Transcribe audio files (MP3, WAV, M4A, etc.) using OpenAI Whisper AI and ffmpeg to produce structured, timestamped transcripts with automatic summarization and action item extraction. Supports multilingual transcription, speaker diarization, and meeting minutes generation.
Installs and runs local executables: pip install openai-whisper (or whisper.cpp) plus ffmpeg, and downloads Whisper model files (39MB-1.5GB) on first use. Review the install and transcription commands and run them in a trusted environment.
Privacy notes
Audio files and the generated transcripts may contain personal or confidential speech. Transcription runs locally (no cloud upload) but transcripts, summaries, and downloaded models are written to local disk; control where these outputs are stored and shared.
Current risk score 0/100. Use staged verification before broader rollout.
Risk 0
Pre-adoption checks
Validate source and review signals before any execution.
Confirm source provenanceRequired
Source URL/provenance metadata is present.
Done
Confirm metadata review state
Listing has review metadata.
Done
Verify install payload
Install/config payload exists and can be inspected.
Done
Security checks
Confirm safety, privacy, and package integrity signals.
Review safety notesRequired
Safety notes are present.
Done
Review privacy notesRequired
Privacy notes are present.
Done
Verify package integrity metadata
Package verification/checksum metadata is available.
Done
Rollout
Adopt in controlled steps based on the selected plan.
Run in isolated sandbox firstRequired
Use a constrained sandbox and observe behavior across multiple tasks.
Pending
Roll out graduallyRequired
Roll out to a small cohort before wider usage.
Pending
Set monitoring and fallback
Define rollback path and monitor errors after adoption.
Pending
Evidence readiness
Evidence readiness matrix · balanced
Required evidence gates are covered (6/6 signals complete).
Risk 0
Source provenance
Present
Source repository/provenance is listed.
Required in this preset
Metadata review
Present
Review metadata is present.
Required in this preset
Safety notes
Present
Safety notes are present.
Required in this preset
Privacy notes
Present
Privacy notes are present.
Optional in this preset
Package integrity
Present
Package integrity metadata is present.
Optional in this preset
Install payload
Present
Install payload is available.
Required in this preset
Required evidence gates are covered for this preset.
Decision timeline
Decision timeline · balanced
6/6 steps complete with no blocking gaps for this preset.
Risk 0
triage
Confirm source provenanceRequired
Source/provenance metadata is available.
Done
triage
Check metadata review statusRequired
Review metadata is available.
Done
verify
Review safety notesRequired
Safety notes are available.
Done
verify
Review privacy notes
Privacy notes are available.
Done
verify
Validate package integrity metadata
Package integrity metadata is available.
Done
rollout
Verify install payload and commandsRequired
Install payload is available.
Done
No required blockers for this timeline preset.
Prerequisite readiness
Prerequisite readiness
6 prerequisites to line up before setup.
0/6 ready
Install & runtime3Permissions & scopes1General2
Safety & privacy surface
Safety & privacy surface
1 safety and 1 privacy notes across 1 risk area. Review closely: network access.
1 area
SafetyNetwork accessInstalls and runs local executables: pip install openai-whisper (or whisper.cpp) plus ffmpeg, and downloads Whisper model files (39MB-1.5GB) on first use. Review the install and transcription commands and run them in a trusted environment.
PrivacyNetwork accessAudio files and the generated transcripts may contain personal or confidential speech. Transcription runs locally (no cloud upload) but transcripts, summaries, and downloaded models are written to local disk; control where these outputs are stored and shared.
Safety notes
Installs and runs local executables: pip install openai-whisper (or whisper.cpp) plus ffmpeg, and downloads Whisper model files (39MB-1.5GB) on first use. Review the install and transcription commands and run them in a trusted environment.
Privacy notes
Audio files and the generated transcripts may contain personal or confidential speech. Transcription runs locally (no cloud upload) but transcripts, summaries, and downloaded models are written to local disk; control where these outputs are stored and shared.
Prerequisites
ffmpeg
Python 3.11+ or whisper.cpp
openai-whisper (pip) or whisper.cpp binary
Sufficient disk space for model downloads (Whisper models range from 39MB small to 1.5GB large model)
Audio file access permissions - read access to input audio files and write access for transcription output files
System resources: Minimum 4GB RAM for small model, 8GB+ recommended for medium/large models, GPU optional but recommended for faster processing
.gemini/skills/<skill-name>/SKILL.md or .agents/skills/<skill-name>/SKILL.md
cursor
Adapter
.cursor/rules/<skill-name>.mdc
cli
Manual
AGENTS.md or tool-specific context file
Full copyable content
# 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
About this resource
What This Skill Enables
Claude can transcribe audio files (MP3, WAV, M4A, etc.) and generate structured summaries with timestamps, action items, and speaker identification. This skill leverages Whisper AI and ffmpeg through Claude's Code Interpreter to process audio locally.
