Some time ago one of the program managers and I sat down with a customer exasperated that we do not support normalizing prompts within our Recording Editing and Design Studio. We were quite puzzled because we do actually support normalizing prompts.
In order to normalize prompts, right click on the prompt project and select properties. You can then select a wave file to normalize against or normalize the peaks of the wave files. When you build the prompt project into a .prompts file, all wave files will be normalized.
From this interaction with the customer we learned several important things.
1) From a usability point, this feature sucked. No one could find it. This was not the first time someone asked us how to normalize prompts. It is such an obscure feature that if we had a certification exam for Speech Server it would make the perfect trick question.
2) Another aspect of the usability of this feature is there is no visible way to determine that your prompts were normalized. This happens at compile time and is built into the .prompts file. The original files in the prompt project are not changed.
3) In general, no one should ever maximize to peaks. This does not give a good audio experience. Normalization should be RMS, which is done if you normalize to a wave file.
OK, so this may sound like I am slamming our own product, and in some ways this feature is certainly not optimal (I can't say much more, other than we know this). But there are some positives here.
1) Recording Editing and Design Studio (REDS) does have prompt normalization.
2) If you use this feature, do not maximize to peaks.
Of course, REDS was never designed to be a full editing studio and most likely never will be. For beginning users it has all of the features one would need, but most professionals will likely want to edit their wave files in a product such as SoundForge and then use REDS only to manage the wave files.