Delete code_to_extract_audio_from_any_mp4_line_by_line_from_srt_file.ipynb
Browse files
code_to_extract_audio_from_any_mp4_line_by_line_from_srt_file.ipynb
DELETED
@@ -1,330 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"cells": [
|
3 |
-
{
|
4 |
-
"cell_type": "code",
|
5 |
-
"execution_count": null,
|
6 |
-
"metadata": {},
|
7 |
-
"outputs": [],
|
8 |
-
"source": [
|
9 |
-
"import pandas as pd\n",
|
10 |
-
"import os, io, re, sys, time, datetime, wave, contextlib, librosa\n",
|
11 |
-
"from glob import glob\n",
|
12 |
-
"import numpy as np\n",
|
13 |
-
"from moviepy.editor import *\n",
|
14 |
-
"import soundfile as sf\n",
|
15 |
-
"from pydub import AudioSegment\n",
|
16 |
-
"\n",
|
17 |
-
"def create_directories():\n",
|
18 |
-
" slice_path = './ready_for_slice'\n",
|
19 |
-
" if not os.path.exists(slice_path):\n",
|
20 |
-
" try:\n",
|
21 |
-
" os.mkdir(slice_path)\n",
|
22 |
-
" except OSError:\n",
|
23 |
-
" print('Creation of directory %s failed' %slice_path)\n",
|
24 |
-
" sliced_audio = './sliced_audio'\n",
|
25 |
-
" if not os.path.exists(sliced_audio):\n",
|
26 |
-
" try:\n",
|
27 |
-
" os.mkdir(sliced_audio)\n",
|
28 |
-
" except OSError:\n",
|
29 |
-
" print('Creation of directory %s failed' %sliced_audio)\n",
|
30 |
-
"\n",
|
31 |
-
" merged_csv_files = './merged_csv'\n",
|
32 |
-
" if not os.path.exists(merged_csv_files):\n",
|
33 |
-
" try:\n",
|
34 |
-
" os.mkdir(merged_csv_files)\n",
|
35 |
-
" except OSError:\n",
|
36 |
-
" print('Creation of directory %s failed' %merged_csv_files)\n",
|
37 |
-
"\n",
|
38 |
-
" final_csv_files = './final_csv'\n",
|
39 |
-
" if not os.path.exists(final_csv_files):\n",
|
40 |
-
" try:\n",
|
41 |
-
" os.mkdir(final_csv_files)\n",
|
42 |
-
" except OSError:\n",
|
43 |
-
" print('Creation of directory %s failed' %final_csv_files)\n",
|
44 |
-
" \n",
|
45 |
-
" audio = './audio'\n",
|
46 |
-
" if not os.path.exists(audio):\n",
|
47 |
-
" try:\n",
|
48 |
-
" os.mkdir(audio)\n",
|
49 |
-
" except OSError:\n",
|
50 |
-
" print('Creation of directory %s failed' %audio)\n",
|
51 |
-
" \n",
|
52 |
-
" srt_files = './srt_files'\n",
|
53 |
-
" if not os.path.exists(srt_files):\n",
|
54 |
-
" try:\n",
|
55 |
-
" os.mkdir(srt_files)\n",
|
56 |
-
" except OSError:\n",
|
57 |
-
" print('Creation of directory %s failed' %srt_files)\n",
|
58 |
-
"\n",
|
59 |
-
"def merge_csv(path):\n",
|
60 |
-
" print('Merging csv-files with transcriptions')\n",
|
61 |
-
" csv_combined = pd.DataFrame()\n",
|
62 |
-
" for entry in glob (path+'*.csv'):\n",
|
63 |
-
" df = pd.read_csv(entry)\n",
|
64 |
-
" csv_combined = csv_combined.append(df)\n",
|
65 |
-
"\n",
|
66 |
-
" csv_combined.to_csv('./merged_csv/Full_Transcript.csv', header=True, index=False, encoding='utf-8')\n",
|
67 |
-
" print('All csv-files merged')\n",
|
68 |
-
"\n",
|
69 |
-
"def change_encoding(srt):\n",
|
70 |
-
" with io.open(srt, 'r', encoding='utf-8') as f:\n",
|
71 |
-
" text = f.read()\n",
|
72 |
-
" # process Unicode text\n",
|
73 |
-
" with io.open(srt, 'w', encoding='utf-8') as f:\n",
