feat: support deepgram audio queries and fix streaming in openai structured outputs

[Gmin2]

Mentioned in #1372

Fixed structured output streaming in openai

audio -> audio

audio -> text (query)

audio -> text(transcribe)

text -> speech(rest)

[vercel]

@Gmin2 is attempting to deploy a commit to the louis030195's projects Team on Vercel.

A member of the Team first needs to authorize it.

[Gmin2]

Hey @louis030195 i think we need to introduce testing in ai-proxy

[louis030195]

Hey @louis030195 i think we need to introduce testing in ai-proxy

can you add tests?

[Gmin2]

can you add tests?

should i add in this pr?

[louis030195]

yes

[louis030195]

also would be nice to support voice input, i want to add ability to query with voice input and output in rewind, search, etc

[louis030195]

/bounty 100

[algora-pbc]

💎 $100 bounty • Screenpi.pe

Steps to solve:

1. Start working: Comment /attempt #1465 with your implementation plan
2. Submit work: Create a pull request including /claim #1465 in the PR body to claim the bounty
3. Receive payment: 100% of the bounty is received 2-5 days post-reward. Make sure you are eligible for payouts

Thank you for contributing to mediar-ai/screenpipe!

Add a bounty • Share on socials

[Gmin2]

@louis030195 i think we should merge this, and do the test and websocket implementation in another PR

wdyt ?