AZLyrics Bulk Lyrics Scraper

Scrape song lyrics in bulk from AZLyrics.com at the cheapest per-song rate on the Apify platform. At $0.0015 per song, scraping 10,000 songs costs just $15. Built for NLP researchers, computational linguists, and dataset builders who need large lyrics corpora without breaking the bank. Translate lyrics to any language with built-in AI translation.

AZLyrics hosts one of the largest collections of song lyrics on the web, covering hundreds of thousands of artists across every genre and language. This Actor extracts clean, plain-text lyrics along with structured metadata (artist, album, title) ready for direct use in machine learning pipelines, text analysis, and research datasets.

Features

• Bulk-optimized scraping -- scrape entire artist discographies or target individual songs
• Clean text output -- lyrics extracted as clean plain text, free of HTML artifacts, ads, and formatting noise
• AI-powered translation -- translate lyrics to any language via OpenAI-compatible APIs, with built-in service or bring your own key (BYOK) for lower cost
• Structured metadata -- song title, artist name, album name, and source URL for every record
• Artist catalog extraction -- full discography listings with album groupings and related artists
• Multi-strategy search -- provide artist + song fields, free-text queries, or direct AZLyrics URLs
• Google fallback -- when direct URL guesses fail, the Actor falls back to site:azlyrics.com Google search
• Anti-bot resilience -- PlaywrightCrawler with Firefox, fingerprint generation, session rotation, and residential proxy support to handle Cloudflare and reCAPTCHA challenges

How to Use

1. Scrape a specific song

Provide the artist name and song title for the most reliable direct lookup.

{
    "artistName": "Radiohead",
    "songTitle": "Creep",
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

2. Scrape all songs by an artist

Set only the artist name. The Actor locates the artist page, extracts the full catalog, then scrapes lyrics for every song (up to maxResults).

{
    "artistName": "Radiohead",
    "maxResults": 0,
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

3. Free-text search

Use a natural query. The Actor tries multiple URL patterns and falls back to Google if needed.

{
    "searchQuery": "Pomme Soleil Soleil",
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

4. Direct AZLyrics URLs

Provide one or more AZLyrics URLs. The Actor auto-detects whether each URL is a song page, artist page, or letter index.

{
    "startUrls": [
        { "url": "https://www.azlyrics.com/lyrics/radiohead/creep.html" },
        { "url": "https://www.azlyrics.com/r/radiohead.html" }
    ],
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

5. Scrape with AI translation

Add translateTo to translate every song's lyrics. Works out of the box at $0.008/song, or provide your own OpenAI-compatible API key for $0.002/song.

{
    "artistName": "Pomme",
    "translateTo": "English",
    "maxResults": 10,
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

With your own API key (lower cost):

{
    "artistName": "Pomme",
    "translateTo": "English",
    "translateApiKey": "sk-your-openai-key",
    "maxResults": 10,
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

You can also use non-OpenAI providers by setting translateBaseUrl and translateModel:

{
    "artistName": "Pomme",
    "translateTo": "English",
    "translateApiKey": "your-api-key",
    "translateBaseUrl": "https://generativelanguage.googleapis.com/v1beta/openai/",
    "translateModel": "gemini-2.0-flash",
    "proxyConfiguration": {
        "useApifyProxy": true,
        "apifyProxyGroups": ["RESIDENTIAL"]
    }
}

Input Parameters

Parameter	Type	Default	Description
artistName	string	--	Artist or band name. Used alone to scrape all songs, or with songTitle for a specific song.
songTitle	string	--	Song title to search for. Requires artistName.
searchQuery	string	--	Free-text search query. The Actor guesses artist/song from the text.
startUrls	array	--	Direct AZLyrics URLs to scrape. Format: [{ "url": "..." }]
maxResults	integer	50	Maximum number of songs to scrape. Set to 0 for unlimited.
translateTo	string	"English"	Target language for AI translation (e.g. "Spanish", "Japanese"). Leave empty to skip.
translateApiKey	string	--	Your OpenAI-compatible API key (BYOK). Omit to use the built-in service.
translateModel	string	"gpt-4o-mini"	Chat completions model. Only relevant when using your own API key.
translateBaseUrl	string	--	API base URL for non-OpenAI providers.
proxyConfiguration	object	--	Proxy settings. Residential proxies are strongly recommended.

