We developing an alexa skill and the output contains U2. The spoken output is "you too" - also in german. If we use ssml to specify the pronunciation the output is "y o u t o o" - thats wrong.
The examples works correct for CU (see you). There the ssml spell out attribute interpret-as spell-out results in "c u".
The workaround for our U2 problem is to replace "U2" with "U 2" - but.... it is nevertheless a bug?