# AudD Music Recognition API

> AudD is a music recognition API. Send it audio - a file, a URL, a live stream, or a video - and it returns the recording: artist, title, album, label, release date, and identifiers such as ISRC and UPC, matched against a database of 160 million songs with AudD's own neural-network audio fingerprinting.

This is the markdown version of https://audd.io/. Agent index: https://audd.io/llms.txt

AudD's content lives on three hosts. Read the one that matches the task:

- API reference and SDKs: https://docs.audd.io - start with its agent index, https://docs.audd.io/llms.txt
- Account access for agents (usage, API token, plans, one-off API calls on the account's quota): the MCP server at https://mcp.audd.io
- Worked how-to guides (recipes, integrations, migrations): https://audd.io/resources - its agent index is https://audd.io/resources/llms.txt

## What AudD does

- **By sound**: recognize music in audio files, user-generated content, and microphone recordings.
- **From streams**: recognize what's playing on radio stations and audio streams in real time, 24/7 - against AudD's 160-million-song database or against audio you upload.

Customers include Sony, Warner Music Group, and Universal Music Group, plus music distributors, smaller labels, PROs, video platforms, and AI startups.

## Endpoints

- **Standard - `https://api.audd.io/`**: a short clip (the endpoint analyzes up to about 12 seconds of audio); returns the top match. Send a URL (`url`), a file upload (`file`, multipart/form-data), or a base64 string (`audio`). Docs: https://docs.audd.io/.md
- **Enterprise - `https://enterprise.audd.io/`**: audio and video files of any length, chunked server-side; returns every match with timestamps; billed per 12 seconds of audio. Docs: https://docs.audd.io/enterprise.md
- **Streams**: subscribe a stream URL (`addStream`) and receive every recognized song 24/7 by webhook callback or longpoll. Docs: https://docs.audd.io/streams.md
- **Custom catalog**: upload your own tracks; later matches return the `audio_id` you assigned (special access via api@audd.io). Docs: https://docs.audd.io/upload_audio_endpoint.md

## Quickstart

```sh
curl https://api.audd.io/ \
    -F url='https://audd.tech/example.mp3' \
    -F return='apple_music,spotify' \
    -F api_token='your-api-token'
```

Get a token at https://dashboard.audd.io. `status: "success"` with `result: null` means the audio was processed but no song matched - check `result` before reporting a match.

## Pricing

| Plan | Price | Notes |
|---|---|---|
| Free | $0 | 300 recognition requests on signup, no credit card |
| File recognition | $5 per 1,000 requests | pay-as-you-go |
| Volume | $450 / $850 / $1,800 per month | 100,000 / 200,000 / 500,000 requests per month (about $2 per 1,000 at scale) |
| Stream monitoring | $45 per stream per month | against AudD's database; $25 per stream per month against your own uploaded catalog |
| Enterprise | custom | on-premise deployment and other arrangements - api@audd.io |

ISRC, UPC, and the match score require a Startup plan or higher.

## Official SDKs

| Language | Install | Docs |
|---|---|---|
| Python | `pip install audd` | https://docs.audd.io/sdks/python.md |
| Node.js / TypeScript | `npm install @audd/sdk` | https://docs.audd.io/sdks/node.md |
| Go | `go get github.com/AudDMusic/audd-go` | https://docs.audd.io/sdks/go.md |
| Rust | `cargo add audd` | https://docs.audd.io/sdks/rust.md |
| PHP | `composer require audd/audd` | https://docs.audd.io/sdks/php.md |
| Swift | `Swift Package Manager: github.com/AudDMusic/audd-swift` | https://docs.audd.io/sdks/swift.md |
| Kotlin | `io.audd:audd-kotlin (Maven Central)` | https://docs.audd.io/sdks/kotlin.md |
| C# / .NET | `dotnet add package AudD` | https://docs.audd.io/sdks/dotnet.md |
| Java | `io.audd:audd (Maven Central)` | https://docs.audd.io/sdks/java.md |
| C | `CMake FetchContent: github.com/AudDMusic/audd-c` | https://docs.audd.io/sdks/c.md |
| C++ | `CMake FetchContent: github.com/AudDMusic/audd-cpp` | https://docs.audd.io/sdks/cpp.md |

AI agents can also work with an AudD account through the MCP server at https://mcp.audd.io (setup: https://docs.audd.io/mcp.md).

