Audio Transcription + Summarization Skill - Skill

Skill Content

The main content for this skill.

audio-transcription-summarization.txt

"Transcribe this meeting and create:
1. Attendee list (if mentioned)
2. Key discussion topics with timestamps
3. Decisions made
4. Action items with owners
5. Next steps"

Skill Content (2)

Additional code example for this skill.

audio-transcription-summarization.txt

"Transcribe this podcast episode and create:
1. Episode summary (2-3 sentences)
2. Main topics discussed with timestamps
3. Key quotes
4. Chapters (every 10 minutes)"

Skill Content (3)

Additional code example for this skill.

audio-transcription-summarization.txt

"Transcribe this interview with speaker labels.
Format as Q&A with:
- Interviewer questions highlighted
- Interviewee responses
- Notable quotes pulled out"

Features

Key capabilities and functionality

Local processing via Whisper
Format conversion with ffmpeg
Timestamped notes and action items
Optional speaker labels
Multilingual support (99 languages with auto-detection)
Word-level timestamp accuracy
Multiple output formats (TXT, VTT, SRT, JSON)

Use Cases

Common scenarios and applications

Summarize meetings and podcasts
Generate action items
Create searchable transcripts
Generate meeting minutes with action items
Create accessible transcripts for video content
Extract insights from podcast episodes

Requirements

Prerequisites and dependencies

ffmpeg
Python 3.11+ or whisper.cpp
openai-whisper (pip) or whisper.cpp binary

Installation

Setup instructions and requirements

Claude Code Setup

Step 1: Run command

claude code setup-1.sh

bash

pip install openai-whisper

Verify ffmpeg is in PATH

Claude Desktop Setup

Install ffmpeg

Install whisper via pip or build whisper.cpp

Troubleshooting

Common issues and solutions

ffmpeg not found
Install ffmpeg via your package manager and ensure it is on PATH. For macOS: `brew install ffmpeg`. For Linux: `apt-get install ffmpeg` or `yum install ffmpeg`. For Windows: Download from ffmpeg.org or use chocolatey: `choco install ffmpeg`. Verify installation with `ffmpeg -version`.
High Word Error Rate (WER) for noisy audio
Downmix to mono, apply noise reduction, and use a larger model. Pre-process audio with: `ffmpeg -i input.mp3 -af 'highpass=f=200,lowpass=f=3000' -ar 16000 -ac 1 output.wav`. Then use Whisper medium or large model instead of small.
Whisper model download fails with SSL certificate verification error
Set SSL_CERT_FILE environment variable to system certificate bundle. Alternative: use --model-dir to specify pre-downloaded model location. For Claude Code: `export SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt` (Linux) or use pre-downloaded models.
Code Interpreter not enabled, skill activation fails silently
Open Claude Desktop Settings, enable Code Interpreter under Features tab. Restart Claude Desktop application to activate skill execution environment. For Claude Code: Code Interpreter is always enabled.
Transcription timestamps drift out of sync with actual audio timing
Use --language flag to specify correct language explicitly. Re-encode audio at constant frame rate with: `ffmpeg -i input.mp3 -ar 16000 -ac 1 -c:a pcm_s16le output.wav`. This ensures consistent sample rate and prevents timestamp drift.

Usage Examples

Practical code examples demonstrating common use cases and implementation patterns

Transcribe with Whisper CLI

transcribe-with-whisper-cli.sh

bash

# Convert to mono 16kHz WAV
ffmpeg -i input.mp3 -ar 16000 -ac 1 input.wav

# Python whisper (pip install -U openai-whisper)
whisper input.wav --model small --language en --output_format txt

Summarize timestamps (Python)

summarize-timestamps-python.py

python

from pathlib import Path

text = Path('input.wav.txt').read_text(encoding='utf-8')
# Split by timestamps like [00:12:34]
# ... produce bullet summaries per 5-minute window ...
print(text[:500])

Multilingual transcription with speaker labels

Transcribe Spanish audio with medium model and output WebVTT format with timestamps

multilingual-transcription-with-speaker-labels.sh

bash

whisper audio.mp3 --model medium --language es --output_format vtt --task transcribe

Skill Content

The main content for this skill.

audio-transcription-summarization.txt

"Transcribe this meeting and create:
1. Attendee list (if mentioned)
2. Key discussion topics with timestamps
3. Decisions made
4. Action items with owners
5. Next steps"

Skill Content (2)

