Transcription Diarization Formatter

# Transcription Diarization Formatter

You are a transcription specialist. Format and clean the multi-speaker transcript below.

## Transcript Details

Recording type: [RECORDING_TYPE — meeting / interview / focus group / panel discussion / podcast]
Number of speakers: [SPEAKER_COUNT]
Speaker names (if known): [SPEAKER_NAMES — or 'Speaker 1, Speaker 2...']
Total duration: [DURATION]

## Raw Transcript

[PASTE_RAW_TRANSCRIPT_HERE]

## Formatting Instructions

1. **Speaker labels** — assign consistent speaker labels. If names are known, use them. Otherwise use Speaker A, Speaker B, etc.
2. **Timestamps** — add approximate timestamps every 1–2 minutes of content
3. **Paragraph breaks** — group each speaker's continuous turn into clean paragraphs
4. **Filler word removal** — remove um, uh, like, you know when they add no meaning
5. **Crosstalk notation** — mark simultaneous speaking as [CROSSTALK]
6. **Unclear passages** — mark as [INAUDIBLE] or [UNCLEAR: possible transcription]
7. **Tone markers** — add [laughs], [pause], [emphasis] only when they carry meaning

## Summary Section

After the cleaned transcript, add:
- Total word count by speaker (estimated percentage of talk time)
- The three most important statements made, with speaker and timestamp