Fix to Timing Overlap Issue #608 and #621

[FredBill1]

Currently the function iterate_subtitles does not compute the start and end time correctly for each combined subtitle segment. All it does is selecting the start and end time of the first segement in the combined segement, without considering the following segments or segment breaks. The start and end time of each word is ignored, causing the --max_line_count and --max_line_width to return segements with overlapped timing, as stated in issue #608 and #621.

whisperX/whisperx/utils.py

Line 274 in f2da2f8

times.append((segment["start"], segment["end"], segment.get("speaker")))

whisperX/whisperx/utils.py

Lines 281 to 284 in f2da2f8

	for subtitle, _ in iterate_subtitles():
	sstart, ssend, speaker = _[0]
	subtitle_start = self.format_timestamp(sstart)
	subtitle_end = self.format_timestamp(ssend)

whisperX/whisperx/utils.py

Line 316 in f2da2f8

yield subtitle_start, subtitle_end, prefix + subtitle_text

---

I added a piece of code to compute the start and end time of the combined segment based on the time of each word in the segment. But I'm not sure how to utilize the times yielded by iterate_subtitles.

[issaMbarki]

I edited the iterate_subtitles function and it gave me really good result now with max_line_count and max_line_width, once I'll have sometime I'll create a pull request

[tormki-uio]

This pr worked well for me, as we've been struggling with issues around max_line_width and max_line_count created duplicated timestamps for split segments