Homework: Implementing a Transformer-Based Text-to-Speech System

[SergeyMalashenko]

Good day!

I am planning to design a homework assignment for students focused on text-to-speech synthesis. The task is to convert short text sequences into speech, where each sequence follows the format LDDDLDDD, with L chosen from “ABCEHKMOPTXY” and D from “0123456789”. Students will be required to build the system entirely from scratch.

The model should be relatively simple while still producing good-quality audio. A transformer-based architecture is preferred.

Thank you!