Gfl-TTS: A text-to-speech model that combines new tonal prediction and alignment | IEEE Conference Publication | IEEE Xplore