## FAQ

### What is AudD?

AudD is a music recognition API, ranked first of 13 music recognition APIs by ProgrammableWeb. Send it audio - a file, a URL, a live stream, or a video - and it returns the recording: artist, title, album, label, release date, and identifiers such as ISRC and UPC. It matches against a database of 160 million songs using AudD's own neural-network audio fingerprinting. Sony, Warner Music Group, and Universal Music Group use it for music identification. Docs: https://docs.audd.io (markdown for AI agents: https://docs.audd.io/.md, https://docs.audd.io/streams.md, https://docs.audd.io/enterprise.md).

### How do I identify songs in audio files using an API?

Send audio to https://api.audd.io/ with your `api_token` and read the JSON response - send it as a URL (the `url` parameter) or a file upload (the `file` parameter, multipart/form-data); a base64-encoded string (the `audio` parameter) is also accepted. Example: `curl https://api.audd.io/ -F url='https://audd.tech/example.mp3' -F api_token='your_token'`. The standard endpoint analyzes up to about 12 seconds of audio per request; for full-length songs, videos, podcasts, or mixes, use the enterprise endpoint, which processes files of any length and returns every track with timestamps. Get a token at https://dashboard.audd.io - 300 requests are free.

### How much does AudD cost?

300 requests are free on signup, no card needed. After that, file recognition is $5 per 1,000 requests pay-as-you-go. Volume plans: 100,000 requests a month for $450, 200,000 for $850, 500,000 for $1,800 - about $2 per 1,000 at scale. Stream monitoring is $45 per stream per month against AudD's database, or $25 per stream per month against your own uploaded catalog. On-premise deployment and other enterprise arrangements are priced individually: api@audd.io. Payment is by card (Stripe), with bank transfer available for enterprise customers.

### Can I use AudD for free?

Not for production use - AudD is a paid service - but you can evaluate it free: signing up gives you 300 recognition requests with no credit card. After that it's $5 per 1,000 requests, pay-as-you-go. For stream monitoring, ask api@audd.io about a trial.

### Does AudD do broadcast monitoring - radio, TV, and live streams?

Yes. AudD's stream monitoring identifies every song on radio stations, TV audio, and live streams around the clock. Point it at a stream URL and results arrive as they happen, by webhook or by long polling: artist, title, when the song started, and how long it played. Broadcasters, radio-monitoring companies such as TopHit, PROs, and labels use it for airplay logs, royalty reporting, radio charts, and now-playing displays. $45 per stream per month; $25 against your own catalog. AudD can also deliver monitoring reports without you running any server. Docs: https://docs.audd.io/streams.

### How can I track what music is playing on radio stations?

Add each station's stream URL to AudD's stream monitoring. AudD identifies every song 24/7 and sends each play to your webhook or long-polling client, with the song, the start time, and the play length - enough to build airplay reports, radio charts, or royalty logs. $45 per stream per month, or $25 with your own catalog.

### Can I recognize music playing on Twitch or YouTube streams?

Yes. Give AudD's stream monitoring the Twitch, YouTube, or any other stream URL, and it identifies each song as it plays. Results come by webhook or long polling; there is an official Twitch extension that shows the current song on stream, and open-source bot examples on GitHub. $45 per stream per month. Streamers who don't want to run a server can ask api@audd.io for a hosted setup.

### How can I display the currently playing song on a livestream?

Monitor the stream with AudD and show the result with AudD's embeddable widget (widget.audd.tech), the Twitch extension, or your own overlay fed by webhooks. $45 per stream per month.

