- Home
- Skills
- Productivity
- Audio Review by shash.builds
Audio Review by shash.builds
Measures a podcast, voice-over, music track or video soundtrack and gives a scored review like a senior audio designer, with ranked issues and fixes an engineer can apply.
Free
Works with the AI Agents you already use
Audio Review by shash.builds
Example session with this skill installed
Is voiceover.wav ready for YouTube?
- Read your context and instructions
- Compiled the audio review by
Verdict: one or two sentences, a score out of 10 and a label (Ready to ship, Minor fixes, Needs work or Not usable).
What I'm hearing: a short paragraph, as a designer would say it.
Scorecard (1-10, one reason each): Loudness and delivery, Dynamics, Tonal balance, Clarity, Stereo and mono, Noise and artifacts, Technical integrity.
Issues by severity, with time range or frequency as evidence.
Fix list with starting values, e.g. "high-pass at 80 Hz".
What's working. Limits of this review.
Connects securely to your tools. The creator never sees your data.
About this skill
Measures a podcast, voice-over, music track or video soundtrack and gives a scored review with a ranked issue list and fixes an engineer can apply.
Claude can't take audio as input, so this skill gives it what an audio designer works from while listening: meters, analysers and a spectrogram. Claude then judges the file against a senior-designer rubric.
What gets measured
| Area | Measurements |
|---|---|
| Loudness | Integrated LUFS, loudness range, true peak (EBU R128 via ffmpeg), a short-term loudness timeline, and pass or fail against Spotify, Apple Music, YouTube, podcast, EBU R128, ATSC A/85, game and social targets |
| Dynamics | Crest factor, macro-dynamics, section-by-section level |
| Tonal balance | Energy in sub, bass, low-mid, mid, high-mid, presence, brilliance and air bands; spectral tilt; centroid |
| Problems | Clipping, DC offset, clicks and pops, 50/60 Hz hum, narrow resonances, sibilance, noise floor and SNR, lossy-codec bandwidth cutoff, leading and trailing silence |
| Stereo | L/R correlation, width, mono-fold loss, low-end width, L/R balance |
Before and after
| Claude without the skill | Claude with the skill |
|---|---|
| Cannot hear the file, so it guesses or declines | Reads real measurements and spectrogram and waveform pictures |
| Says "sounds a bit quiet" | Reports integrated loudness, loudness range and true peak, with pass or fail against platform targets |
| Gives general mixing tips | Ranks issues by severity with time ranges and gives fix values to start from |
| Misses hum, clipping and codec cutoffs | Checks clipping, DC offset, clicks, hum, sibilance, noise floor and bandwidth |
Examples
- "Is this podcast episode ready to upload?" A readiness label, loudness against podcast targets and the fixes that matter first.
- "Compare my mix with this reference track." Both files analysed the same way, with the gaps in tonal balance and loudness.
- "Check the sound of my reel." A review of the video's audio, including how it folds to mono.
- "Why does the voice sound harsh?" The sibilance and presence-band evidence, and an EQ move with a frequency and amount to try.
Three ways to run it
- As a skill: share a file and ask "how does this sound?".
review.py: the same review through the Claude API, for scripts and batch jobs.initiation_prompt.md: a paste-in prompt for any Claude chat.
Free and MIT licensed. Part of the free shash.builds skill collection by Shashwat Jain.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Productivity.
Claude cannot hear audio. It reads meters, analyser results and spectrogram and waveform pictures, then judges them. It cannot judge performance, emotion, lyrics or taste, and treats deliberate creative choices (lo-fi, heavy limiting) as choices. Click and resonance detection are heuristics, checked against the spectrogram before they are reported.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 13 days ago
- Free with an Agensi account
- Software
Python 3NumPyFFmpeg
- Optional
- anthropic package
Anthropic API key
The anthropic package and ANTHROPIC_API_KEY are only for review.py.
Permissions
Allowed Hosts
File Scopes
Audio is measured locally with ffmpeg; nothing is sent to shash.builds.