[T2S] Add VISinger

[A-Quarter-Mile]

Hi, here are updates on T2S task:

• Add VISinger (from ESPnet) for singing voice synthesis task.

It now only works with default input.
Maybe we can let VISinger be an optional choice for a music score input with different format. More options for music score format will be added in the next version.

[ftshijt]

Many thanks!

• can you also add espnet to the requirements.txt?
• could you try to use VISinger to support svs systems with different languages (and different voices, e.g., male, female etc) For argument type, you may follow the one for Sound Extract