YouTube Audio Extractor

Extract audio and timed transcripts from YouTube and YouTube Music — MP3, M4A, WebM, FLAC, WAV, Ogg Vorbis or ALAC — as public, no-login download links, at a flat $0.006 per track with the first 50 MB of every track included. M4A and WebM are copied straight from YouTube's native audio stream (no re-encode, ~5–10x faster than converter actors), every track comes with a permanent public URL, and captions are fetched in the same run.

Built for podcasters, researchers building audio datasets, archivists, music lovers and AI projects that need clean audio files.

Main Use Cases

• YouTube to MP3 — 192 kbps default (96–320 selectable) for playback anywhere
• YouTube Music audio download — public music tracks in M4A/WebM kept losslessly native, or MP3/FLAC/WAV
• Audio datasets for AI — batch-download thousands of tracks with metadata (views, likes, captions, duration, published date) alongside the files
• Podcast and interview extraction — pull episode audio from YouTube or YouTube Music with optional section trimming
• YouTube captions and transcripts — same run, one language, billed only when that language actually exists
• Archival in lossless formats — FLAC, WAV or ALAC for permanent collections, metadata included

How It Works — Native Stream Copy, Not Re-Encoding

YouTube and YouTube Music serve audio as progressive download streams; the actor fetches that stream over plain HTTP — no browser, no JavaScript, no login — and saves it as M4A or WebM without touching a single sample (that is why M4A/WebM is the "fastest" format option). Only when you ask for MP3, FLAC, WAV, Ogg Vorbis or ALAC does the actor re-encode the saved source with FFmpeg (MP3 at 192 kbps by default). Failures are per-track and independent: one bad link never stops the rest of the run, and a failed download is never charged for the track itself — only the metered proxy bandwidth it actually consumed, if any, is billed.

Download YouTube Audio Without a YouTube Audio API

YouTube has no official audio-download API, so converter actors typically run a headless browser and pay per video minute — which is why the segment-downloader competitor bills $0.05/video plus $0.04/min. This actor talks directly to the stream URLs YouTube's own pages use, so a typical 3–5 minute track costs a flat $0.006 including the file (50 MB included; extra megabytes $0.001/MB) and finishes in seconds, not minutes. Transcripts ride along at $0.0012 each instead of a separate paid actor, and every output file is a public link you can hand to anyone — no Apify login or token required.

Input

Field	Type	Required	Default	Notes
urls	array	Yes	—	YouTube Music tracks, YouTube videos, Shorts, live, embed or youtu.be links, or bare 11-character video IDs; each processed independently
format	enum	No	m4a	m4a/webm = native stream copy (fastest); mp3 = 192 kbps re-encode; flac/wav/vorbis/alac = lossless or open re-encode
country	string	No	US	Two-letter ISO code — fetch public audio offered in that market (region restrictions)
captionLanguage	string	No	""	Captions/timed transcript language, e.g. en or en-US; empty = skip
proxyConfiguration	object	No	off	Leave off: direct download works and is cheapest; the actor falls back to metered residential proxy on block
maxConcurrency	integer	No	4	Tracks processed in parallel (1–20; >10 rarely helps)
maxDurationSeconds	integer	No	14400	Safety cap (4 h) — longer tracks rejected before any download
retries	integer	No	3	Failed-track retries, each with rotated proxy + longer backoff (0–8)
mp3Bitrate	integer	No	192	MP3-only target (96–320); other formats ignore it
startTimeSeconds	integer	No	0	Skip this many seconds and save only what follows (0 = whole track)
endTimeSeconds	integer	No	0	Stop saving here (90 = first 90 seconds); 0 = to the end

Output — Audio Tracks With Complete Metadata

One dataset row per input URL (the overview view shows the essentials; the KVS audio collection holds the binary files, key prefix audio-):

Field	Description
videoId, title, artistName	Track identity (artist = channel for videos, artist for music)
submittedUrl, sourceUrl	As-submitted vs resolved canonical URL
audioFormat, audioCodec, bitrateKbps, sampleRateHz	What you actually got (native stream or re-encode)
durationSeconds, audioDurationSeconds	Source duration vs length of the saved audio (differs when trimmed)
fileSizeBytes, sizeMb, billedMb, sourceFileSizeBytes	Size; billedMb = MB charged above the 50 included
proxyMb	Metered proxy bandwidth (0 for direct/self-proxy runs)
fileName, downloadUrl	Public, no-login link to the audio file
thumbnailUrl, publishedDate, viewCount, likeCount, commentCount, isLive	Content metadata
transcoded, requestedBitrateKbps, section	Encoding provenance: was it re-encoded, target kbps, requested start/end
accessCountry, captions	Market used; captions object (null when none requested/exist)
error, errorMessage	Per-track failure codes (track itself never charged; proxyMb shows any metered bandwidth billed)

Example Input

{
  "urls": [
    "https://music.youtube.com/watch?v=jNQXAC9IVRw",
    "https://www.youtube.com/watch?v=aqz-KE-bpKQ",
    "youtu.be/9bZkp7q19f0"
  ],
  "format": "mp3",
  "captionLanguage": "en",
  "maxConcurrency": 4
}

Trim a section and keep it lossless: "format": "flac", "startTimeSeconds": 90, "endTimeSeconds": 300.

