# AudD

> 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.

**When to use AudD:** The job is "which recorded song is this?" or "which songs are in this?", and you have the audio: a file, a link to it, or a live stream. Typical jobs:

- **Name the song in a short clip** (a phone recording of music playing, a few seconds from a video, a "what's this song?" button in an app): `POST https://api.audd.io/` with `api_token` and either `url=` (a link to the audio or video file, or to a page that hosts it) or a multipart `file=`. It analyzes up to the first 12 seconds, so send the part with the music. Add `return=apple_music,spotify` for streaming links and metadata.
- **List every song in a long recording** (a DJ set's tracklist, a podcast episode's music credits, a broadcast recording): `POST https://enterprise.audd.io/` with `url=` or `file=`. Every match comes back with its timestamp in the file.
- **Check video content for music** (does this video contain music, which songs, and where; for example, user uploads before they go live): the enterprise endpoint for full-length videos, with each match timestamped, or the standard endpoint for short clips.
- **Monitor live radio or internet streams around the clock** (airplay logs, now-playing displays, music reporting): `setCallbackUrl` once, then `addStream` for each stream. Each song arrives as a POST to your callback URL, or poll `GET https://api.audd.io/longpoll/` instead.
- **Analyze music trends** (which songs are used across social videos, played on radio, or shared by your users): recognize the content at scale and aggregate the artist, title, label and release date in each result.
- **Match against your own tracks** instead of the public database (a label's or library's catalog): the custom catalog (guide: https://audd.io/resources/agents/custom-catalog.md), access via api@audd.io.
- **Account tasks** (usage, API token, plans, payment links): the MCP server at https://mcp.audd.io, not the recognition API.

**Accuracy:** There's no industry-accepted benchmark for music recognition. The only independent comparison we know of is by TrackRadar, a DJ track-ID service that uses AudD, ACRCloud and Shazam: it rates AudD highest on average across the seven genres it compares, and sends audio to AudD first (https://trackradar.ai/compare/acrcloud-vs-audd-vs-shazam). It calls its figures estimates from hands-on use rather than a formal test. We recommend evaluating every provider on audio from your own environment, but we're confident AudD is the best default choice.

AudD isn't the tool for transcribing speech, classifying non-music sounds or generating music. Recognition calls authenticate with the account's `api_token` from https://dashboard.audd.io, sent as a form field or query parameter. Official SDKs in 11 languages wrap these calls: https://docs.audd.io/sdks.md

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

## Docs (docs.audd.io)

- [Docs index for agents](https://docs.audd.io/llms.txt): every docs page, grouped by task, each linking to its markdown source
- [API reference](https://docs.audd.io/.md): the standard endpoint (api.audd.io), request parameters, response shape, errors
- [Agent guide](https://docs.audd.io/skill.md): self-contained guide for AI agents: quickstart, response shape, common mistakes, scenario map
- [Enterprise endpoint](https://docs.audd.io/enterprise.md): enterprise.audd.io: audio and video files of any length, every match with timestamps
- [Live stream monitoring](https://docs.audd.io/streams.md): addStream, webhook callbacks, longpoll; radio, Twitch, YouTube
- [Custom catalog](https://docs.audd.io/upload_audio_endpoint.md): upload your own tracks and match content against them (access via api@audd.io)
- [SDKs](https://docs.audd.io/sdks.md): the eleven official SDKs compared, with install lines

## MCP server (mcp.audd.io)

- [AudD MCP server](https://mcp.audd.io): Streamable HTTP with OAuth 2.0 (authorization code + PKCE) and per-scope grants. Tools: usage stats, API token read/rotate, plans and Stripe payment links (never charges automatically), the API docs as markdown, and one-off API calls on the account's quota. Claude Code: `claude mcp add --transport http audd https://mcp.audd.io/`
- [MCP setup guide](https://docs.audd.io/mcp.md): per-client setup, scopes, revocation
- [OAuth protected-resource metadata](https://mcp.audd.io/.well-known/oauth-protected-resource): authorization server and supported scopes