You must provide at least one of artistName, searchQuery, or startUrls.

Output

The Actor produces two types of records in the same dataset. Use the dataset views in the Apify Console to filter them.

Song result

Each song produces one record with lyrics and metadata.

{
    "title": "Creep",
    "artist": "Radiohead",
    "album": "Pablo Honey",
    "lyrics": "When you were here before\nCouldn't look you in the eye\nYou're just like an angel\nYour skin makes me cry\n\nYou float like a feather\nIn a beautiful world\nI wish I was special\nYou're so very special\n\nBut I'm a creep\nI'm a weirdo\nWhat the hell am I doing here?\nI don't belong here",
    "translatedLyrics": null,
    "url": "https://www.azlyrics.com/lyrics/radiohead/creep.html",
    "scrapedAt": "2025-02-15T10:30:00.000Z"
}

Field	Type	Description
title	string	Song title
artist	string	Artist or band name
album	string or null	Album name when available on the page
lyrics	string	Full lyrics as clean plain text
translatedLyrics	string or null	AI-translated lyrics, or null if translation was not requested
url	string	Source URL on AZLyrics
scrapedAt	string	ISO 8601 timestamp

Artist catalog result

When an artist page is scraped, the Actor produces a catalog record with the full discography and then enqueues individual song pages.

{
    "type": "artist-catalog",
    "artist": "Radiohead",
    "artistUrl": "https://www.azlyrics.com/r/radiohead.html",
    "totalSongs": 162,
    "songs": [
        {
            "title": "You",
            "url": "https://www.azlyrics.com/lyrics/radiohead/you.html",
            "album": "Pablo Honey"
        },
        {
            "title": "Creep",
            "url": "https://www.azlyrics.com/lyrics/radiohead/creep.html",
            "album": "Pablo Honey"
        }
    ],
    "relatedArtists": ["Thom Yorke", "Atoms for Peace", "Muse"],
    "scrapedAt": "2025-02-15T10:25:00.000Z"
}

Field	Type	Description
type	string	Always "artist-catalog"
artist	string	Artist or band name
artistUrl	string	Artist page URL
totalSongs	number	Total songs in catalog
songs	array	List of { title, url, album } entries
relatedArtists	array	Names of related artists listed on the page
scrapedAt	string	ISO 8601 timestamp

Dataset views

The Apify Console provides four pre-configured views:

• Songs -- table of scraped songs with title, artist, album, URL (filtered to song records only)
• Lyrics -- table with full lyrics text and translations side-by-side
• Artist Catalogs -- summary table of artist catalogs with song counts and related artists
• Catalog Songs (expanded) -- the songs array from each catalog unwound into individual rows

Pricing

This Actor uses Pay Per Event (PPE) pricing. You only pay for successfully scraped data. There are no platform fees, no subscriptions, and no minimum spend.

Event	Price	Description
Song scraped	$0.0015	One song with lyrics and metadata successfully extracted
Artist catalog scraped	$0.015	One artist's full discography listing extracted
Translation (BYOK)	$0.002	Lyrics translated using your own OpenAI API key
Translation (integrated)	$0.008	Lyrics translated using the built-in AI service

Cost examples

Scenario	Songs	Catalogs	Actor cost	Est. proxy cost	Total
Single song lookup	1	0	$0.002	~$0.01	~$0.01
Small artist	50	1	$0.09	~$0.05	~$0.14
Large artist	200	1	$0.32	~$0.20	~$0.52
10 artists	1,000	10	$1.65	~$0.75	~$2.40
Bulk corpus	10,000	100	$16.50	~$7.50	~$24.00
Large corpus	50,000	500	$82.50	~$40.00	~$122.50
1,000 songs + translation (BYOK)	1,000	10	$3.65	~$0.75	~$4.40
1,000 songs + translation (integrated)	1,000	10	$9.65	~$0.75	~$10.40

Proxy costs are estimates based on Apify residential proxy pricing and assume a moderate retry rate. Actual costs depend on blocking rates and geographic targeting.

Proxy Requirements

Residential proxies are required for reliable operation. AZLyrics employs Cloudflare protection and Google reCAPTCHA. Datacenter proxies will be blocked within a few requests.

