you for the plugin!

[vlietz]

@nikdanilov hey Nik, just figured I'd follow up on this PR since it's been open for a little while — this is my "one crucial feature" for the plugin so I wanted to ask. :) Would you be able to review? Thank you for the plugin!

Just wanted to echo this — I’d also really love to see this PR approved. It’s a key feature for me too. Appreciate all the work on the plugin!

[ezuk]

@nikdanilov hey Nik, just figured I'd follow up on this PR since it's been open for a little while — this is my "one crucial feature" for the plugin so I wanted to ask. :) Would you be able to review? Thank you for the plugin!

Just wanted to echo this — I’d also really love to see this PR approved. It’s a key feature for me too. Appreciate all the work on the plugin!

fwiw, what I ended up doing is going with VoiceInk. It's a one-time purchase, $20 (or you can build it from source since it's open source), works systemwide, and supports "AI enhancement" (ChatGPT post processing) with several different providers and models.

[peritus]

@vlietz @ezuk @heimoshuiyu @drkpxl @nikdanilov

I've just open sourced https://github.com/peritus/obsidian-content-pipeline — which builds on top of the idea that made me submit this pull request, but in a more configurable way, enabling more complex workflows.

It's still a little rough around the edges (haven't submitted this to the official plugin repository yet, but plan to) .. happy to get feedback whether this solves your use cases as well.