[install] compile kaldi static

[released model]updata the ds2 released model

[asr] nproc to ngpu

[asr] dataset to root dir

[asr] rename to result.md

[asr] test wav for u2

[TTS]update ipynb

[TTS]update ipynb, add eval loss

[asr] update ds2 offline result

[asr] fix preprocess of libri asr2

[asr] put vocab into data/lang_char

[TTS]fix readme

[ASR] update librispeech transformer result

[bugfix] Kaldi Feature using dither in train