AudioTag Identify Song Upload Audio: Official Checklist

If you typed some version of "audiotag identify song upload audio official" into a search bar, you're trying to do one of two things: name a mystery track, or figure out why a free fingerprinting tool won't tag the music you actually own. Both are solvable in an afternoon, but they need different tools. Here is the checklist for both halves of that job.

What the official AudioTag upload actually does

The official site is audiotag.info, and its one job is recognition, not description. After you upload an audio file, it is analyzed by the underlying audio engine, an audio fingerprint is extracted from the acoustic content and identified using a fingerprint matching algorithm against a large database of music fingerprints, and information about matched song candidates is displayed. The service is free for non-commercial use.

That database is large but it only knows songs that were already fingerprinted. The underlying core audio fingerprinting technology works with purely acoustic features of the sound and does not consider melodic or musical properties, so it can only find tracks with exactly the same sound. It cannot recognize a cover, a remix, or a track that was never entered into its database, and it has no concept of key, tempo, or mood, because it isn't listening for those things.

The afternoon checklist for identifying a track

Work through these in order and you'll get a clean answer or a clean "no match" in under ten minutes per clip.

1. Confirm the file plays cleanly. No clipping, no dead air at the start.

2. Trim to the right length. Even a 10 second fragment is enough for the robot to analyze, though a full song works too.

3. Check the format. Uncompressed and ADPCM WAV, MP3, OGG, FLAC, FLV, AMR, and MP4 are all supported. Convert obscure formats before uploading rather than guessing.

4. Use a link if you have one. AudioTag can also recognize audio files directly from YouTube and other video sharing services, which saves a download-and-reupload step.

5. If the first attempt returns nothing, don't give up on the clip. If the fragment isn't informative enough, uploading a longer recording may increase the chances of successful recognition.

6. For recordings with several songs back to back, know that the engine can recognize different songs contained in different fragments of the same recording, returning results with time intervals relative to the start of the file.

If you're automating a lot of lookups rather than doing them one at a time, AudioTag also ships an API. The API offers full automation of music recognition queries and allows integrating AudioTag music identification functionality with third-party systems, services and software applications. That's worth knowing if a workflow needs to check incoming files against known releases before anything else happens to them.

Where the official tool stops helping you

Here's the wall almost everyone hits eventually: you run your own catalog through AudioTag hoping for tags, and it either returns nothing or, at best, confirms the track is yours by matching it to an earlier upload. That's expected behavior, not a bug. Fingerprinting tools answer "what is this song and who made it," by matching against previously catalogued recordings, not "what does this song sound like." An unreleased demo, an unmastered mix, or a stem has no prior fingerprint to match against, so there's nothing for the engine to find.

If that's the gap you actually ran into, the rest of this checklist is for you. If you want the longer breakdown of what fingerprinting can and can't do, the AudioTag alternatives guide covers it in more depth.

Finish the afternoon: tag the tracks that are actually yours

Describing your own catalog is a different job with a different tool. TrackTag Studio batch audio analyzer listens to the audio itself and returns up to 35 fields per track: BPM, key, genres, subgenres, moods, emotions, themes, occasions, instruments, vocals, song structure, production notes, and a full written description.

Two analysis levels run on the same engine, so accuracy doesn't change between them, only how much of the answer you get back. Core, at 1 credit, returns the 9 fields that let you file and find a track, keyword tags included. Ultra, at 2 credits, unlocks all 35 fields including the description. You can start a batch on Core to get everything organized quickly, then send the tracks that need licensing pitches or catalog write-ups back through at Ultra later for the 1-credit difference.

Tempo and key aren't guessed from genre averages either. Precision Mode measures BPM and musical key directly from the audio signal on-device, benchmarked at 15 out of 15 tempo agreement against the leading industry analyzer. If your catalog spans multiple tempos and time signatures, that measured approach matters more than it sounds like it should, and it's worth understanding how the detection actually works before you trust a tool with hundreds of files. The BPM and key detection guide walks through the mechanics.

Once tracks are tagged, export in whatever format the next step needs: TXT, CSV or Excel, XML, JSON including schema.org JSON-LD, Markdown, or PDF, as a full batch, a hand-picked subset, or one file per track zipped up.

Turn the checklist into a standing workflow

A single afternoon session clears a backlog once. The tools below keep it clear.

My Library connects a local folder and shows every file as Tagged or Untagged with sortable columns, folder aliases, and an "Analyze untagged" action that queues up just the backlog. The scan itself is local: file names and sizes are read on your own machine, and nothing uploads until you actually analyze a track. Browser folder access works in Chrome, Edge, and Brave; Safari and Firefox don't support it, which is why the desktop app exists for Mac and Windows, keeping folders connected permanently without repeated permission prompts.

For teams and pipelines, the public API takes a POST request and returns the same JSON analysis the app produces, on the same credit balance, with keys created in Studio and a default of 10 requests per minute and 2,000 analyses a day. It's built for marketplaces auto-tagging uploads and labels enriching deliveries, the kind of use case people originally hoped a fingerprinting API could cover. The batch tagging guide covers folder structures and naming conventions that keep large libraries consistent as they grow.

If your team already lives in Claude, Cursor, or Claude Code, the MCP server lets those assistants analyze a track, search your local files by name, and check your credit balance from inside a conversation, running entirely on your own machine. And if the rest of your stack is Zapier, the "Analysis Finished" trigger, "Analyze Track" action, and "Get Analysis" lookup connect straight into Google Sheets, Dropbox, Airtable, Notion, or Slack with no code.

The bottom line for one afternoon of work

Use the official AudioTag upload for what it's built for: naming a mystery clip against its fingerprint database. Then switch tools for the job it was never designed to do, describing and organizing the music you already own. Run the same folder through TrackTag Studio, pick Core or Ultra depending on how much detail you need right now, and export whatever your DAW, spreadsheet, or delivery system expects. That's a realistic afternoon: one tool to identify, one tool to describe, and a library that's actually searchable by the end of it.