Compatibility
Native
Claude Code / Claude: native skill usage via SKILL.md.
Codex/OpenAI workflows: compatible with Agent Skills-style SKILL.md content as reusable workflow instructions.
Manual Adaptation
Gemini CLI: native skill usage via .gemini/skills/<skill-name>/SKILL.md or .agents/skills/<skill-name>/SKILL.md where supported.
Cursor: use the generated .cursor/rules/*.mdc adapter for project rules.
OpenClaw and similar agents: use the same skill content as a reusable prompt/workflow file when native skill import is unavailable.
Prerequisites
Required:
Claude Pro subscription
Code Interpreter feature enabled in Claude Desktop settings
Audio file uploaded to conversation (drag and drop)
What Claude handles automatically:
Installing and running Whisper AI models
Audio format conversion with ffmpeg
Timestamp extraction and alignment
Summary generation and structuring
How to Use This Skill
Basic Transcription
Prompt: "Transcribe this audio file and give me a clean text transcript."
Claude will:
Detect the audio format
Convert to optimal format for transcription
Run Whisper AI transcription
Return formatted text
Timestamped Summary
Prompt: "Transcribe this meeting recording and create a timestamped summary with key discussion points every 5 minutes."
Claude will:
Transcribe the full audio
Chunk by time intervals
Summarize each segment
Present with timestamps
Action Items Extraction
Prompt: "Transcribe this audio and extract all action items, decisions, and to-dos mentioned."
Claude will:
Transcribe the audio
Analyze for actionable items
List action items with timestamps
Identify who was assigned what (if mentioned)
Speaker Diarization
Prompt: "Transcribe this conversation and identify different speakers. Label them as Speaker 1, Speaker 2, etc."
Claude will:
Detect speaker changes in the audio
Segment by speaker
Label each segment
Present as a conversation transcript
Transcription engines compared
This skill defaults to OpenAI Whisper, but you can pair it with hosted APIs when you need streaming or built-in audio intelligence:
Engine
Type
Runs locally
Notable for
Whisper (OpenAI)
Open-source model
Yes
Free, self-hosted, strong multilingual accuracy
Deepgram
Hosted API
No
Low-latency streaming and speaker diarization
AssemblyAI
Hosted API
No
Built-in summaries, topic detection, and audio intelligence
Use Whisper for local, no-cost, privacy-preserving transcription; reach for Deepgram for real-time streaming or AssemblyAI when you want higher-level audio analysis out of the box.
File Size: For files over 25MB, mention if you want a specific time range transcribed first
Language: Specify the language if it's not English (e.g., "Transcribe this Spanish audio...")
Model Selection: For better accuracy on difficult audio, ask Claude to use the "medium" or "large" Whisper model
Post-Processing: Ask Claude to clean up transcription artifacts like repeated words or filler sounds
Common Workflows
Meeting Minutes Generation
"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"
Podcast Summary
"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)"
Interview Transcription
"Transcribe this interview with speaker labels.
Format as Q&A with:
- Interviewer questions highlighted
- Interviewee responses
- Notable quotes pulled out"
Troubleshooting
Issue: Transcription is inaccurate
Solution: Ask Claude to use a larger Whisper model or pre-process the audio for noise reduction
Issue: Wrong language detected
Solution: Explicitly specify the language in your prompt ("Transcribe this French audio...")
Issue: Timestamps are off
Solution: Ask Claude to re-align timestamps or specify the desired timestamp interval
Issue: Speaker diarization missing
Solution: Request it explicitly: "Please identify different speakers and label them"
Multilingual support (99 languages with auto-detection)
Word-level timestamp accuracy
Multiple output formats (TXT, VTT, SRT, JSON)
Real-time streaming transcription support with live audio processing, continuous transcription updates, and low-latency transcription for live events or meetings
Show that Audio Transcription + Summarization Skill is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/skills/audio-transcription-summarization)
How it compares
Audio Transcription + Summarization Skill side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
3 trust signals differ across this comparison (Package trust, Source provenance, Submitter).
Next steps differ across entries — use the actions in the table below to copy install commands and source links per resource.
Transcribe audio files (MP3, WAV, M4A, etc.) using OpenAI Whisper AI and ffmpeg to produce structured, timestamped transcripts with automatic summarization and action item extraction. Supports multilingual transcription, speaker diarization, and meeting minutes generation.
