* fix tacotron2 dygraph to static , test=tts

* fix tacotron2 dygraph to static , test=tts

* simplify synthesize_e2e.py , test=tts

* add tacotron2 readme, test=doc

* update changelog.md, test=doc

move link_wav.py to paddlespeech/t2s/exps/gan_vocoder/ so that different vocoders can use one script.

add --dataset and --rootdir to voc3 finetune script, also remove --speaker-dict in voc5.

* update mb melgan

* update mb melgan, test=tts

* add style melgan pretraied model

* add style melgan pretraied model, test=tts
Co-authored-by: NHui Zhang <zhtclz@foxmail.com>

* remve useless conf

* format code

* update conf

* update conf

* update conf

* add hifigan

* add hifigan

* integrate synthesize synthesize_e2e, inference for tts, test=tts

* add some python files, test=tts

* update readme, test=doc_fix