[Frontend] Set MAX_AUDIO_CLIP_FILESIZE_MB via env var instead of hardcoding

[deven-labovitch]

Purpose

Set MAX_AUDIO_CLIP_FILESIZE_MB via env var in SpeechToTextConfig instead of hardcoding to 25

Test Plan

Startup vllm locally with and without VLLM_MAX_AUDIO_CLIP_FILESIZE_MB set. Ensure VLLM_MAX_AUDIO_CLIP_FILESIZE_MB defaults to 25 when unset

Test Result

Passed

[gemini-code-assist]

Code Review

The pull request successfully makes MAX_AUDIO_CLIP_FILESIZE_MB configurable via an environment variable. However, the current implementation is not robust against invalid values for the environment variable, which could lead to a crash on startup. I've suggested a change to handle this case gracefully.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[DarkLight1337]

cc @NickLucche

[NickLucche]

Hey, thanks a lot for contributing!

So the intended usage of SpeechToTextConfig is to hold configs for speech-to-text models.
The way I am thinking about MAX_AUDIO_CLIP_FILESIZE_MB is more from a server managing point of view, purely to limit the amount of bytes streamed and handle traffic/networking rates.

In my view this shouldn't be model-dependent so I would either stick with a configurable env var or a startup argument.
In both cases it would be great if you could add a line to the openai server docs, I am pretty sure I left a TODO.

[deven-labovitch]

Hey, thanks a lot for contributing!

So the intended usage of SpeechToTextConfig is to hold configs for speech-to-text models. The way I am thinking about MAX_AUDIO_CLIP_FILESIZE_MB is more from a server managing point of view, purely to limit the amount of bytes streamed and handle traffic/networking rates.

In my view this shouldn't be model-dependent so I would either stick with a configurable env var or a startup argument. In both cases it would be great if you could add a line to the openai server docs, I am pretty sure I left a TODO.

Thanks for taking a look! Makes sense that this isn't per model config, just updated the PR to remove it from `SpeechToTextConfig.

Are these the openai server docs? I can add a line! docs/serving/openai_compatible_server.md

[deven-labovitch]

Hey, thanks a lot for contributing!

So the intended usage of SpeechToTextConfig is to hold configs for speech-to-text models. The way I am thinking about MAX_AUDIO_CLIP_FILESIZE_MB is more from a server managing point of view, purely to limit the amount of bytes streamed and handle traffic/networking rates.

In my view this shouldn't be model-dependent so I would either stick with a configurable env var or a startup argument. In both cases it would be great if you could add a line to the openai server docs, I am pretty sure I left a TODO.

@NickLucche Just added in a note to docs/serving/openai_compatible_server.md. Thanks!

[NickLucche]

let's just store the env var value once then we're good here

[NickLucche]

LGMT, thanks for your work!

[DarkLight1337]

Otherwise LGTM