Example Output — Track Record (trimmed)

{
  "videoId": "aqz-KE-bpKQ",
  "title": "Big Buck Bunny 60fps 4K - Official Blender Foundation Short Film",
  "artistName": "Blender",
  "audioFormat": "mp3",
  "audioCodec": "mp3",
  "bitrateKbps": 192,
  "sampleRateHz": 44100,
  "durationSeconds": 635,
  "audioDurationSeconds": 635,
  "fileSizeBytes": 15420416,
  "sizeMb": 15,
  "billedMb": 0,
  "proxyMb": 0,
  "transcoded": true,
  "requestedBitrateKbps": 192,
  "section": null,
  "fileName": "audio-7f8d1a2b-3c4d-5e6f-9a8b-0c1d2e3f4a5b.mp3",
  "downloadUrl": "https://api.apify.com/v2/key-value-stores/.../records/audio-7f8d1a2b....mp3?disableRedirect=1",
  "thumbnailUrl": "https://i.ytimg.com/vi/aqz-KE-bpKQ/...",
  "publishedDate": "2015-01-14",
  "viewCount": 11762113,
  "likeCount": 158274,
  "commentCount": 10212,
  "isLive": false,
  "accessCountry": "US",
  "captions": { "language": "en", "transcript": "a line of timed caption text" },
  "error": null,
  "errorMessage": null
}

YouTube Audio at $0.006 Per Track

Billing is strictly per completed track (from .actor/pay-per-event.json, all real):

Event	Price	Billed when
Audio track	$0.006	Track saved as a downloadable file — first 50 MB included; a failed track is never charged for the track itself
Extra saved MB	$0.001	Per MB above the 50 MB included (rounded up)
Transcript/subtitle	$0.0012	Only when the track actually has captions in the requested language
Metered proxy MB	$0.02	Only on a block, when the actor falls back to residential proxy, and includes bandwidth spent on a retry that still failed (never for direct runs)

A 4-minute 128 kbps M4A ≈ 4 MB ≈ $0.006 flat. A 100 MB FLAC ≈ $0.006 + $0.05 = $0.056. A link that fails costs no track fee; the metered proxy bandwidth it spent, if any, still shows up as proxyMb. The flat per-track price makes volume predictable: 1,000 regular tracks = $6.00.

Integrations & Automation

• API — run and fetch via the Apify REST API, or call it from AI agents through the Apify MCP server
• Storage — audio files land in the actor's key-value store (audio collection, public links); dataset rows carry metadata + downloadUrl
• Scheduling — periodic sweeps for channels you track (new uploads → audio + transcript)
• Webhooks — notify your pipeline each time a track finishes; export dataset as JSON/CSV/Excel

Recommended schedule: on-demand for one-offs; daily/weekly sweeps for archival or dataset builds. The maxDurationSeconds safety cap guarantees a mistaken link can never produce a huge bill.

Related Actors

• YouTube Audio Segment Downloader — entertained_rattlesnake/youtube-audio-segment-downloader — most-used audio actor (114 users, 39 monthly); $0.05/video + $0.04/min
• YouTube Link — utils/youtube-link — link extractor with audio option, $0.013/track (104 users, 5.0★)
• YouTube Video Audio Downloader — apple_yang/youtube-video-audio-downloader — $0.0025/result (69 users, 5.0★)
• YouTube Video Audio Downloader Reliable — vidkraken/youtube-video-audio-downloader-reliable — per-MB billing ($0.002/MB; a 50 MB track ≈ $0.10)
• YouTube Audio Downloader Lite — alpha-scraper/youtube-audio-downloader-lite — $0.0025/result

This actor is the only one with 7 formats, bundled transcripts, section trimming, and flat per-track pricing — the per-MB competitor charges ~$0.10 for a 50 MB track where this costs $0.006 flat.

FAQ

Why use this actor instead of an official YouTube audio API?

There is no official audio-download API. YouTube serves progressive audio streams that any client with the right request shape can consume; this actor reads them directly over HTTP — no browser, no API key, no token.

What are alternatives to this actor for YouTube audio?

On the Store: entertained_rattlesnake/youtube-audio-segment-downloader ($0.05/video + $0.04/min), utils/youtube-link ($0.013/track, link extractor), vidkraken/youtube-video-audio-downloader-reliable (per-MB). Ours is the only flat per-track actor with transcripts + trimming + 7 formats.

MP3 320 kbps vs 192 — which should I pick?

