The data is PCM format, so floats. It is not clear whether you are treating it as such but the name "ByteArray" means it stores binary data, but in a format depending on the application. See the documentation for extract() and loadPCMFromByteArray() for more details.
If you can't get it to work you could try representing what you are doing graphically. E.g. draw graphs of both input sounds, by just graphing the PCM data over time, and the calculated sum. This should make it much clearer what is going on. It's also how low level sound editing apps often work, as it is much easier to see what is happening when you have a visual representation of sound data.