|
74 |
-
" f.write(text)\n",
|
75 |
-
"\n",
|
76 |
-
"def convert_srt_to_csv(file):\n",
|
77 |
-
" with open(file, 'r', encoding='utf-8') as h:\n",
|
78 |
-
" sub = h.readlines() #returns list of all lines\n",
|
79 |
-
"\n",
|
80 |
-
" re_pattern = r'[0-9]{2}:[0-9]{2}:[0-9]{2},[0-9]{3} --> [0-9]{2}:[0-9]{2}:[0-9]{2},[0-9]{3}'\n",
|
81 |
-
" regex = re.compile(re_pattern)\n",
|
82 |
-
" # Get start times\n",
|
83 |
-
" times = list(filter(regex.search, sub))\n",
|
84 |
-
" end_times = [time.split('--> ')[1] for time in times] #returns a list\n",
|
85 |
-
" start_times = [time.split(' ')[0] for time in times] #returns a list\n",
|
86 |
-
"\n",
|
87 |
-
" # Get lines\n",
|
88 |
-
" lines = [[]]\n",
|
89 |
-
" for sentence in sub:\n",
|
90 |
-
" if re.match(re_pattern, sentence):\n",
|
91 |
-
" lines[-1].pop()\n",
|
92 |
-
" lines.append([])\n",
|
93 |
-
" else:\n",
|
94 |
-
" lines[-1].append(sentence)\n",
|
95 |
-
"\n",
|
96 |
-
" lines = lines[1:] #all text in lists\n",
|
97 |
-
"\n",
|
98 |
-
" column_names = ['id','start_times', 'end_times', 'sentence']\n",
|
99 |
-
" df_text = pd.DataFrame(columns=column_names)\n",
|
100 |
-
"\n",
|
101 |
-
" df_text['start_times'] = start_times\n",
|
102 |
-
" df_text['end_times'] = end_times\n",
|
103 |
-
" df_text['sentence'] = [\" \".join(i).replace('\\n', '') for i in lines]\n",
|
104 |
-
" df_text['end_times'] = df_text['end_times'].replace(r'\\n', '', regex=True)\n",
|
105 |
-
"\n",
|
106 |
-
" df_text['id'] = np.arange(len(df_text))\n",
|
107 |
-
" id_extension = os.path.basename(file).replace('.srt', '_')\n",
|
108 |
-
" id_extension = id_extension.replace(' ', '_')\n",
|
109 |
-
" id_extension = id_extension.replace('-', '_')\n",
|
110 |
-
" id_extension = id_extension.replace('.', '_')\n",
|
111 |
-
" id_extension = id_extension.replace('__', '_')\n",
|
112 |
-
" id_extension = id_extension.replace('___', '_')\n",
|
113 |
-
" df_text['id'] = id_extension + df_text['id'].map(str)\n",
|
114 |
-
" file_extension = id_extension[:-1]\n",
|
115 |
-
"\n",
|
116 |
-
" def convert_to_ms(time):\n",
|
117 |
-
" h_ms = int(time[:2])*3600000\n",
|
118 |
-
" m_ms = int(time[3:5])*60000\n",
|
119 |
-
" s_ms = int(time[6:8])*1000\n",
|
120 |
-
" ms = int(time[9:12])\n",
|
121 |
-
" ms_total = h_ms + m_ms + s_ms + ms\n",
|
122 |
-
" return(ms_total)\n",
|
123 |
-
"\n",
|
124 |
-
" def conv_int(start):\n",
|
125 |
-
" new_start = int(start)\n",
|
126 |
-
" return(new_start)\n",
|
127 |
-
"\n",
|
128 |
-
" df_text['start_times'] = df_text['start_times'].apply(convert_to_ms)\n",
|
129 |
-
" df_text['end_times'] = df_text['end_times'].apply(convert_to_ms)\n",
|
130 |
-
" df_text['start_times'] = df_text['start_times'].apply(conv_int)\n",
|
131 |
-
" df_text.to_csv('./ready_for_slice/' + file_extension + '.csv', index=False, header=True, encoding='utf-8-sig')\n",
|
132 |
-
"\n",
|
133 |
-
"def wmv_to_wav(entry):\n",
|
134 |
-
" video = VideoFileClip(entry)\n",
|
135 |
-
" audio = video.audio\n",
|
136 |
-
