制作FastWordQuery可以使用的语音库
本帖最后由 instrumental 于 2021-3-18 04:18 编辑FastWordQuery
在为[]制作单词卡片的时候，发音一直是一个比较头痛的问题。
有道的发音是比较容易获取的，但是有道的发音不纯正，很多音读得不是非常正确。
TTS也不能与真人发音相比，而且机器合成音在背单词的时候会有不良影响。
相对来讲朗文词典的真人发音就非常的纯正。
所以FastWQ特别针对朗文mdx词典制作对应的字典查询服务，以便快速的为单词添加读音并下载音频到本地。
按照以下步骤，可以很方便的为整个单词本快速的添加音频。
使用的时候切记不要导入词典而是直接去设置py
特制的朗文 6 本地词典
1b5de81464fb46ca892eab4a698a207e#1b5de81464fb46ca892eab4a698a207e#1206#/LDOCE6双解 修改 可提取音频/entry.js
f24817ae7bc736d4365908da3f87f77f#efd11ff3580a0f7dc0735119d296cf98#1297932816#/LDOCE6双解 修改 可提取音频/L6mp3.mdd
ee13707d1b966a49bc3149b111cd6b4d#60a7f4d9a40eb2359b4ece19876f584d#26550272#/LDOCE6双解 修改 可提取音频/L6mp3.mdd.db
7f301751b90d10f33412de45a599f2da#60a24ed264dc49dd7a8b05a1ace848cf#124057083#/LDOCE6双解 修改 可提取音频/L6mp3.mdx
aefbf1c0cae35980122ace73f533409e#df04641926e6d88694ecee7dcada3d19#11501568#/LDOCE6双解 修改 可提取音频/L6mp3.mdx.db
e79f484789815550da723c944b8f27d8#e79f484789815550da723c944b8f27d8#12642#/LDOCE6双解 修改 可提取音频/LDOCE6.css
修改程序的YHCD,然后用正则表达式制卡就好了
https://i.loli.net/2021/03/12/GWjiQILVaP2yUE5.png
- (https://github.com/sth2018/FastWordQuery)
- [为单词添加真人发音（朗文 mdx 词典）](https://sth2018.github.io/FastWordQuery/docs/get_mdx_ldoce6_sounds.html)
- (https://www.yuque.com/purequant/anki/sudl9z)
#-*- coding:utf-8 -*-
import os
import re
import random
from ..base import *
VOICE_PATTERN = r'<a href="sound://([\w/]+\w*\.mp3)"><img src="img/spkr_%s.png"></a>'
VOICE_PATTERN_WQ = r'<span class="%s"><a href="sound://([\w/]+\w*\.mp3)">(.*?)</span %s>'
MAPPINGS = [
['br', ],
['us', ]
]
LANG_TO_REGEXPS = {lang: regexps for lang, regexps in MAPPINGS}
DICT_PATH ='D:\\111111111111111111111111111111.mdx.mdx'
@register()
class xxxxx(MdxService):
def __init__(self):
dict_path = DICT_PATH
# if DICT_PATH is a path, stop auto detect
if not dict_path:
from ...service import service_manager, service_pool
for clazz in service_manager.mdx_services:
service = service_pool.get(clazz.__unique__)
title = service.builder._title if service and service.support else u''
service_pool.put(service)
if title.startswith(u'LDOCE6'):
dict_path = service.dict_path
break
super(xxxxx, self).__init__(dict_path)
@property
def title(self):
return getattr(self, '__register_label__', self.unique)
@export('PHON')
def fld_phonetic(self):
html = self.get_html()
m = re.search(r'<span class="pron">(.*?)</span>', html)
if m:
return m.groups()
return ''
def _fld_voice(self, html, voice):
"""获取发音字段"""
for regexp in LANG_TO_REGEXPS:
match = regexp.search(html)
if match:
val = '/' + match.group(1)
name = get_hex_name('mdx-'+self.unique.lower(), val, 'mp3')
name = self.save_file(val, name)
if name:
return self.get_anki_label(name, 'audio')
return ''
@export('BRE_PRON')
def fld_voicebre(self):
return self._fld_voice(self.get_html(), 'br')
@export('AME_PRON')
def fld_voiceame(self):
return self._fld_voice(self.get_html(), 'us')
def _fld_image(self, img):
val = '/' + img
# file extension isn't always jpg
file_extension = os.path.splitext(img).strip().lower()
name = get_hex_name('mdx-'+self.unique.lower(), val, file_extension)
name = self.save_file(val, name)
if name:
return self.get_anki_label(name, 'img')
return ''
@export('IMAGE')
def fld_image(self):
html = self.get_html()
m = re.search(r'<span class="imgholder"><img src="(.*?)".*?></span>', html)
if m:
return self._fld_image(m.groups())
return ''
@export('EXAMPLE')
def fld_sentence(self):
return self._range_sentence()
def _fld_audio(self, audio):
name = get_hex_name('mdx-'+self.unique.lower(), audio, 'mp3')
