Added transcript for 1h to 1h25 by Anonymous.

Removed music/notes confused as words by auto-subber.
Corrected tiemstamps.
Added some punctuation.

Makefile

@@ -1,14 +1,20 @@
 
 ORIGINAL_SUBS=autogen-subs
-VIDEO=1ere-introcut.mp4
+VIDEO=1ere.mkv
+# VIDEO=1ere-introcut.mp4
+
+test: getvideo
+	ffmpeg -i $(VIDEO) -i mangen-subs.fr.vtt -vcodec copy -acodec copy -c:s mov_text test.mp4 -y
+
+getvideo:
+	yt-dlp https://www.youtube.com/watch?v=WRq2197FlMw -o $(VIDEO)
 
 cleansubs:
 	python scripts/parse-subs.py $(ORIGINAL_SUBS).fr.vtt
-# 	ffmpeg -i 1ere.mkv -i output.vtt -vcodec copy -acodec copy -c:s mov_text test.mp4 -y
 
 getsubs:
 	yt-dlp --write-auto-subs --sub-lang=fr --skip-download  https://www.youtube.com/watch?v=WRq2197FlMw -o autogen-subs
 
 # omit the music
-clip:
-	ffmpeg -ss 3:18 -i 1ere-combined.mp4 -vcodec copy -acodec copy $(VIDEO)
+# clip:
+# 	ffmpeg -ss 3:18 -i 1ere-combined.mp4 -vcodec copy -acodec copy 1ere-introcut.mp4

README.md

@@ -5,9 +5,23 @@ This repository tracks the transcription and translation for the video conferenc
 
 The original video and the auto-generated captions in French can be sourced from [https://www.youtube.com/watch?v=WRq2197FlMw](https://www.youtube.com/watch?v=WRq2197FlMw).
 
-## Setup
+## Contributing
 
-Run the following to download auto-generated captions from YouTube to a new file named `autogen-subs.fr.vtt`.
+The automatically generated French transcript needs to be checked over by a human. The French transcript then needs to be translated to English, also with human oversight.
+
+* Verify and edit the file `mangen-subs.fr.vtt` which contains the current draft of the human-edited auto-generated transcript.
+* Run `make test` to add subtitles to the video and ensure the format is correct.
+
+### Contributors
+
+* tuxmain, [https://txmn.tk/](https://txmn.tk/)
+* 1 anonymous contributor(s)
+
+### Getting the original auto-generated transcript
+
+*To see how much a southern accent and speech quirks can throw off an AI.*
+
+Run the following to download the auto-generated transcript from YouTube to a new file named `autogen-subs.fr.vtt`.
 
     yt-dlp --write-auto-subs --sub-lang=fr --skip-download  https://www.youtube.com/watch?v=WRq2197FlMw -o autogen-subs
 

scripts/update-subs.py

@@ -0,0 +1,50 @@
+
+import re
+import sys
+import argparse
+
+def offset_timestamp(timestamp, offset):
+    hours, minutes, seconds, microseconds = map(float, re.split('[:.]', timestamp))
+    total_seconds = hours * 3600 \
+        + minutes * 60 \
+        + seconds  \
+        + (microseconds * 0.001) \
+        + offset
+    new_hours = int(total_seconds // 3600)
+    new_minutes = int((total_seconds % 3600) // 60)
+    new_seconds = total_seconds % 60
+    return f"{new_hours:02}:{new_minutes:02}:{new_seconds:.3f}"
+
+def update_webvtt(file_path, offset):
+    with open(file_path, 'r', encoding='utf-8') as file:
+        lines = file.readlines()
+
+    updated_lines = []
+    for line in lines:
+        ts = line[:line.find("align")]
+        if '-->' in ts:
+            start, end = ts.split(' --> ')
+            updated_start = offset_timestamp(start.strip(), offset)
+            updated_end = offset_timestamp(end.strip(), offset)
+            updated_lines.append(f"{updated_start} --> {updated_end}\n")
+        else:
+            updated_lines.append(line)
+
+    with open('updated_' + file_path, 'w', encoding='utf-8') as file:
+        file.writelines(updated_lines)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument('input_file', type=str, help='The input file name')
+    parser.add_argument('time_offset', type=str, help='The output file name')
+    args = parser.parse_args()
+
+    vtt = args.input_file
+    t   = args.time_offset
+    
+    minutes, seconds = map(float, re.split('[:.]', t))
+    offset = minutes * 60 + seconds
+
+    print("bumping timestamps in file %s by %f seconds" % (vtt,  offset))
+    update_webvtt(vtt, -offset)  # Use -40 to offset back by 40 seconds
+