When browsing the list of speakers and TVs available as playback destinations, VoiceOver does not recognise or speak the graphic which indicates the type of device.
Consequently, if you have both a speaker and an Apple TV in the same room, there is no feedback from VoiceOver as to which is which - both will have the same label, e.g. “Bedroom”.
You should find that for each available device VoiceOver will speak the room location of the device (and the current/last audio if available).
Visually, there is a graphic which indicates to sighted users whether the device is a speaker or TV. This graphic is not recognised or spoken by VoiceOver.
This can be an issue when you have both a speaker and TV in the same room, as VoiceOver will identify each by the room name, meaning that you may not be certain which is which when making a selection.
The list of available devices appears to always be presented in the same order, so you can work around this issue by remembering the order of any devices which are in the same room.