When configuring the Actor, select Apify Proxy with the RESIDENTIAL group or provide your own residential proxy URLs. Budget approximately $0.50--1.00 in proxy costs per 1,000 songs scraped.

The Actor uses conservative request pacing (3--7 second random delays, maximum 12 requests per minute, 3 concurrent sessions) combined with session rotation, browser fingerprint generation, and automatic CAPTCHA detection with retry. These measures are effective only when paired with residential proxy IPs.

How It Works

1. URL resolution -- the Actor converts your input (artist name, song title, search query, or direct URLs) into AZLyrics page URLs. For name-based inputs, it constructs direct URLs from the normalized name. If the direct URL returns a 404 or empty page, it falls back to a site:azlyrics.com Google search.

2. Artist page processing -- when an artist page is found, the Actor extracts the full song catalog organized by album, pushes the catalog record to the dataset, and enqueues individual song URLs for lyrics extraction.

3. Lyrics extraction -- on each song page, the Actor waits for Cloudflare challenges to clear, then extracts lyrics from the DOM using comment-marker navigation (AZLyrics wraps lyrics between HTML comments rather than using semantic markup). The raw text is cleaned: HTML stripped, whitespace normalized, and line breaks preserved.

4. Translation -- if translateTo is set, each song's lyrics are sent to an OpenAI-compatible chat completions API for translation. The translated text is stored alongside the original.

5. Anti-bot handling -- the Actor uses PlaywrightCrawler with real Firefox to pass browser fingerprint checks. Sessions are rotated through a pool of 20, each limited to 15 requests or 5 minutes. CAPTCHA and block responses are detected automatically, causing the session to be discarded and the request retried with a fresh IP.

Use Cases

• NLP and text analysis -- build lyrics corpora for sentiment analysis, topic modeling, language studies, and text classification
• Music information retrieval -- structured lyrics data for MIR research pipelines and music recommendation systems
• Dataset building -- create training datasets for generative models, search indexes, or annotation projects
• Discography research -- catalog complete artist discographies with album groupings and song counts
• Translation datasets -- generate parallel lyrics corpora across languages for machine translation research
• Cultural analysis -- study lyrical trends, vocabulary evolution, and thematic patterns across genres, decades, or regions

Tips for Reliable Large Runs

• Use maxResults to break large jobs into manageable batches (e.g. 200 songs per run)
• Run during off-peak hours (UTC late night / early morning) for lower blocking rates
• Monitor the run log for CAPTCHA retry rates; if retries spike, pause and resume later
• For very large corpora (10,000+ songs across many artists), use the Apify scheduler to spread runs over multiple days

FAQ

Can I use datacenter proxies? You can try, but expect very high failure rates. AZLyrics actively blocks datacenter IP ranges. Residential proxies are strongly recommended.

How fast is the scraping? With default settings (3--7 second delays, 3 concurrent sessions, max 12 requests/minute), expect roughly 10--15 songs per minute. The Actor prioritizes reliability over speed.

What happens when a CAPTCHA is triggered? The Actor detects the CAPTCHA response, discards the current session (IP + cookies), and retries the request with a fresh session. Most CAPTCHA blocks resolve automatically within 1--2 retries.

Are the lyrics cleaned? Yes. The Actor extracts raw text content from the browser DOM, strips all HTML, trims whitespace, and normalizes line breaks. The output is clean plain text ready for NLP pipelines.

Can I scrape lyrics without artist catalog data? Yes. Provide individual song page URLs via startUrls. The Actor will scrape only those songs without generating catalog records.

What if an artist name does not match? The Actor first constructs a direct URL from the normalized artist name. If that page returns no results (404 or empty song list), it falls back to a Google site:azlyrics.com search to find the correct page.

Can I use a non-OpenAI translation provider? Yes. Set translateBaseUrl to any OpenAI-compatible API endpoint and translateModel to the model you want. For example, use Google Gemini by setting the base URL to https://generativelanguage.googleapis.com/v1beta/openai/ and the model to gemini-2.0-flash.

What is the difference between integrated and BYOK translation? With integrated translation ($0.008/song), the Actor uses a built-in OpenAI API key -- you do not need your own. With BYOK translation ($0.002/song), you provide your own API key and pay your LLM provider directly, so the Actor charges a lower fee.