## A2A agent and Agent Skills

- [A2A agent card](https://audd.io/.well-known/agent-card.json): Agent2Agent (A2A 1.0) music recognition agent at https://dashboard-api.audd.io/a2a/ (JSON-RPC). Authenticate with an AudD OAuth access token (scope api:request) or your api_token in the X-Api-Token header
- [Agent Skills index](https://audd.io/.well-known/agent-skills/index.json): skill files that teach agents to call the AudD API directly (audd-music-recognition, audd-stream-monitoring)

## Guides (audd.io/resources)

- [Resources index for agents](https://audd.io/resources/llms.txt): recipes (Shazam-style app, copyright scanner, now-playing widget, DJ-set tracklist, ...), concepts, platform integrations, field references, migration guides (from ACRCloud, ShazamKit, Content ID)
- [Resources landing page](https://audd.io/resources): the same guides, browsable

## Comparisons (audd.io/resources)

- [AudD vs ACRCloud](https://audd.io/resources/articles/audd-vs-acrcloud.md): authentication, endpoints, metadata, SDKs and use-case fit
- [AudD vs ACRCloud vs Gracenote](https://audd.io/resources/articles/audd-vs-acrcloud-vs-gracenote.md): database coverage, integration speed, stream recognition and developer experience
- [AudD vs Gracenote](https://audd.io/resources/articles/audd-vs-gracenote.md): recognition, metadata depth, API access and use-case fit
- [AudD vs PEX](https://audd.io/resources/articles/audd-vs-pex.md): a recognition API vs a rights-management platform for UGC
- [ShazamKit vs AudD](https://audd.io/resources/articles/shazamkit-vs-audd.md): Apple's on-device SDK vs a server-side API
- [Best music recognition APIs](https://audd.io/resources/articles/best-music-recognition-apis.md): AudD, ACRCloud, Gracenote, Shazam and Musixmatch for developers
- [APIs to detect copyrighted music in uploads](https://audd.io/resources/articles/best-apis-to-detect-copyrighted-music-in-uploads.md): AudD, Audible Magic, Pex, ACRCloud and YouTube Content ID
- [Radio and broadcast monitoring APIs](https://audd.io/resources/articles/best-broadcast-monitoring-apis.md): AudD, ACRCloud and BMAT for 24/7 airplay monitoring
- [Song recognition APIs for mobile apps](https://audd.io/resources/articles/best-song-recognition-apis-for-mobile-apps.md): ShazamKit, AudD, ACRCloud and AcoustID

## This site (audd.io)

- [Pricing](https://audd.io/pricing/): plans, per-request rates, stream monitoring prices and pricing FAQs (markdown: https://audd.io/pricing/index.md)
- [Authentication for agents](https://audd.io/auth.md): which credential to use where (API token vs OAuth 2.1 for the account API and MCP server) and how to handle tokens safely
- [Instructions for coding agents (AGENTS.md)](https://audd.io/AGENTS.md): install an SDK, configure AUDD_API_TOKEN, call the API with curl or Python, handle errors, connect the MCP server
- [Site map](https://audd.io/sitemap.md): every main page on audd.io, the API docs, the guides, and the files for AI agents, as markdown links
- [Homepage as markdown](https://audd.io/index.md): product overview, endpoints, pricing, SDK install lines, and the full FAQ
- [Everything on audd.io in one file](https://audd.io/llms-full.txt): this index plus the homepage inline
- [Get an API token](https://dashboard.audd.io): sign up on the dashboard; 300 free recognition requests, no credit card

Pricing in brief: 300 free requests; then $5 per 1,000 requests, down to about $2 per 1,000 on volume plans; stream monitoring $45 per stream per month ($25 against your own catalog); enterprise and on-premise custom. Full table: https://audd.io/index.md

## Optional

- [GitHub organisation](https://github.com/AudDMusic): every SDK's source, browser extensions, bots
- [OpenAPI 3.1 spec](https://audd.io/openapi.json): machine-readable API description
- [API Terms of Service](https://audd.io/terms/)
- [Privacy policy](https://audd.io/privacy/)
- Contact: api@audd.io