" filename = os.path.basename(entry)\n",
|
137 |
-
" filename = filename.replace(' ', '_')\n",
|
138 |
-
" filename = filename.replace('-', '_')\n",
|
139 |
-
" filename = filename.replace('.', '_')\n",
|
140 |
-
" filename = filename.replace('__', '_')\n",
|
141 |
-
" filename = filename.replace('___', '_')\n",
|
142 |
-
" filename = filename[:-4] + '.wav'\n",
|
143 |
-
" audio.write_audiofile('./audio/' +filename)\n",
|
144 |
-
"\n",
|
145 |
-
"def mp4_to_wav(entry):\n",
|
146 |
-
" video = VideoFileClip(entry)\n",
|
147 |
-
" #extract audio from video\n",
|
148 |
-
" audio = video.audio\n",
|
149 |
-
" filename = os.path.basename(entry)\n",
|
150 |
-
" filename = filename.replace(' ', '_')\n",
|
151 |
-
" filename = filename.replace('-', '_')\n",
|
152 |
-
" filename = filename.replace('.', '_')\n",
|
153 |
-
" filename = filename.replace('__', '_')\n",
|
154 |
-
" filename = filename.replace('___', '_')\n",
|
155 |
-
" filename = filename[:-4] + '.wav'\n",
|
156 |
-
" #filename = filename[:-4]+'.wav'\n",
|
157 |
-
" #filename = filename[:10] + '_' + filename[-9:]\n",
|
158 |
-
" audio.write_audiofile('./audio/' +filename)\n",
|
159 |
-
"\n",
|
160 |
-
"def pre_process_audio(audio_path):\n",
|
161 |
-
" path_audio_processed = './ready_for_slice/'\n",
|
162 |
-
" if not os.path.exists(path_audio_processed):\n",
|
163 |
-
" try:\n",
|
164 |
-
" os.mkdir(path_audio_processed)\n",
|
165 |
-
" except OSError:\n",
|
166 |
-
" print('Creation of directory %s failed' %path_audio_processed)\n",
|
167 |
-
" else:\n",
|
168 |
-
" print('Successfully created the directory %s' %path_audio_processed)\n",
|
169 |
-
" start_sub = time.time()\n",
|
170 |
-
" n = 0\n",
|
171 |
-
" print('Downsampling wav files...')\n",
|
172 |
-
" for file in os.listdir(audio_path):\n",
|
173 |
-
" if(file.endswith('.wav')):\n",
|
174 |
-
" try:\n",
|
175 |
-
" nameSolo_1 = file.rsplit('.', 1)[0]\n",
|
176 |
-
" y, s = librosa.load(audio_path + file, sr=16000) # Downsample 44.1kHz to 8kHz\n",
|
177 |
-
" sf.write(path_audio_processed + nameSolo_1 + '.wav', y, s)\n",
|
178 |
-
" n = n+1\n",
|
179 |
-
" print('File ', n , ' completed:', nameSolo_1)\n",
|
180 |
-
" except EOFError as error:\n",
|
181 |
-
" next\n",
|
182 |
-
"\n",
|
183 |
-
" s = 0\n",
|
184 |
-
" print('Changing bit pro sample...')\n",
|
185 |
-
" for file in os.listdir(path_audio_processed):\n",
|
186 |
-
" if(file.endswith('.wav')):\n",
|
187 |
-
" try:\n",
|
188 |
-
" nameSolo_2 = file.rsplit('.', 1)[0]\n",
|
189 |
-
" #nameSolo_2 = nameSolo_2.replace('')\n",
|
190 |
-
" data, samplerate = sf.read(path_audio_processed + file)\n",
|
191 |
-
" sf.write(path_audio_processed + nameSolo_2 + '.wav', data, samplerate, subtype='PCM_16')\n",
|
192 |
-
" s = s + 1\n",
|
193 |
-
" print('File ' , s , ' completed')\n",
|
194 |
-
" except EOFError as error:\n",
|
195 |
-
" next\n",
|
196 |
-
"\n",
|
197 |
-
" end_sub = time.time()\n",
|
198 |
-
" print('The script took ', end_sub-start_sub, ' seconds to run')\n",
|
199 |
-
" \n",
|
200 |
-
"def create_DS_csv (path):\n",
