try to fix transcript stuff

[andmoredev]

No description provided.

[Copilot]

Pull request overview

This PR refactors the transcript processing logic in the get-clip handler to improve how speaker labels are handled in SRT transcript files. The change addresses issues with inline speaker prefixes by parsing them out and grouping consecutive dialogue by speaker into clean labeled blocks.

Changes:

• Import and use the detectSpeaker utility function for parsing speaker information from SRT entries
• Replace simple text concatenation with speaker-aware grouping logic that consolidates consecutive entries by the same speaker
• Move speaker label formatting to the fallback code path where it's actually needed

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

When an SRT entry contains only a speaker label with no dialogue (e.g., "Speaker:" with just whitespace after), the detectSpeaker function returns the entire text as dialogue with speaker set to null. This causes the speaker label itself to be included as dialogue text in the transcript output. Consider checking if the dialogue contains only the speaker pattern before adding it, or handle this edge case explicitly.

Suggested change

	if (dialogue.trim()) currentLines.push(dialogue);
	const trimmedDialogue = (dialogue || '').trim();
	// Skip empty dialogue lines.
	if (!trimmedDialogue) {
	continue;
	}
	// If detectSpeaker did not find a speaker, but the dialogue itself
	// looks like just a "Speaker:" label with no content, skip it so
	// that the label is not included as transcript text.
	const speakerLabelOnlyPattern = /^\s*[^:]+:\s*$/;
	if (!speaker && speakerLabelOnlyPattern.test(trimmedDialogue)) {
	continue;
	}
	currentLines.push(trimmedDialogue);