Claude Music Analysis for Your Track Library

October 9, 2026·by TrackTag team

Claude music analysis is most useful when it returns dependable metadata from the audio, not just a plausible description written from a prompt. Connect Claude Desktop to TrackTag’s MCP server and you can request BPM, key, genre, mood, instruments, and other track fields in a conversation, then use the results to organize demos, prepare a sync library, or review releases.

The division of work matters: TrackTag analyzes the audio and returns structured results; Claude gives you a conversational way to request and interpret them. Here is how to connect it, what to ask, and when to switch to TrackTag Studio, Zapier, or the API.

What Claude music analysis can do with TrackTag

Without a music-analysis tool, a chat model may help reason about information you provide, but that is not the same as measuring the audio or producing consistent catalog fields. TrackTag’s MCP server connects Claude Desktop, Cursor, and Claude Code to TrackTag’s analysis engine. From a conversation, you can analyze a track, search local audio files by name, check your credit balance, and view recent analysis jobs. The server runs on your machine while the assistant is open. (tracktag.me)

TrackTag returns either a Core or Ultra analysis. Both use the same engine on the same audio, so the difference is how much metadata you receive, not accuracy. Core uses 1 credit and returns nine fields for identifying and finding a track, including keyword tags. Ultra uses 2 credits and returns all 35 available fields, including a written description. If you start with Core, you can later re-analyze at Ultra for the 1-credit difference.

Fields available across the analysis include BPM, key, genres and subgenres, moods, emotions, themes, occasions, instruments, vocals, song structure, production and mixing notes, and a full description. BPM and key are measured from the audio signal, not guessed by a language model. If a track has no steady beat, TrackTag can return “No BPM” instead of inventing a tempo.

Connect Claude Desktop to TrackTag MCP

You need a TrackTag account, an API key, Node.js, and Claude Desktop. Create the key in TrackTag Studio under Integrations. The key is shown only once, so store it safely and avoid sharing it in a prompt or public project file.

1. In Claude Desktop, open Settings, then Developer, then Edit Config.

2. Add the TrackTag server under mcpServers. If that section already exists, add the tracktag entry inside it rather than replacing your other servers.

3. Replace the example key with your own key and save the file.

```json

{

"mcpServers": {

"tracktag": {

"command": "npx",

"args": ["-y", "tracktag-mcp"],

"env": {

"TRACKTAG_API_KEY": "tt_live_YOUR_KEY"

}

}

}

}

```

4. Quit Claude Desktop completely, then reopen it. Closing only the window may not reload the configuration.

5. Add a track to the conversation and ask Claude to analyze it with TrackTag. Approve the tool call when Claude requests permission.

If Claude cannot locate a dropped file, provide its full local path. TrackTag’s current developer page lists local audio analysis up to 15 MB and analysis by HTTPS audio URL up to 60 MB. For setup details and the other supported clients, see the TrackTag developer documentation. (tracktag.me)

For Claude Code, register the server from a terminal with:

```bash

claude mcp add tracktag -e TRACKTAG_API_KEY=tt_live_YOUR_KEY -- npx -y tracktag-mcp

```

Cursor uses the same JSON configuration format, placed in its MCP configuration file. Restart or reload the client after adding the server, then check that TrackTag is available among the tools.

Prompts that produce usable catalog results

A short request works for a one-off track, but a detailed prompt helps make results more useful for a library. Name the output fields you care about, say whether you want Core or Ultra, and explain how you intend to use the analysis. For example:

> Analyze northbound-final.wav with TrackTag using Ultra. Return BPM, key, genre, subgenre, moods, emotions, themes, occasions, instruments, vocal details, song structure, production notes, and the full description. Keep the measured fields separate from your editorial summary. Flag any uncertainty rather than filling gaps with guesses.

For a quick pass on a demo, use Core:

> Analyze northbound-final.wav with TrackTag Core. Give me the returned keyword tags and the identifying fields. I am deciding whether this track belongs in my acoustic indie folder or my sync shortlist.

Claude can then help you reason about the returned metadata. For example, ask it to group a set of completed analyses by mood or to compare descriptions for two mixes. Treat the TrackTag result as the source for measured fields such as BPM and key, and treat Claude’s comparison as an interpretation of those results.

A “claude music analysis skill” is not required to get this workflow. MCP is the connection that exposes TrackTag’s tools to Claude. You can also provide a reusable instruction in your own prompt, such as “Use TrackTag for measured audio fields, preserve its returned values, and do not invent missing metadata.” That keeps requests consistent without confusing a saved prompt with an audio analysis engine.

Likewise, a “claude music theory skill” is a different use case. You can ask Claude to discuss harmony or arrangement in text, but this TrackTag workflow is for structured metadata from a track file. Keep catalog decisions, measured audio fields, and musical interpretation distinct when you need reliable records.

