Issue Description

Okay there are few issues that I am facing related to the AudioLM. And also there is some discrepancies in the documentation as well.

1. Transcription end-point is not available after running the local server instead below endpoints where available"

/v1/audio/processing
/v1/audio/processing_stream

But these two models require whisper purely.

2. nexa run omniaudio - I ran this to test out the omniaudio which is an AudioLM as mentioned in the README.md. After this I ran the server loading this model and tried the API endpoint /v1/audiolm/chat/completions
   Response:

{
  "detail": "The model that is loaded is not an AudioLM model. Please use an AudioLM model for audio chat completions."
}

So what I did was tried to check how the Audio is passed to omniaudio via Streamlit by running nexa run omniaudio -st but when I tried to do that I got permission error on MAC.

Error during audio processing: Error during inference: Error processing audio file: [Errno 1] Operation not permitted: '/Applications/Nexa.app/Contents/Frameworks/tmp'

Steps to Reproduce

Just running and trying out this will work simply. AudioLM related things are not working via API and Streamlit as well.

OS

MacOS

Python Version

Python-3.12

Nexa SDK Version

0.1.1.0

GPU (if using one)

No response