Add logic to convert to srt #1
Conversation
| logger.info("Starting Transcription of Video File: %s" % video_file) | ||
| comment = transcribe_video_file(video_file) | ||
| output[video_file] = comment | ||
| convert_transcribe_to_srt(video_file) |
There was a problem hiding this comment.
It would be nice to do a best-effort post-processing of comments within a try-catch block, so that if the post-processing of a single transcript fails we can still move on and complete the rest.
| else: | ||
| end = format_time(items[len(items)-1]['end_time']) | ||
|
|
||
| with open(transcript_file_name_from_video_file_name(video_file).replace('.json', '.srt'), 'w', encoding='utf-8') as f: |
There was a problem hiding this comment.
Ideally we'd like to create one new srt file and one new txt file containing the topmost ranked transcript.
| logger.info("Conversion to srt started for video file: %s" % video_file) | ||
| with open(transcript_file_name_from_video_file_name(video_file), encoding='utf-8') as f: | ||
| raw = json.load(f) | ||
| items = raw['results']['items'] |
There was a problem hiding this comment.
if results are null or empty and items are null or empty, log the same and move on.
| start = format_time(current) | ||
| if token['type'] == 'punctuation': | ||
| next_line = next_line[0:-1] + token['alternatives'][0]['content'] | ||
| end = format_time(items[counter - 1]['end_time']) |
There was a problem hiding this comment.
If the punctuation is at the beginning of the transcript, won't this throw an index-out-of-bounds exception ?
| next_line = token['alternatives'][0]['content'] + ' ' | ||
| current = float(token['start_time']) | ||
| else: | ||
| next_line += token['alternatives'][0]['content'] + ' ' |
There was a problem hiding this comment.
It seems like any token with end_time - start_time > 5.0 will be written to the srt file. However, I'm not seeing how a sequence of tokens with smaller individual time spans will be strung together in a single sentence ?
ramnanib2
left a comment
ramnanib2
left a comment
There was a problem hiding this comment.
Great work ! Thanks for making the changes. Some minor comments. I think the logic within can be simplified a little within the items loop.
2 participants
Things left to do: