TikTok Sound Scraper

Give it a TikTok sound and it returns the videos using that sound, one row per video.

You pass either the sound's numeric ID or a full /music/ link. The run walks that sound's feed and hands back each video with its counters, caption, author and file links. TikTok serves this feed to signed-out visitors, so the run doesn't use an account of yours.

What you get per video

Field	Notes
id	TikTok's video id.
webVideoUrl	https://www.tiktok.com/@handle/video/<id>.
text	The caption.
createTime	ISO timestamp.
hashtags	Parsed out of the caption.
region	Two-letter code, e.g. US.
isAd	Whether TikTok flags the video as an ad.
playCount, diggCount, commentCount, shareCount, collectCount, downloadCount	Views, likes, comments, shares, saves, downloads, as the feed reports them at run time.

Three nested objects:

• video — duration, cover, dynamicCover, playAddr (the MP4 without the watermark) and downloadAddr (the watermarked one).
• author — id, uniqueId, nickname, avatar.
• music — id, title, authorName, playUrl, original, duration. This one comes back null on videos where the feed doesn't attach music info, so don't key anything off it without a null check.

Input

{
  "musicIds": ["7654341603238103821"],
  "resultsPerPage": 200
}

musicUrls does the same job with full links, and the ID gets pulled out of the slug:

{
  "musicUrls": ["https://www.tiktok.com/music/original-sound-7654341603238103821"]
}

The ID is the trailing digits of any /music/ URL. You'll also find it in the music.id field of anything this scraper (or a TikTok video scraper) returns.

Field	Default	Notes
musicIds	—	Sound IDs. Several per run is fine.
musicUrls	—	Or /music/ links. Mix both if you like.
resultsPerPage	50	Cap on videos per sound, 1 to 2000.
sessionCookies	empty	Optional, see below.
proxyConfiguration	off	Optional. The scraper works without one.
notionConnector / notionParentId	—	Optional delivery into Notion via an Apify MCP connector.

Limits worth knowing

• 2,000 videos per sound is the ceiling the input box accepts. Quiet sounds run out well before that and the run just stops.
• You get the feed's order, which is roughly newest first. There's no sort option, because the source doesn't offer one.
• CDN links expire. playAddr, cover and avatar are signed URLs with an expiry a few days out. Download what you need in the same run or soon after; don't store them as permanent links.
• Counts are a snapshot at the moment of the run, not a history.
• A sound that returns nothing gives you one ok: false row with NO_RESULTS, BLOCKED or RATE_LIMITED rather than an empty dataset, so you can tell "no videos" apart from "something broke".

Your own cookie (optional)

Runs work with nothing from you. If you're tracking a lot of sounds and want a rate limit nobody else shares, paste your own TikTok sessionid into sessionCookies (Chrome: F12 → Application → Cookies → tiktok.com → sessionid), one per line. Several get rotated. It's only ever sent to TikTok and never written to the log. An expired cookie isn't fatal, the run just carries on without it.

Billing

$0.001 when a run starts, then $0.00025 per video returned, which is $0.25 per 1,000. Flat on every plan.

You pay for rows you got. Duplicates are dropped before the charge fires, and diagnostic rows (ok: false) are never charged, so a sound with no videos costs you the $0.001 start fee and nothing more.