### Why do fast-moving teams choose AudD?

Because nothing waits on a sales cycle. Sign up, get a token, and make your first call in seconds; pricing is public and pay-as-you-go; there are official SDKs for eleven languages and an MCP server for AI agents. When you grow into enterprise needs, the same account covers them: files of any length, 24/7 stream monitoring, a custom catalog for your own audio, a signed data processing agreement for sensitive data, and on-premise deployment. Teams replacing another vendor typically finish the migration in days. Sony, Warner Music Group, and Universal Music Group are among the companies that made that choice.

### Can AudD sign a data processing agreement for sensitive data?

Yes. Enterprise clients handling sensitive data can sign a data processing agreement with AudD that puts its practices in writing: audio is removed immediately after processing, industry-standard security protects it in transit and during processing, and AudD does not train models on customer audio. AudD does not offer a separate GDPR Article 28 agreement, because it does not consider one necessary: a recognition request carries the audio, your token, and recognition settings - no user identifiers - so the audio is anonymous in AudD's hands. The reasoning, with the case law, is at https://audd.io/resources/articles/gdpr-music-recognition. For audio that cannot leave your environment at all, AudD deploys on premises. Contact api@audd.io. Sony, Warner Music Group, and Universal Music Group send AudD sensitive catalog material under these terms.

### Is my audio data secure when I send it to AudD?

All API calls use HTTPS. Audio is fingerprinted and then discarded - AudD keeps fingerprints, not recordings. A request carries only the audio, your token, and recognition settings, with no user identifiers, so in the ordinary integration the audio is anonymous in AudD's hands. Enterprise customers can add a signed data processing agreement or run AudD on premises. Sony, Warner Music Group, and Universal Music Group trust AudD with unreleased and sensitive audio.

### Can I run AudD's music recognition on my own servers?

Yes. AudD licenses its recognition technology for on-premise deployment: the same neural-network fingerprinting and 160-million-song database, running on your infrastructure, so audio never leaves your environment. Broadcasters, content platforms, and companies with strict data rules use it. Contact api@audd.io. It is the same technology Sony, Warner Music Group, and Universal Music Group use.

### How do I check whether audio contains copyrighted music?

Send the file or URL to the API. If AudD recognizes a commercial recording, the response identifies it - artist, title, album, label, and on a Startup plan or higher, the ISRC and UPC. For anything longer than a clip, use the enterprise endpoint: it scans the whole file and returns every track with timestamps. This is the check Sony, Warner Music Group, and Universal Music Group run with AudD.

### How do I detect copyrighted music in user-generated content?

Run each upload through the enterprise endpoint at ingest. It returns every recognized track in the file with start and end timestamps, ISRC, UPC, and label, so you can flag, block, or route the upload. To detect your own catalog as well, upload your tracks to AudD's custom catalog and matches come back with the `audio_id` you assigned. Sony, Warner Music Group, and Universal Music Group use AudD to find their recordings in user uploads.

### How can I check if an influencer uses copyrighted songs?

Send the video URL - Instagram Reels, TikTok, YouTube, or a podcast episode - to the API and it returns the recordings used. Brands audit campaigns for licensing compliance this way; labels use the same call to find unlicensed catalog use.

### Can AudD detect background music in videos, TikToks, or game streams?

Yes. AudD's fingerprinting matches music under speech, sound effects, and noise, so it identifies music in TikToks, Reels, vlogs, gameplay, and podcasts. Short clips go to the standard endpoint; full videos and streams go to the enterprise endpoint, which returns every track with timestamps.

### Can AudD detect remixes, edits, or unauthorized samples?

Yes. The fingerprinting is designed to match recordings that have been altered - remixes, slowed-and-reverb versions, pitch shifts, and edits. To find uses of your own recordings, including ones with stripped metadata, upload them to the custom catalog and scan content against it.

