I've reproduced this and fixed it for the next release.
macOS provides several voices with the same name for different languages and sometimes different quality levels. Objo was listing only the name, which made those variants look like duplicates. Selecting one also picked the first voice matching that name, so the individual variants couldn't be selected correctly.
The next release will distinguish them with names such as Eddy (en-GB) and Eddy (en-US), adding quality information where needed. Selecting an entry from Voices will use that exact variant for both speech and WAV export. Existing code using bare voice names will remain supported.
There's also a separate issue in the demo project: it populates the popup but never assigns the selection to the synthesiser. Before speaking or exporting, it needs:
speech.Voice = PopupMenu1.SelectedRowText
Without that assignment, it keeps using the system default voice regardless of the popup selection.
