Spaces:

AbeShinzo0708
/

AI_Abe_Suga_Kishida_Bert_VITS2

Running

App Files Files Community

AI_Abe_Suga_Kishida_Bert_VITS2 / oldVersion /V101 /text /japanese.py

AbeShinzo0708

Upload 227 files

060d192 7 months ago

raw history blame contribute delete

No virus

2.98 kB

	# modified from https://github.com/CjangCjengh/vits/blob/main/text/japanese.py
	import re
	import sys

	import pyopenjtalk

	from . import symbols

	# Regular expression matching Japanese without punctuation marks:
	_japanese_characters = re.compile(
	r"[A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]"
	)

	# Regular expression matching non-Japanese characters or punctuation marks:
	_japanese_marks = re.compile(
	r"[^A-Za-z\d\u3005\u3040-\u30ff\u4e00-\u9fff\uff11-\uff19\uff21-\uff3a\uff41-\uff5a\uff66-\uff9d]"
	)

	# List of (symbol, Japanese) pairs for marks:
	_symbols_to_japanese = [(re.compile("%s" % x[0]), x[1]) for x in [("％", "パーセント")]]


	# List of (consonant, sokuon) pairs:
	_real_sokuon = [
	(re.compile("%s" % x[0]), x[1])
	for x in [
	(r"Q([↑↓]*[kg])", r"k#\1"),
	(r"Q([↑↓]*[tdjʧ])", r"t#\1"),
	(r"Q([↑↓]*[sʃ])", r"s\1"),
	(r"Q([↑↓]*[pb])", r"p#\1"),
	]
	]

	# List of (consonant, hatsuon) pairs:
	_real_hatsuon = [
	(re.compile("%s" % x[0]), x[1])
	for x in [
	(r"N([↑↓]*[pbm])", r"m\1"),
	(r"N([↑↓]*[ʧʥj])", r"n^\1"),
	(r"N([↑↓]*[tdn])", r"n\1"),
	(r"N([↑↓]*[kg])", r"ŋ\1"),
	]
	]


	def post_replace_ph(ph):
	rep_map = {
	"：": ",",
	"；": ",",
	"，": ",",
	"。": ".",
	"！": "!",
	"？": "?",
	"\n": ".",
	"·": ",",
	"、": ",",
	"...": "…",
	"v": "V",
	}
	if ph in rep_map.keys():
	ph = rep_map[ph]
	if ph in symbols:
	return ph
	if ph not in symbols:
	ph = "UNK"
	return ph


	def symbols_to_japanese(text):
	for regex, replacement in _symbols_to_japanese:
	text = re.sub(regex, replacement, text)
	return text


	def preprocess_jap(text):
	"""Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html"""
	text = symbols_to_japanese(text)
	sentences = re.split(_japanese_marks, text)
	marks = re.findall(_japanese_marks, text)
	text = []
	for i, sentence in enumerate(sentences):
	if re.match(_japanese_characters, sentence):
	p = pyopenjtalk.g2p(sentence)
	text += p.split(" ")

	if i < len(marks):
	text += [marks[i].replace(" ", "")]
	return text


	def text_normalize(text):
	# todo: jap text normalize
	return text


	def g2p(norm_text):
	phones = preprocess_jap(norm_text)
	phones = [post_replace_ph(i) for i in phones]
	# todo: implement tones and word2ph
	tones = [0 for i in phones]
	word2ph = [1 for i in phones]
	return phones, tones, word2ph


	if __name__ == "__main__":
	for line in open("../../../Downloads/transcript_utf8.txt").readlines():
	text = line.split(":")[1]
	phones, tones, word2ph = g2p(text)
	for p in phones:
	if p == "z":
	print(text, phones)
	sys.exit(0)