|
201 |
-
" print('Extracting filepath and -size for every .wav file in ./sliced_audio')\n",
|
202 |
-
" data = pd.DataFrame(columns=['file_name', 'duration'])\n",
|
203 |
-
" df = pd.DataFrame(columns=['file_name', 'duration'])\n",
|
204 |
-
"\n",
|
205 |
-
" for entry in glob(path +'*.wav'):\n",
|
206 |
-
" filename = os.path.basename(entry)\n",
|
207 |
-
" with contextlib.closing(wave.open(entry, 'rb')) as f:\n",
|
208 |
-
" frames = f.getnframes()\n",
|
209 |
-
" rate = f.getframerate()\n",
|
210 |
-
" duration = frames / float(rate)\n",
|
211 |
-
" df['file_name'] = [filename]\n",
|
212 |
-
" df['duration'] = [duration]\n",
|
213 |
-
" data = data.append(df)\n",
|
214 |
-
" data.to_csv('./merged_csv/Filepath_Filesize.csv', header=True, index=False, encoding='utf-8')\n",
|
215 |
-
"\n",
|
216 |
-
"def split_files(item, wav_item):\n",
|
217 |
-
" song = AudioSegment.from_wav(wav_item)\n",
|
218 |
-
" df = pd.read_csv(item)\n",
|
219 |
-
"\n",
|
220 |
-
" def audio_split(df):\n",
|
221 |
-
" split = song[df['start_times']:df['end_times']]\n",
|
222 |
-
" split.export('./sliced_audio/' + df['id'] + '.wav', format ='wav')\n",
|
223 |
-
"\n",
|
224 |
-
" df.apply(audio_split, axis=1)\n",
|
225 |
-
"\n",
|
226 |
-
"def merge_transcripts_and_wav_files(transcript_path, DS_csv):\n",
|
227 |
-
" df_final = pd.DataFrame()\n",
|
228 |
-
" df_transcripts = pd.read_csv(transcript_path)\n",
|
229 |
-
" df_files = pd.read_csv(DS_csv)\n",
|
230 |
-
" def remove_path(path):\n",
|
231 |
-
" path = path.split('/')[-1]\n",
|
232 |
-
" return path\n",
|
233 |
-
" df_files['id'] = df_files['file_name'].apply(remove_path)\n",
|
234 |
-
" #filter out duration of less than 10 seconds\n",
|
235 |
-
" def convert(duration):\n",
|
236 |
-
" time = float(duration)\n",
|
237 |
-
" return time\n",
|
238 |
-
" df_files['duration'] = df_files['duration'].apply(convert)\n",
|
239 |
-
" df_files = df_files[df_files['duration']<10.00]\n",
|
240 |
-
" #drop unnecessary columns\n",
|
241 |
-
" df_transcripts.drop(['start_times','end_times'], axis=1, inplace=True)\n",
|
242 |
-
" df_files.drop(['duration'], axis=1, inplace=True)\n",
|
243 |
-
" df_files['id'] = df_files['id'].replace('.wav', '', regex=True)\n",
|
244 |
-
" #merge on column id\n",
|
245 |
-
" df_final = pd.merge(df_transcripts, df_files, on='id')\n",
|
246 |
-
" df_final.drop(['id'], axis=1, inplace=True)\n",
|
247 |
-
" #rearrange columns\n",
|
248 |
-
" df_final = df_final[['file_name', 'sentence']]\n",
|
249 |
-
" df_final.to_csv('./final_csv/metadata.csv', header=True, index=False, encoding='utf-8')\n",
|
250 |
-
" \n",
|
251 |
-
" create_directories()\n",
|
252 |
-
" \n",
|
253 |
-
"print(\"Put your video or audio files into the audio folder and srt files into the srt_files folder when you're ready...\")"
|
254 |
-
]
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"cell_type": "code",
|
258 |
-
"execution_count": null,
|
259 |
-
"metadata": {},
|
260 |
-
"outputs": [],
|
261 |
-
"source": [
|
262 |
-
"start_time = time.time()\n",