Additional code example for this skill.

audio-transcription-summarization.txt

"Transcribe this podcast episode and create:
1. Episode summary (2-3 sentences)
2. Main topics discussed with timestamps
3. Key quotes
4. Chapters (every 10 minutes)"

Skill Content (3)

Additional code example for this skill.

audio-transcription-summarization.txt

"Transcribe this interview with speaker labels.
Format as Q&A with:
- Interviewer questions highlighted
- Interviewee responses
- Notable quotes pulled out"

Features

Key capabilities and functionality

Local processing via Whisper
Format conversion with ffmpeg
Timestamped notes and action items
Optional speaker labels
Multilingual support (99 languages with auto-detection)
Word-level timestamp accuracy
Multiple output formats (TXT, VTT, SRT, JSON)

Use Cases

Common scenarios and applications

Summarize meetings and podcasts
Generate action items
Create searchable transcripts
Generate meeting minutes with action items
Create accessible transcripts for video content
Extract insights from podcast episodes

Requirements

Prerequisites and dependencies

ffmpeg
Python 3.11+ or whisper.cpp
openai-whisper (pip) or whisper.cpp binary

Installation

Setup instructions and requirements

Claude Code Setup

Step 1: Run command

claude code setup-1.sh

bash

pip install openai-whisper

Verify ffmpeg is in PATH

Claude Desktop Setup

Install ffmpeg

Install whisper via pip or build whisper.cpp

Troubleshooting

Common issues and solutions

ffmpeg not found
Install ffmpeg via your package manager and ensure it is on PATH. For macOS: `brew install ffmpeg`. For Linux: `apt-get install ffmpeg` or `yum install ffmpeg`. For Windows: Download from ffmpeg.org or use chocolatey: `choco install ffmpeg`. Verify installation with `ffmpeg -version`.
High Word Error Rate (WER) for noisy audio
Downmix to mono, apply noise reduction, and use a larger model. Pre-process audio with: `ffmpeg -i input.mp3 -af 'highpass=f=200,lowpass=f=3000' -ar 16000 -ac 1 output.wav`. Then use Whisper medium or large model instead of small.
Whisper model download fails with SSL certificate verification error
Set SSL_CERT_FILE environment variable to system certificate bundle. Alternative: use --model-dir to specify pre-downloaded model location. For Claude Code: `export SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt` (Linux) or use pre-downloaded models.
Code Interpreter not enabled, skill activation fails silently
Open Claude Desktop Settings, enable Code Interpreter under Features tab. Restart Claude Desktop application to activate skill execution environment. For Claude Code: Code Interpreter is always enabled.
Transcription timestamps drift out of sync with actual audio timing
Use --language flag to specify correct language explicitly. Re-encode audio at constant frame rate with: `ffmpeg -i input.mp3 -ar 16000 -ac 1 -c:a pcm_s16le output.wav`. This ensures consistent sample rate and prevents timestamp drift.

Usage Examples

Practical code examples demonstrating common use cases and implementation patterns

Transcribe with Whisper CLI

transcribe-with-whisper-cli.sh

bash

# Convert to mono 16kHz WAV
ffmpeg -i input.mp3 -ar 16000 -ac 1 input.wav

# Python whisper (pip install -U openai-whisper)
whisper input.wav --model small --language en --output_format txt

Summarize timestamps (Python)

summarize-timestamps-python.py

python

from pathlib import Path

text = Path('input.wav.txt').read_text(encoding='utf-8')
# Split by timestamps like [00:12:34]
# ... produce bullet summaries per 5-minute window ...
print(text[:500])

Multilingual transcription with speaker labels

Transcribe Spanish audio with medium model and output WebVTT format with timestamps

multilingual-transcription-with-speaker-labels.sh

bash

whisper audio.mp3 --model medium --language es --output_format vtt --task transcribe

Audio Transcription and Summarization

Claude Code Setup

Claude Desktop Setup

Transcribe with Whisper CLI

Summarize timestamps (Python)

Multilingual transcription with speaker labels

Reviews (0)

Audio Transcription and Summarization

Claude Code Setup

Claude Desktop Setup

Transcribe with Whisper CLI

Summarize timestamps (Python)

Multilingual transcription with speaker labels

Reviews (0)

Category

Source

Markdown Knowledge Base Composer

V0 Rapid UI Prototyping Workflow

Playwright E2E Testing Automation

Hire Claude Developers