python使用phonemizer实现获取英文文本句子单词音素

pip install phonemizer

# -*-coding:utf-8-*-
# date:2024-11
# Author: DataBall - XIAN
# Function: 获取英文文本句子单词音素
from phonemizer import phonemize
from phonemizer.backend import EspeakBackend
from phonemizer.punctuation import Punctuation
from phonemizer.separator import Separator
 
# en-gb – 英国英语
# en-us – 美国英语
 
language_ = "en-us" # 语言
 
print("语言类型 ：{}".format(language_))
backend = EspeakBackend(language_)
print("\n-------------------------------->>")
text = "hello Tom , my name is Lily  !"
# 整句音素
phonemes = phonemize(text, language=language_, backend='espeak', strip=True)
print("整句文本： {}: ".format(text))
print("整句音素：{}".format(phonemes))
 
print("\n-------------------------------->>")
# 按词分割音素
words = [word for word in text.lower().split(' ')]
# 音素之间用空格分割
phonemes_word = Separator(phone=' ', word=None)
dict_ = {word: backend.phonemize([word], separator=phonemes_word, strip=True)[0] for word in words}
 
print("打印音素字典:")
i = 0
for key_ in dict_.keys():
    i += 1
    if len(dict_[key_])>0: # 去除标点符号干扰
        print(" [{}] {} : {}".format(i,key_,dict_[key_]))