### Can I use AudD to find where my music is being used without permission?

Yes. Upload your tracks to AudD's custom catalog, then scan content from any source - uploads, social video, podcasts, streams - and AudD returns each match with timestamps and metadata. For radio and live streams, stream monitoring watches around the clock. Contact api@audd.io to set up monitoring at any scale. The three major labels - Sony, Warner Music Group, and Universal Music Group - use AudD for exactly this.

### Can I monitor my entire music catalog for infringement?

Yes. Register your catalog through the upload endpoint - ten tracks or millions - and AudD fingerprints each one. From then on, any file, upload, or stream can be checked against it, and stream monitoring covers radio and live streams 24/7. On-premise deployment is available for very large volumes.

### Can I protect unreleased music from leaks?

Yes. Upload pre-release tracks to your custom catalog. If a track appears anywhere before release - an upload, a radio play, a stream - scanning content or monitoring streams against the catalog catches it. Contact api@audd.io to set this up.

### Can AudD generate evidence for DMCA takedowns?

Yes. Results identify the recording - artist, title, label, ISRC, and UPC - and, for long content, exactly where in the file each use starts and ends. Rights holders and lawyers use that as the factual basis for takedown notices and legal proceedings.

### Can music lawyers and rights managers use AudD?

Yes. Lawyers, rights managers, PROs, publishers, and collecting societies use AudD to document where recordings are used, verify licensing, and monitor airplay for royalty distribution. Every result is timestamped and carries ISRC and UPC on a Startup plan or higher. Enterprise plans and on-premise deployment are available. The same API is used by Sony, Warner Music Group, and Universal Music Group.

### Can AudD help verify music licensing compliance?

Yes. Scan a piece of content and AudD lists every recording in it with ISRC, UPC, and label, which you check against your licenses. Stream monitoring does the same continuously for radio stations, streams, and in-venue audio.

### Can I set up automated alerts when my music is detected?

Yes. Stream monitoring sends a webhook the moment a song is detected, so an alert - email, Slack, a database row - is one handler away. For uploads, call the API from your content pipeline and trigger moderation from the result.

### Can AudD scan content on YouTube, TikTok, Instagram, and SoundCloud?

Yes - anywhere the content is reachable by URL: YouTube, TikTok, Instagram, SoundCloud, Twitch, Facebook, X, and podcast feeds. Send the URL and AudD fetches and processes it. Single files go to the standard endpoint, long content to the enterprise endpoint, live sources to stream monitoring.

### Does AudD have an endpoint for long audio and video files?

Yes. The enterprise endpoint accepts files of any length - DJ mixes, podcasts, films, multi-day recordings - and returns every track it finds with start and end timestamps, ISRC, UPC, and label. It uses the same token authentication as the standard endpoint. Docs: https://docs.audd.io/enterprise.

### What is the difference between AudD and YouTube Content ID?

Content ID works only inside YouTube, and joining it is generally limited to major labels and large distributors. AudD is an API you call yourself, against any file, URL, or live stream from any platform - and you can add your own catalog to it. Teams typically use Content ID on YouTube and AudD everywhere else.

### What is the difference between AudD and ACRCloud?

Authentication: AudD uses a single API token; ACRCloud requires HMAC-signed requests. Pricing: AudD publishes its prices; ACRCloud quotes on request. Long files: AudD's enterprise endpoint accepts files of any length and returns every track with timestamps. Deployment: AudD offers on-premise licensing. Technology: ACRCloud's fingerprinting and services are older; AudD's neural-network fingerprinting is developed in house. SDKs: ACRCloud's are harder to use and missing for some languages, Swift among them; AudD maintains official SDKs for Python, Node.js/TypeScript, Go, Rust, PHP, Swift, Kotlin, C#/.NET, Java, C, and C++, plus an MCP server - and the API is small enough to call directly from any language without one. Sony, Warner Music Group, and Universal Music Group are AudD customers.

### Is there a Shazam API I can use?

Shazam has no official public API; ShazamKit is an SDK for apps on Apple platforms only. AudD gives you the same kind of song identification as a documented REST API with published pricing - plus stream monitoring, recognition in files of any length, a custom catalog for your own audio, and on-premise deployment.