192 kbps (default) is the widely accepted quality/size balance for MP3. Pick 320 kbps for archival on the MP3 codec, or — better — keep the native stream (m4a) or FLAC/WAV/ALAC for true lossless. mp3Bitrate only applies when format: "mp3".

M4A vs MP3 — why is M4A the fastest?

M4A and WebM are YouTube's native audio formats, so the actor downloads the stream and saves it as-is — no re-encoding, no FFmpeg pass. MP3/FLAC/WAV/Vorbis/ALAC require a conversion pass over the saved file. For identical quality, M4A usually beats MP3 at the same bitrate anyway.

Why am I being charged proxy MB?

Only when a direct download was blocked and the actor had to retry through Apify residential proxy. Direct runs (the default) never see proxy charges. If you're being blocked often, set country to the market where the track is public, or force a proxy — never to avoid the fee, but to get through.

How do I get captions/transcripts?

Set captionLanguage (e.g. en, en-US). The dataset row's captions object carries the transcript; it is null when no captions exist in that language — and you are only charged the $0.0012 transcript fee when captions are actually returned.

What about region-blocked or very long tracks?

country fetches the public audio offered in that market. maxDurationSeconds (default 14 400 = 4 h) rejects anything longer before downloading a byte — a mistaken link can never run up a bill. startTimeSeconds/endTimeSeconds let you save just a section (great for podcasts).

Where are the files?

In the run's key-value store, audio collection (keys audio-…). Each dataset row's downloadUrl is a public link — anyone can download without an Apify account. The dataset rows themselves are the index, with full metadata.

For AI Agents & LLM Apps

• Purpose: given YouTube/YouTube Music URLs, returns one dataset row per track — format/codec/bitrate/sample rate/size, duration, metadata (views, likes, captions), the public downloadUrl — and stores the audio file itself in the audio KVS collection.
• Minimal working input:

{ "urls": ["https://music.youtube.com/watch?v=jNQXAC9IVRw"] }

• Variant inputs — MP3 + captions: { "urls": ["..."], "format": "mp3", "captionLanguage": "en" }; trimmed lossless: { "urls": ["..."], "format": "flac", "startTimeSeconds": 90, "endTimeSeconds": 300 }; batch: any number of URLs, "maxConcurrency": 8.
• Output fields: videoId, title, artistName, submittedUrl, sourceUrl, durationSeconds, audioFormat, audioCodec, bitrateKbps, sampleRateHz, fileSizeBytes, sizeMb, audioDurationSeconds, requestedBitrateKbps, section, billedMb, sourceFileSizeBytes, proxyMb, fileName, downloadUrl, thumbnailUrl, publishedDate, viewCount, likeCount, commentCount, isLive, transcoded, accessCountry, captions, error, errorMessage.
• Behaviors an agent should know: urls is required and accepts full URLs or bare 11-char IDs; each URL is independent — failures land in error/errorMessage and are never charged for the track itself (any metered proxy bandwidth is billed as proxyMb), so keep failed rows out of downstream billing. Audio files live in the run's KVS audio collection (key prefix audio-); downloadUrl needs no auth but is ephemeral like any Apify key-value record — persist the bytes (or re-run) for long-term use. M4A/WebM = native stream, transcoded: false; MP3/FLAC/WAV/Vorbis/ALAC = re-encoded (transcoded: true), requestedBitrateKbps applies to MP3 only. The safety cap maxDurationSeconds (default 14 400) rejects longer tracks before download — raise it consciously for long podcasts. billedMb = MB over the 50 included; proxyMb nonzero only when metered Apify residential was used. Never request a caption language you don't need: the $0.0012 transcript fee applies per returned transcript.
• Billing: pay-per-event — $0.006 per saved track (first 50 MB included; a failed track is not charged for the track itself) + $0.001 per extra MB + $0.0012 per transcript + $0.02 per metered proxy MB (only when proxy is used, failures included) + standard $0.00005 actor start.

Legal & Compliance Disclaimer

This actor is an independent tool and is not affiliated with, endorsed by, or sponsored by Google LLC, YouTube, or YouTube Music. It accesses only publicly available audio streams served to anonymous visitors of public watch pages — no login, no bypass of paywalls or DRM, and no extraction of private content. Downloading audio may violate YouTube's Terms of Service; the actor is intended for legitimate uses such as archiving content you have rights to, research, and personal use. Users are responsible for their own compliance with YouTube's Terms of Service, applicable copyright law, and data-protection rules — do not redistribute copyrighted audio without permission.

SEO Keywords

youtube audio extractor, youtube to mp3, youtube music audio download, youtube audio api alternative, download youtube audio m4a, youtube audio flac, youtube audio wav, youtube video to audio, youtubetranscript api, youtube caption download, youtube audio metadata, audio dataset for ai, youtube shorts audio, podcast audio extraction, youtube audio trimming, youtube lyrics captions, region locked youtube audio, youtube audio archival, public audio download link, youtube mp3 192 kbps