The *picture* is in another scene, but the audio might be global. You probably want to disable it in Settings -> Audio, and re-add it to the specific scene(s).
Like pretty much *every* media production tool, OBS also treats video and audio separately. Just because one works like you want, doesn't mean anything about the other. You have to think of them separately, as if they come from completely separate sources with no connection between them, because to OBS, they do.
Another alternative is to use the
Advanced Scene Switcher plugin to automatically fade down a global audio source, and back up again, according to certain triggers. This makes it sound better than the hard cut that the scenes do, but you do have to script it separately. Possibly triggered by the scene switch, but technically a separate action.