File size: 1,742 Bytes
8d24b90 9df835b 8d24b90 9df835b 8d24b90 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
import os
import json
import warnings
from typing import List
import pyopenjtalk
def preprocess_input(src_str, seg_syb=" "):
src_str = src_str.replace("\n", seg_syb)
src_str = src_str.replace(" ", seg_syb)
return src_str
def pyopenjtalk_g2p(text) -> List[str]:
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
# phones is a str object separated by space
phones = pyopenjtalk.g2p(text, kana=False)
if len(w) > 0:
for warning in w:
if "No phoneme" in str(warning.message):
return False
phones = phones.split(" ")
return phones
def split_pinyin(pinyin: str, zh_plan: dict) -> tuple[str]:
# load pinyin dict from local/pinyin.dict
pinyin = pinyin.lower()
if pinyin in zh_plan["dict"]:
return zh_plan["dict"][pinyin]
elif pinyin in zh_plan["syllable_alias"]:
return zh_plan["dict"][zh_plan["syllable_alias"][pinyin]]
else:
return False
def get_tokenizer(lang):
if lang == "zh":
with open(os.path.join("resource/all_plans.json"), "r") as f:
all_plan_dict = json.load(f)
for plan in all_plan_dict["plans"]:
if plan["language"] == "zh":
zh_plan = plan
return lambda text: split_pinyin(text, zh_plan)
elif lang == "jp":
return pyopenjtalk_g2p
def load_pitch_dict(file_path = "resource/midi-note.scp"):
pitch_dict = {}
with open(file_path, "r", encoding="utf-8") as f:
for line in f:
items = line.strip().split()
pitch_dict[items[0]] = int(items[1])
pitch_dict[items[1]] = int(items[1])
return pitch_dict
|