-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathSRT_for_Chinese.py
More file actions
189 lines (152 loc) · 8.26 KB
/
SRT_for_Chinese.py
File metadata and controls
189 lines (152 loc) · 8.26 KB
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
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
'''
SRT_for_Chinese 2.0
Author:kindkind.com
脚本编写于 python 3.7
说明：
谷歌云服务及SRT转换均参考了该项目：
https://github.com/darshan-majithiya/Generate-SRT-File-using-Google-Cloud-s-Speech-to-Text-API.git
使用过程中发现，对于上面项目中仅仅按照时间段分割英语单词的方式，非常不适合中国人以字为单位的语言习惯，
经过反复测试修改，始终有语句中断错乱、时码偏差较大的问题，
最后干脆将分割方式按照我的预想进行了重写，并添加了自动判断发音习惯参考值的功能（比较简陋，但是可以自适应不同人的语速了），
现在的版本是我正在使用，感觉比较良好的一个版本，现只分享出转换SRT字幕部分的代码，
完整的谷歌云语音识别服务的上传、接收脚本，请参考上面的项目。
已知问题：
1.谷歌声音识别服务转换出来的内容，会分为几个段落，但鉴于转换之后的一些缺陷（个人感觉是谷歌开发人员不怎么用这个服务吧）
SRT转换以后，对于每段最后一个字的时长无法与下一段落的起始字时间对接，导致每个段落最后一句显示的时长不够。
现将毫秒数x2以便不至于过短，我也懒得继续尝试，反正还需要用字幕软件校对一遍内容及部分时码。
2.语音识别服务产生的时码，跟音频时码有偏差，基本上整体会慢不到1秒（也有部分句子时码刚好），
使用SRT字幕编辑软件进行整体提前修正即可。
3.因为没有使用单词模型进行对照断句（感觉个人使用，弄个模型库不太现实），有些字说的快的时候会出现超长的句子，
而有些字又会超短，当然也跟语音识别服务的时码缺陷有关，暂时无法解决，只有字幕软件校对进行断句或合并句子来解决。
'''
#coding=utf8
import sys,os
import datetime,time
import srt
from pprint import pprint
##############################################################
def wordSpeed(result): # 计算发音习惯，统计单字速度，得出平均值阈值
wn = 0
wCount = 0
wBalance = 0
wMax = 0
for i in range(len(result.alternatives[0].words) - 1):
wStart = int((result.alternatives[0].words[i].start_time.seconds + result.alternatives[0].words[i].start_time.nanos*0.001*0.001*0.001)*100)
wEnd = int((result.alternatives[0].words[i].end_time.seconds + result.alternatives[0].words[i].end_time.nanos*0.001*0.001*0.001)*100)
wt = wEnd-wStart
if wt<80: #过滤超长的
wn += 1 #统计字数
wMax = wMax if wMax>wt else wt
wCount += wt
##print(wn,result.alternatives[0].words[i].word," : ",wStart,wEnd," = ",wEnd-wStart)
if wn>200 : #限制取样数量
break
##time.sleep(0.5)
if wn:
wBalance = int(wCount / wn)
print("平均：",wBalance, " 最大：",wMax)
print(" = "*20)
return wBalance , wMax
##############################################################
def subtitle_generation(speech_to_text_response): # 转SRT过程
response = speech_to_text_response
transcriptions = []
index = 0
sclip = 1
##tNum = 1
for result in response.results:
try:
start_sec = 0
start_microsec = 0
# for last word of result
last_word_end_sec = result.alternatives[0].words[-1].end_time.seconds
last_word_end_microsec = result.alternatives[0].words[-1].end_time.nanos * 0.001
# bin transcript
transcript = "" #result.alternatives[0].words[0].word
##index += 1 # subtitle index
wBalance, wMax = wordSpeed(result) #计算发音习惯
whold = int((wMax+wBalance)/2) #断句阈值
for i in range(len(result.alternatives[0].words)):
try:
if i==0 :
word=""
else:
word = result.alternatives[0].words[i-1].word
##if i==len(result.alternatives[0].words)-1 : word += result.alternatives[0].words[i].word
if 0==len(transcript) :
start_sec = result.alternatives[0].words[i].start_time.seconds
start_microsec = result.alternatives[0].words[i].start_time.nanos * 0.001
word_start_sec = result.alternatives[0].words[i].start_time.seconds
word_start_microsec = result.alternatives[0].words[i].start_time.nanos * 0.001 # 0.001 to convert nana -> micro 毫微秒转换微秒microsecond
word_end_sec = result.alternatives[0].words[i].end_time.seconds
word_end_microsec = result.alternatives[0].words[i].end_time.nanos * 0.001
