The document discusses progress on Bangla text-to-speech systems. It describes how text-to-speech works by converting text to speech through linguistic analysis, phonetic transcription, and acoustic processing. It then summarizes existing Bangla text-to-speech systems like KATHA and SUBACHAN, and discusses challenges like homograph disambiguation and improper concatenation of diphones. Solutions proposed include improving concatenation methods and developing duration models to generate more natural sounding speech.