|
263 |
-
"srt_path = './srt_files'\n",
|
264 |
-
"audio_path = './audio/'\n",
|
265 |
-
"srt_counter = len(glob('./srt_files/' + '*.srt'))\n",
|
266 |
-
"\n",
|
267 |
-
"#Extracting information from srt-files to csv\n",
|
268 |
-
"print('Extracting information from srt_file(s) to csv_files')\n",
|
269 |
-
"for file in glob('./srt_files/*.srt'):\n",
|
270 |
-
" convert_srt_to_csv(file)\n",
|
271 |
-
"print('%s-file(s) converted and saved as csv-files to ./csv' %srt_counter)\n",
|
272 |
-
"print('---------------------------------------------------------------------')\n",
|
273 |
-
"\n",
|
274 |
-
"#extract audio (wav) from mp4\n",
|
275 |
-
"for entry in glob('./audio/*.mp4'):\n",
|
276 |
-
" mp4_to_wav(entry)\n",
|
277 |
-
"print('MP4 to WAV convert complete')\n",
|
278 |
-
"print('---------------------------------------------------------------------')\n",
|
279 |
-
"\n",
|
280 |
-
"\n",
|
281 |
-
"#Pre-process audio for folder in which wav files are stored\n",
|
282 |
-
"pre_process_audio(audio_path)\n",
|
283 |
-
"print('Pre-processing of audio files is complete.')\n",
|
284 |
-
"print('---------------------------------------------------------------------')\n",
|
285 |
-
"\n",
|
286 |
-
"print('Slicing audio according to start- and end_times of transcript_csvs...')\n",
|
287 |
-
"for item in glob('./ready_for_slice/*.csv'):\n",
|
288 |
-
" wav_item = item.replace('.csv','.wav')\n",
|
289 |
-
" if os.path.exists(wav_item):\n",
|
290 |
-
" split_files(item, wav_item)\n",
|
291 |
-
" else:\n",
|
292 |
-
" next\n",
|
293 |
-
"wav_counter = len(glob('./sliced_audio/' + '*.wav'))\n",
|
294 |
-
"print('Slicing complete. {} files in dir \"sliced_audio\"'.format(wav_counter))\n",
|
295 |
-
"print('---------------------------------------------------------------------')\n",
|
296 |
-
"\n",
|
297 |
-
"create_DS_csv('./sliced_audio/')\n",
|
298 |
-
"\n",
|
299 |
-
"#now join all seperate csv files\n",
|
300 |
-
"merge_csv('./ready_for_slice/')\n",
|
301 |
-
"print('Merged csv with all transcriptions created.')\n",
|
302 |
-
"print('---------------------------------------------------------------------')\n",
|
303 |
-
"transcript_path = './merged_csv/Full_Transcript.csv'\n",
|
304 |
-
"DS_path = './merged_csv/Filepath_Filesize.csv'\n",
|
305 |
-
"merge_transcripts_and_wav_files(transcript_path, DS_path)"
|
306 |
-
]
|
307 |
-
}
|
308 |
-
],
|
309 |
-
"metadata": {
|
310 |
-
"kernelspec": {
|
311 |
-
"display_name": "Python 3",
|
312 |
-
"language": "python",
|
313 |
-
"name": "python3"
|
314 |
-
},
|
315 |
-
"language_info": {
|
316 |
-
"codemirror_mode": {
|
317 |
-
"name": "ipython",
|
318 |
-
"version": 3
|
319 |
-
},
|
320 |
-
"file_extension": ".py",
|
321 |
-
"mimetype": "text/x-python",
|
322 |
-
"name": "python",
|
323 |
-
"nbconvert_exporter": "python",
|
324 |
-
"pygments_lexer": "ipython3",
|
325 |
-
"version": "3.10.0"
|
326 |
-
}
|
327 |
-
},
|
328 |
-
"nbformat": 4,
|
329 |
-
"nbformat_minor": 2
|
330 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|