A Claude skill that fetches web pages, strips boilerplate with Mozilla Readability, respects robots.txt, removes duplicate content, and turns the result into structured summaries.
Expert subagent foreground background delegation capability pack for choosing when to run Claude Code subagents interactively versus in the background, coordinating parallel work, and returning summarized results safely.
Microsoft .NET team skill marketplace for AI coding agents working on .NET, C#, ASP.NET Core, Blazor, MAUI, diagnostics, MSBuild, NuGet, upgrades, tests, AI workflows, RAG pipelines, and C# MCP servers.
✓Installs and runs local executables: pip install openai-whisper (or whisper.cpp) plus ffmpeg, and downloads Whisper model files (39MB-1.5GB) on first use. Review the install and transcription commands and run them in a trusted environment.
✓Executes network requests to crawl third-party websites and installs scraping libraries (Playwright, BeautifulSoup) on first use; respect robots.txt, rate limits, and each site's terms of service, and review target URLs before running.
Crawled pages are written to local files (Markdown, JSON, CSV); review output paths so the skill does not overwrite existing files.
✓Background subagents reduce interactive oversight; do not use them for destructive edits, production deploys, or credential-handling tasks.
Parallel background subagents multiply token cost and can race on the same files or external systems.
Foreground subagents keep the parent waiting; overusing them erases the latency benefits of delegation.
Subagents isolate context but still consume tokens in their own windows; unbounded parallel research can become expensive quickly.
This skill recommends delegation modes; it must not spawn background subagents for high-risk tasks without explicit approval.
✓.NET build, test, upgrade, package, template, publish, and migration tasks can modify project files, lock files, generated code, packages, app settings, and deployment artifacts.
Diagnostics skills may suggest collecting traces, dumps, counters, crash data, MSBuild binlogs, or performance profiles; collect those artifacts only with explicit approval and storage controls.
MCP server skills can expose local code, files, APIs, credentials, or production services as callable tools; review tool descriptions, parameter validation, authorization, and transport choice before connecting clients.
NuGet and publish workflows can push packages or artifacts to public or private feeds; verify package IDs, versions, API keys, feed targets, and release policy before publishing.
Upgrade and modernization guidance should be verified against each application's framework support window, deployment target, package compatibility, and rollback plan.
Privacy notes
✓Audio files and the generated transcripts may contain personal or confidential speech. Transcription runs locally (no cloud upload) but transcripts, summaries, and downloaded models are written to local disk; control where these outputs are stored and shared.
✓Fetches and stores content from external websites locally and may use site cookies or auth tokens to reach gated pages; do not crawl private or authenticated data without authorization, and keep any credentials out of saved output.
✓Background subagent transcripts may accumulate sensitive file contents, customer examples, and internal URLs before summarization.
Summaries returned to the parent session can still leak data if subagents paste raw tool output instead of redacted findings.
Public delegation notes should describe mode choice and oversight level, not full subagent transcripts.
✓.NET repositories may contain connection strings, appsettings secrets, user secrets, certificates, environment variables, telemetry keys, logs, traces, dumps, package credentials, and production data.
MSBuild binlogs, crash dumps, profiler output, and test artifacts can contain source paths, dependency graphs, request data, exception payloads, configuration values, and environment details.
MCP servers created with these skills may forward prompts and tool inputs to local processes, HTTP services, databases, cloud APIs, or third-party model providers depending on the implementation.
Keep private NuGet credentials, signing keys, deployment secrets, customer data, dumps, and proprietary source out of public prompts, issues, pull requests, and shared artifacts.
Prerequisites
ffmpeg
Python 3.11+ or whisper.cpp
openai-whisper (pip) or whisper.cpp binary
Sufficient disk space for model downloads (Whisper models range from 39MB small to 1.5GB large model)
Node.js 18+ or Python 3.11+
Playwright or Puppeteer (for JavaScript rendering)
readability or newspaper3k (for content extraction)
BeautifulSoup or Cheerio (for HTML parsing)
A concrete parent-session task that may benefit from subagent delegation.
Understanding of expected outputs, deadlines, and whether interactive approval is required mid-task.
Inventory of tools, MCP servers, and file scopes the subagent will need.
Agreement on how subagent results will be summarized back into the parent session.
An AI coding assistant or skill host that supports Agent Skills, plugin marketplaces, or direct skill installation.
.NET SDK and project-local build/test tooling appropriate for the repository being edited.
For MCP server work, the official C# MCP SDK, MCP project templates, and a target transport choice such as stdio or HTTP.
For diagnostics, permission to collect traces, dumps, logs, counters, binlogs, or test output from the target environment.