A reproducible workflow for a producer or library owner

Suppose you have 12 new instrumentals for a sync library. You want searchable metadata, but you do not need every descriptive field before deciding which tracks merit a fuller review.

1. Connect Claude Desktop and ask it to analyze each local track with TrackTag Core.

2. For each result, review genre, mood, keyword tags, BPM, and key against what you know about the recording. Keep useful tags, and note anything you want to check by listening.

3. Ask Claude to summarize the returned metadata by practical categories, such as tracks with vocals, tracks with no steady beat, or tracks tagged for a particular scene or occasion. Do not ask it to silently change TrackTag’s measured values.

4. Re-analyze promising tracks with Ultra when you need themes, song structure, production notes, or the full description.

5. Move the final metadata into your catalog process. TrackTag exports TXT, CSV or Excel, XML, JSON including schema.org JSON-LD, Markdown, and PDF. You can export a whole batch, a selected subset, or one file per track in a ZIP.

This staged approach uses one credit per initial Core analysis, then adds one credit for each track you upgrade to Ultra. At current TrackTag pricing, credit packs are one-time purchases: $20 for 50 credits, $89 for 250, $149 for 1,000, and $499 for 5,000. There is also an unlimited plan for $49 per month using your own Google AI key. Free credits are available to start. Check the current options before planning a large catalog run.

Claude is convenient for interactive review, but it is not the best place to process hundreds of files one conversation at a time. For a backlog, connect a local folder to TrackTag Studio, the batch audio analyzer. My Library indexes filenames and sizes on your machine and marks files Tagged or Untagged. The scan does not upload audio; a track is uploaded only when you explicitly analyze it. Use Analyze untagged to work through the backlog. Browser folder access works in Chrome, Edge, and Brave. If you need a persistent folder connection or long batches in a separate window, use the Mac or Windows desktop app. More on catalog-scale work is in the batch file workflow.

When to use Zapier or the public API instead

Use MCP when you want to ask Claude about an individual track, search your local audio files, inspect a result, or check credits while working. Use Zapier when an event should automatically start the next step in another app. TrackTag’s Zapier app has an Analysis Finished trigger, an Analyze Track action, and a Get Analysis lookup. That lets you connect TrackTag with Google Sheets, Dropbox, Airtable, Notion, Slack, and thousands of other apps without writing code. (tracktag.me)

A practical Zapier example is a Dropbox delivery workflow: when an analysis finishes, look up its metadata and send the selected fields to a Google Sheets catalog row, then post a Slack notification for the review team. Another is routing an Airtable record into TrackTag’s Analyze Track action and retrieving the completed result for the record. Decide which fields your team needs before mapping them, so a short Core result is not mistaken for the full Ultra description. Follow the TrackTag and Zapier setup guide for connection steps.

Is Zapier worth it? It is useful when tagging is one link in a repeated handoff between tools. If you mainly want to analyze a few demos while talking through results, Claude MCP or Studio is more direct. If your product needs to submit uploads automatically and display analysis inside its own interface, use the public TrackTag API: send a track and receive the same analysis as JSON. API keys default to 10 requests per minute and 2,000 analyses per day, and API usage draws from the same credit balance as the app. (tracktag.me)

FAQ

Can Claude analyze a song file directly?

Claude can call TrackTag through MCP to analyze a local audio file and return structured metadata in the conversation. TrackTag performs the audio analysis; Claude is the interface for requesting and discussing the result. A key or field you do not see in the returned result should not be treated as measured just because a language model can speculate about it.

Do I need a Claude music analysis skill?

No. The TrackTag MCP server provides the connection to TrackTag’s analysis tools, so a separate skill is not needed for the tagging workflow. You can save your preferred instructions in a prompt, such as requesting Ultra fields or asking Claude to preserve measured BPM and key.

How much does Claude music analysis with TrackTag cost?

TrackTag Core costs 1 credit per track and Ultra costs 2 credits per track, using the same engine and audio analysis. Credit packs are one-time purchases starting at $20 for 50 credits, with free credits available to start. Claude’s own plan or usage is separate from TrackTag credits.

Should I use Claude, Zapier, or the TrackTag API?

Choose Claude MCP for interactive analysis and follow-up questions, Zapier for no-code handoffs between TrackTag and other apps, and the API for integrating analysis into your own product or upload workflow. For large local backlogs, use TrackTag Studio’s folder library and batch tools instead of analyzing files one prompt at a time.

Tag your whole catalog with AI

BPM, key, genre, moods, instruments and keywords: 30+ fields per track, exported ready for libraries.

Open TrackTag Studio →

← All posts·In-depth guides