name = self.save_file(audio, name)
if name:
return self.get_anki_label(name, 'audio')
return ''
@export()
def fld_sentence_audio(self):
return self._range_sentence_audio()
@export('DEF')
def fld_definate(self):
m = m = re.findall(r'<span class="def"\s*.*>\s*.*<\/span>', self.get_html())
if m:
soup = parse_html(m)
el_list = soup.findAll('span', {'class':'def'})
if el_list:
maps = [u''.join(str(content) for content in element.contents)
for element in el_list]
my_str = ''
for i_str in maps:
my_str = my_str + '<li>' + i_str + '</li>'
return self._css(my_str)
return ''
@export()
def fld_random_sentence(self):
return self._range_sentence()
@export()
def fld_first2_sentence(self):
return self._range_sentence()
@export()
def fld_random_sentence_audio(self):
return self._range_sentence_audio()
@export()
def fld_first2_sentence_audio(self):
return self._range_sentence_audio()
def _range_sentence(self, range_arr=None):
m = re.findall(r'<span class="example"\s*.*>\s*.*<\/span>', self.get_html())
if m:
soup = parse_html(m)
el_list = soup.findAll('span', {'class':'example'})
if el_list:
maps = [u''.join(str(content) for content in element.contents)
for element in el_list]
my_str = ''
range_arr = range_arr if range_arr else
for i, i_str in enumerate(maps):
if i in range_arr:
i_str = re.sub(r'<a[^>]+?href=\"sound\:.*\.mp3\".*</a>', '', i_str).strip()
my_str = my_str + '<li>' + i_str + '</li>'
return self._css(my_str)
return ''
def _range_sentence_audio(self, range_arr=None):
m = re.findall(r'<span class="example"\s*.*>\s*.*<\/span>', self.get_html())
if m:
soup = parse_html(m)
el_list = soup.findAll('span', {'class':'example'})
if el_list:
maps = []
for element in el_list:
i_str = ''
for content in element.contents:
i_str = i_str + str(content)
sound = re.search(r'<a[^>]+?href=\"sound\:\/(.*?\.mp3)\".*</a>', i_str)
if sound:
maps.append()
my_str = ''
range_arr = range_arr if range_arr else
for i, e in enumerate(maps):
if i in range_arr:
i_str = e
sound = e
mp3 = self._fld_audio(sound.groups())
i_str = re.sub(r'<a[^>]+?href=\"sound\:.*\.mp3\".*</a>', '', i_str).strip()
my_str = my_str + '<li>' + i_str + ' ' + mp3 + '</li>'
return self._css(my_str)
return ''
@export()
def fld_extra_examples(self):
lst = re.findall(r'href="/(@examples_.*?)\">.*?<', self.get_html())
if lst:
str_content = u''
for m in lst:
content = self.builder.mdx_lookup(m)
if len(content) > 0:
for c in content:
str_content += c.replace("\r\n","").replace("entry:/","")
return self._css(str_content)
return ''
@with_styles(cssfile='_ldoce6.css')
def _css(self, val):
return val
^([^\t]*)\t(.*)
\1\n<link type="text/css" rel="stylesheet" href="LDOCE6.css"/><div id="LDOCE6_Zzz_1"><span class="entry" id="zzz" new="NewInLdoce"><span class="entryhead"><span class="hwd"></span><span class="hyphenation"></span> <span class="brevoice"><a href="sound://\2"></span brevoice><img src="img/spkr_r.png"></a> <span class="amevoice"><a href="sound://\2"></span amevoice><img src="img/spkr_b.png"></a><span class="buttons"></span></span><span class="sense" id="zzz_s1"> <span class="def"></span def></span></span></div><script src="entry.js"></script>\n</>
有点复杂就说下原理,懂的人自然懂
FastWordQuery的作者特别做了朗文 6的支持,包含 .py 文件 特制的 mdx mdd
我就根据那个特制的MDX直接造了新的mdx出来,然后就支持了.
py代码比较复杂看不懂,我只是改了里面的词典名 xxxxxxx
然后正则表达式是用来制作mdx的
词头\tMP3
想想应该也没几个人用得上,学英语以外的外语的人应该不多
Thanks a lot. 给力！感谢分享！ FastWordQuery不是本身就支持朗文6的发音和命名查询么？我下载了朗文6的词典文件就可以直接用了。 这个真不错啊，还有图片呢？ 下来看看吧， 我就属于看不懂的那个
页:
[1]