The author Keith Casey compares and contrasts two text-to-speech (TTS) voices, one being the traditional female voice and the other being a new Alice TTS voice. The Alice voice is found to be better in terms of pronunciation, especially when it comes to handling acronyms, heteronyms, and uncommon words. To improve the accuracy of the traditional female voice, tips are provided such as inserting spaces between letters or using regular expressions to handle specific cases like acronyms and homonyms. The article also discusses the use of homonyms in place of heteronyms to solve some problem cases. Finally, it highlights the importance of considering pronunciation over spelling when working with TTS voices.