Comment by kennywinker

That doesn't sound like the same bar as "most authentic full-duplex speech model ever built"