Besides the usual .po files mechanism, translators have to know that the sounds can be localized too. If the various translators have recording devices, they can register their voice and store it into .wav files in a new subdirectory of the sounds directory. Then they can translate the sound file name in the .po file.
For example, if destination language is italian, translators can register their voice in .wav files located in /opt/kde/share/apps/ktuberling/sounds/it. Then they can translate file names like "en/hat.wav" to names like "it/capello.wav".