Show HN: Combining LLMs and Voice Models – Part 1 (opens in new tab)

This is a guide that I wrote to showcase a new batch inference feature for an OSS framework that I author (nitric.io).

I know things like Podcast generation via NotebookLM and also NotebookLlama exist, but wanted to demonstrate a case where an API could be built, and subsequently orchestrated in the cloud.

This is just the first part for producing audio using suno/bark via an API. I'm currently working on a part 2 that will introduce an LLM to make scripts from short prompt, which will be piped to the code introduced in Part 1.

Looking for feedback on improving this, there are a few things I'd like to clean up but overall am pretty happy with the outputs it produces so far.

Thanks in advance for any feedback given.