Cues with overlapping timings are doubled in the Transcript output

I noticed that a lot of lines in my episode transcript are shown double in the Web Player’s Transcript view. This makes it very ugly to read, and kind of defeats the purpose of the transcript. Example episode where this happens here, see also this screenshot:

When debugging the error, I found out that I’m using a WebVTT file for the episode transcript that has overlapping timings of cues, both because of using Speaker Identification and simply because my WebVTT timings are generated by Amazon Transcribe and there the cues do have some overlap.

Since overlapping cues are explicitly allowed by the WebVTT standard, I would expect Podlove Web Player to handle this more gracefully. It’s obvious, for instance, that no cue should ever be shown multiple times because there is no situation where this would be beneficial for the user.

Is this something you’d plan to fix?

The Web Player itself doesn’t implement the WebVTT standard. That is part of the Publisher implementation. So to address this correctly first we need to get an example WebVTT. In the current implementation overlapping cues are support when the speaker varies. Keep in mind that the WebVTT standard is used for Video subtitles with long frames where multiple speakers are a viable option. A podcast is focused on audio content but humans are very poor at parsing two voices at the same time I wouldn’t project any expectations from the standard to audio content.