Thanks, I think I understand now.
The way events and the event volume slider work is that each event key plays the audio file. If you set an event key wanting to change the volume, that will instead play a second instance of the audio file at that volume.
Volume is not interpolated between event keys. That's why the Spine editor only shows the keyed volume on the frame that keys the event, otherwise it shows the volume for the setup pose.
This is also true for the integer, float, and string fields. They are properties of the event key and are not interpolated between event keys.
Are you planning to use a runtime? Please note that the Spine Runtimes don't provide audio support at runtime. Instead, you receive the events at runtime and initiate audio playback yourself.
If you want to modulate volume at runtime, you could create a second event that does not have an audio path and set the volume there. At runtime when you see this event, you'd modify the volume for the related sound, if it is being played. You could use the integer, float, or string event fields to specify how many seconds it should take to change the volume and implement changing it over time. Those fields could also be used for other purposes, like panning left/right.