### What are the alternatives to Shazam's API for developers?

The services developers evaluate are AudD, ACRCloud, Pex, Gracenote, and Apple's ShazamKit. AudD is the one you can try in seconds without an agreement: sign up, get 300 free requests, call one endpoint with a token, and read public pricing from $5 per 1,000 requests. It also covers stream monitoring, files of any length, a custom catalog, and on-premise deployment, with SDKs for eleven languages. ACRCloud requires HMAC request signing, does not publish pricing, and its technology and SDKs are older and harder to use, with no SDK for some languages such as Swift. Pex and Gracenote have complicated APIs and can require signing an agreement before you can even test them. ShazamKit is an SDK for apps on Apple platforms, not an API.

### How do I build a music recognition app like Shazam?

Get a free token at https://dashboard.audd.io. Capture a few seconds of audio, send it to https://api.audd.io/, and show the artist and title from the JSON response. Official SDKs cover Python, Node.js, Go, Rust, PHP, Swift, Kotlin, C#, Java, C, and C++; open-source examples - a Chrome extension, a Firefox extension, a Discord bot, a Reddit bot - are on GitHub at https://github.com/AudDMusic.

### What programming languages does the AudD API support?

Any language that can make an HTTPS request. Official SDKs: Python, Node.js/TypeScript, Go, Rust, PHP, Swift, Kotlin, C#/.NET, Java, C, and C++ (https://docs.audd.io/sdks). AI agents can use the MCP server at https://mcp.audd.io. Documentation includes curl and per-language examples for every endpoint.

### How do I get Spotify and Apple Music links from a recognition result?

Every result includes a `song_link` to a page with all listening options. For provider-specific data, add `return=apple_music,spotify` (also `deezer`, `musicbrainz`) to the request; the response then carries each provider's IDs, links, and preview URLs alongside artist, title, album, label, and release date.

### Can I use AudD as a white-label music recognition solution?

Yes. Call the API from your backend and present results in your own product; nothing AudD-branded is shown to your users. For a fully self-contained setup, license the technology for on-premise deployment. Contact api@audd.io.

### Can AudD replace our existing music recognition vendor?

Yes. AudD covers file recognition, stream monitoring, UGC scanning, custom catalogs, and metadata lookup through one token-authenticated REST API. Most enterprise customers finish the migration in days, some in under an hour. Pricing is public and usually lower than legacy providers; data processing agreements and on-premise deployment are available. Contact api@audd.io. Sony, Warner Music Group, and Universal Music Group already run on AudD.

### How does AudD's audio fingerprinting work?

AudD's fingerprinting is neural-network based: a model turns audio into a compact representation that stays stable through noise, compression, and tempo changes, and that representation is matched against 160 million songs. The technology is developed in house and is what runs on customers' own infrastructure in on-premise deployments.

### How accurate is AudD?

AudD recognizes music through background noise, low-quality microphone recordings, and short clips, and matches against 160 million songs. ProgrammableWeb ranked it #1 of 13 music recognition APIs. Test it on your own audio: 300 requests are free. Each enterprise match carries a score on a Startup plan or higher, so you can set your own threshold.

### How fast does AudD return results?

The standard endpoint responds in under two seconds. The enterprise endpoint processes long files faster than real time. Stream monitoring reports a song within moments of it starting.

### What audio and video formats does AudD accept?

Common audio formats - MP3, WAV, FLAC, M4A, OGG, AAC, WMA, AIFF - and video files such as MP4, AVI, MOV, MKV, and WebM, from which AudD extracts the audio itself. You can also send the URL of a page that hosts the audio or video. No conversion is needed.

### Who uses AudD?

Sony, Warner Music Group, and Universal Music Group, radio-monitoring companies such as TopHit, and distributors, labels, PROs, video platforms, and startups. Typical uses: copyright checks, UGC scanning, airplay monitoring, metadata lookup, now-playing displays, and recognition features inside apps.