I have a library of different words/phrases, and in order to build sentences at present I add a combination of these phrases into a playlist to make a sentence. Unfortunate
As MP3s are a compressed audio source, I imagine that you can't just concatenate them into a single file without decoding each one first to the wave form that it would play. This may be quite intensive. Perhaps you could cheat by using a critical section when playing back your phrase so that the CPU is not stolen from you until the phrase was complete. This isn't necessarily playing nice with other threads but might work if your phrases are short.