The ibus-speech-to-text package (https://src.fedoraproject.org/rpms/ibus-speech-to-text) claims to be an offline speech-to-text engine. However, at a glance I see some code in its repo that appears to be establishing online connections at runtime.

From https://github.com/Manish7093/IBus-Speech-To-Text/blob/main/engine/sttvoskmodelmanagers.py#L126 :
<snip>
    def _download_model_thread(self, download_link, destination, status):
        with urllib.request.urlopen(download_link) as response:
            length_str = response.getheader('content-length')
            blocksize = 4096
<snip>

IMHO, software that claims to work offline should not establish a network connection of any sort after the package is installed. I am also concerned about the possibility of private user data (this application has access to the user's microphone) being leaked to online services, possibly in future versions even if the current version does not leak such data.

Furthermore, I do not see any matches for the ibus-engine-stt or ibus-setup-stt binaries when searching the SELinux contexts.

$ grep -r 'ibus-\(engine\|setup\)-stt' /etc/selinux/targeted/contexts

I would like SELinux policies to be provided with the ibus-speech-to-text package and I would like those policies to deny any online connections at runtime.

I mentioned this concern in the change proposal that introduced the ibus-speech-to-text package:

https://discussion.fedoraproject.org/t/f42-change-proposal-ibus-speech-to-text-self-contained/143151/11

Reproducible: Didn't try