##print(f"{start_sec}\t-\t{word_end_sec}\t{word}")
##pprint()
##print("")
transcript += word
##print(f"{i}/{len(result.alternatives[0].words)-1}\t{word_start_sec},{'%6.1f' % word_start_microsec}\t-\t{word_end_sec},{'%6.1f' % word_end_microsec}\t\t{word}")
#################### 按照标点断句 #########################
"""tt = result.alternatives[0].transcript
if tt[i+tNum:i+1+tNum]=="。" or tt[i+tNum:i+1+tNum]=="！" : ## or tt[i+tNum:i+1+tNum]=="，" or tt[i+tNum:i+1+tNum]=="、"
tNum += 1
if len(transcript)<4 :
transcript += " "
else :
sclip = 0
"""
##print(i,tNum,tt[i+tNum:i+1+tNum],result.alternatives[0].words[i + 1].word,transcript)
#################### 按照停顿断句 #########################
wThis = int((result.alternatives[0].words[i].start_time.seconds + result.alternatives[0].words[i].start_time.nanos*0.001*0.001*0.001)*100)
wEnd = int((result.alternatives[0].words[i].end_time.seconds + result.alternatives[0].words[i].end_time.nanos*0.001*0.001*0.001)*100)
wt = wEnd - wThis
##print(" "*20,int((wNext-wThis)*100))
if wt>wMax : sclip = 0 #间隔过长断句
if sclip>0 and wt>whold and len(transcript)>3 : sclip = 0 #阈值断句
if sclip>0 and wt>wBalance and len(transcript)>10 : sclip = 0 #长度过长断句
##print(" "*20,"Len = ",len(result.alternatives[0].words))
##time.sleep(0.5)
#################### 自定义规则结束 #######################
if sclip == 0 or i==len(result.alternatives[0].words)-1:
previous_word_end_sec = result.alternatives[0].words[i].end_time.seconds
previous_word_end_microsec = result.alternatives[0].words[i].end_time.nanos * 0.001
# append bin transcript
if i>=len(result.alternatives[0].words)-1 :
transcript += result.alternatives[0].words[i].word
else:
index += 1
transcriptions.append(srt.Subtitle(index, datetime.timedelta(0, start_sec, start_microsec), datetime.timedelta(0, previous_word_end_sec, previous_word_end_microsec), transcript))
##print(f"{index}\t{start_sec},{'%6.1f' % start_microsec}\t-\t{previous_word_end_sec},{'%6.1f' % previous_word_end_microsec}\t\t{transcript}")
transcript = ""
sclip = 1
##pprint(transcriptions)
##print(f"{start_sec}\t-\t{previous_word_end_sec}\t{transcript}\n")
###start_sec = word_start_sec
###start_microsec = word_start_microsec
##print(" "*20,transcript)
##time.sleep(3)
except IndexError as e:
pass
print("Err1:",e," ?? "*10)
print('文件', e.__traceback__.tb_frame.f_globals['__file__'], '行号', e.__traceback__.tb_lineno)
##time.sleep(0.05)
##tNum = 1
# append transcript of last transcript in bin
index += 1
transcriptions.append(srt.Subtitle(index, datetime.timedelta(0, start_sec, start_microsec), datetime.timedelta(0, last_word_end_sec, last_word_end_microsec*2), transcript))
##pprint(transcriptions)
##print("最后一句 ",f"{index}\t{start_sec},{'%6.1f' % start_microsec}\t-\t{last_word_end_sec},{'%6.1f' % last_word_end_microsec}\t\t{transcript}")
##time.sleep(5)
except IndexError as e:
pass
print("Err2:",e," ?? "*10)
print('文件', e.__traceback__.tb_frame.f_globals['__file__'], '行号', e.__traceback__.tb_lineno)
print(f"SRT Index-Count :{index-1}")
# turn transcription list into subtitles
subtitles = srt.compose(transcriptions)
return subtitles
if __name__ == '__main__':
##调用及保存SRT字幕
subtitles = subtitle_generation(response)
timer = 1 ##str(time.strftime("%Y%m%d_%H%M%S",time.localtime()))
with open(f"subtitles_{timer}.srt", "w") as f:
f.write(subtitles)