jeffvincent avatar

video-transcript-analyzer

Analyze customer interview transcripts (SRT or plain text) to generate thematic breakdowns with summ

by jeffvincent|Open Source

Video Transcript Analyzer Skill

Overview

This Claude Skill analyzes customer interview and feedback call transcripts to generate comprehensive structured documents with thematic analysis, key insights, chapter markers, and full transcripts.

What It Does

Takes video transcripts (SRT or plain text format) and produces a single markdown document containing:

  • Video link (if provided)
  • Call summary (2-3 paragraphs)
  • Key verbatim quotes from customer/interviewee (5-8 quotes)
  • Topical breakdown with detailed context bullets
  • Chapter markers with timestamps (if SRT format)
  • Complete original transcript

When Claude Will Use This Skill

Claude automatically uses this skill when:

  • You provide an SRT or plain text transcript file
  • You say "analyze this customer call/interview"
  • You ask for "chapter markers" or "topical breakdown"
  • You want to extract insights from a recorded conversation

How to Use

Basic Usage

Upload: customer-call.srt
Message: "Can you analyze this transcript?"

With Video Link

Upload: transcript.srt
Message: "Here's the video: https://zoom.us/rec/abc123 - please analyze"

Without Video Link

If you don't provide a video URL initially, Claude will prompt you for one:

Upload: transcript.srt
Message: "Analyze this transcript"

Claude: "Do you have a link to the video recording? If so, please share it and I'll include it in the analysis."

You: "https://vimeo.com/123456" (or "No, not available")

Timestamp Adjustment

Upload: transcript.srt
Message: "I cut off the first 55 seconds, can you adjust the timestamps and analyze?"

Example Output Structure

Output filename:

  • With last name: YYYY-MM-DD_FirstName_LastName_CompanyName.md
    • Example: 2025-11-15_Martin_McKenna_HubSpot.md
  • Without last name: YYYY-MM-DD_FirstName_CompanyName.md
    • Example: 2025-11-11_Emma_Ax.md
# Customer Call Title

## Video
[Link or "Not provided"]

## Call Summary
[2-3 paragraphs covering who, what, why, and outcomes]

## Key Quotes
> "Verbatim quote from customer"
>
> — Speaker, context

[5-8 total quotes]

## Topical Breakdown

**00:02:57 - Topic Title** (if SRT)
Brief description of topic.
- Detailed bullet with specifics
- Another bullet with context
- Impact or consequence
[3-6 bullets per topic]

## Full Transcript
[Complete original transcript]

Features

Smart Format Detection

  • Automatically detects SRT (with timestamps) vs plain text
  • Parses timestamps and dialogue from SRT files
  • Handles various transcript formats

Thematic Organization

  • Groups related topics even if discussed at different times
  • Organizes chronologically for intro/context
  • Switches to thematic grouping for main issues
  • Ideal for product feedback and research synthesis

Verbatim Quotes

  • Preserves exact customer language (no paraphrasing)
  • Selects quotes showing emotion, specifics, and impact
  • Provides brief context for each quote
  • Focuses on customer voice, not interviewer questions

Comprehensive Output

  • Everything in one file: analysis + original transcript
  • Easy to share with team members
  • Reference original transcript while reading analysis
  • Includes video link for full context

Files in This Skill

  • Skill.md - Main skill definition with instructions for Claude
  • resources/EXAMPLES.md - Full annotated examples of inputs and outputs
  • resources/REFERENCE.md - Detailed guide for transcript analysis patterns
  • README.md - This file (usage documentation)

Tips for Best Results

Provide Context

If participant names aren't clear in transcript, tell Claude:

"This is a call between Hannah (HubSpot researcher) and Emma (customer admin)"

Request Timestamp Adjustments

If you edited the video:

"I removed the first 2 minutes where no one was talking, subtract 120 seconds from all timestamps"

Specify Focus Areas

If you want emphasis on certain topics:

"Focus on pain points and feature requests in your analysis"

Plain Text Format

If providing plain text without speaker labels:

"This is a transcript of an interview. Interviewer questions start with 'Q:' and responses with 'A:'"

What This Skill Doesn't Do

  • Won't create transcripts from audio/video files (needs pre-transcribed text)
  • Won't translate or clean up transcript errors
  • Won't add information not present in the transcript
  • Won't paraphrase quotes (always verbatim)
  • Won't remove sensitive information (unless explicitly requested)

Output Format

All outputs are markdown-formatted documents suitable for:

  • Sharing with product/research teams
  • Publishing to internal wikis or documentation
  • Converting to PDF or other formats
  • Using as input for further analysis

Privacy and Security

  • Customer names and company information are preserved as-is
  • Warn if transcript contains credentials or sensitive PII
  • Don't modify quotes or sanitize content
  • Full transcript always included (nothing redacted by default)

Examples

See resources/EXAMPLES.md for:

  • Complete SRT analysis with all sections
  • Plain text transcript analysis
  • Minimal input handling

See resources/REFERENCE.md for:

  • Detailed format specifications
  • Quote selection guidelines
  • Topic organization strategies
  • Quality check criteria
Download Skill