sxhzju/sfx / skills/sfx-sources
sfx-sources
VerifiedFind and download free sound effects (音效/拟音/SFX) from Mixkit, Kenney, Freesound, Pixabay. Use for any request to find or add sound effects.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 1.3 KB |
| references/local-setup.md | markdown | 647 B |
| scripts/sfx.pyexecutable | script | 12.3 KB |
scripts/sfx.py
View on GitHub#!/usr/bin/env python3
"""Search & download free SFX from Mixkit / Kenney / Freesound / Pixabay.
Stdlib only, except Pixabay: it sits behind Cloudflare, so those calls re-exec under a private
venv with curl_cffi (Chrome TLS impersonation), created on first use — see ../references/local-setup.md.
sfx.py mixkit <keyword> [-n N]
sfx.py kenney [pack]
sfx.py freesound "<query>" [--license cc0|by|any] [--sort downloads|rating|relevance|short] [-n N] [-p PAGE]
sfx.py pixabay "<query>" [-n N] [-p PAGE] [--pages K] ranked by likes across K pages
sfx.py get <out_dir> <key>... keys: mk:<id> fs:<id> kn:<pack>/<file> px:<id>
"""
import argparse, html, json, os, re, shutil, subprocess, sys, time, urllib.parse, urllib.request, zipfile
from concurrent.futures import ThreadPoolExecutor
from pathlib import Path
UA = {"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) sfx-sources"}
CACHE = Path.home() / ".cache" / "sfx-sources"
INDEX = CACHE / "index.json" # search results, so `get` knows title + download url
VENV = CACHE / "venv" # curl_cffi for Pixabay
FS_LICENSE = {"cc0": '"Creative Commons 0"', "by": '("Attribution" OR "Creative Commons 0")', "any": None}
FS_LIC_SHORT = {"Creative Commons 0": "CC0", "Attribution": "BY", "Attribution NonCommercial": "BY-NC", "Sampling+": "Sampling+"}
FS_SORT = {"downloads": "Downloads (most first)", "rating": "Rating (highest first)",
"relevance": "Automatic by relevance", "short": "Duration (shortest first)"}
def fetch(url, timeout=60, tries=3):
for n in range(tries):
try:
if "pixabay.com" in url: # Cloudflare 403s plain urllib
from curl_cffi import requests as cr
r = cr.get(url, impersonate="chrome", timeout=timeout, headers={"Referer": "https://pixabay.com/"})
if r.status_code != 200:
raise OSError(f"HTTP {r.status_code} {url}")
return r.content
with urllib.request.urlopen(urllib.request.Request(url, headers=UA), timeout=timeout) as r:
return r.read()
except Exception as e:
if n == tries - 1:
raise e if isinstance(e, OSError) else OSError(repr(e))
time.sleep(1 + n) # Cloudflare 403s are sporadic; a short pause usually clears them
def need_cffi():
"""Re-exec this script under the private venv unless curl_cffi is already importable."""
try:
import curl_cffi # noqa: F401
return
except ImportError:
pass
py = VENV / "bin" / "python"
if not py.exists():
print("setting up curl_cffi venv for Pixabay (one-time)…", file=sys.stderr)
subprocess.run([sys.executable, "-m", "venv", str(VENV)], check=True)
subprocess.run([str(py), "-m", "pip", "-q", "install", "curl_cffi"], check=True)
os.execv(str(py), [str(py), os.path.abspath(__file__), *sys.argv[1:]])
def load_index():
try:
return json.loads(INDEX.read_text())
except (OSError, ValueError):
return {}
def save_index(rows):
CACHE.mkdir(parents=True, exist_ok=True)
idx = load_index()
idx.update({r["key"]: r for r in rows})
INDEX.write_text(json.dumps(idx, ensure_ascii=False))
def show(rows, a, lic=False, hint=""):
save_index(rows)
for r in rows:
print("\t".join([r["key"], r["title"], r["dur"]] + ([r["stats"]] if "stats" in r else []) + [r["license"]] * lic + [r["preview"]] * a.urls))
if not rows:
print("(none)" + hint)
def mixkit(a):
kw = re.sub(r"\s+", "-", a.keyword.strip().lower())
h = fetch(f"https://mixkit.co/free-sound-effects/discover/{urllib.parse.quote(kw)}/").decode()
rows = []
for blk in h.split('data-audio-player-item-id-value="')[1:]:
i = blk.split('"', 1)[0]
t = re.search(r'item-grid-card__title">\s*(.*?)\s*</h2>', blk, re.S)
d = re.search(r'data-test-id="duration">\s*([\d:]+)', blk)
rows.append({"key": f"mk:{i}", "title": html.unescape(t.group(1)) if t else "", "dur": d.group(1) if d else "",
"license": "Mixkit", "preview": f"https://assets.mixkit.co/active_storage/sfx/{i}/{i}-preview.mp3",
"url": f"https://assets.mixkit.co/active_storage/sfx/{i}/{i}.wav"})
show(rows[: a.n], a, hint=" try a broader tag, or freesound")
def freesound(a):
q = {"q": a.query, "s": FS_SORT[a.sort], "page": a.page}
if FS_LICENSE[a.license]:
q["f"] = "license:" + FS_LICENSE[a.license]
h = fetch("https://freesound.org/search/?" + urllib.parse.urlencode(q)).decode()
rows = []
for blk in h.split('class="bw-search__result"')[1:]:
g = lambda k: (re.search(rf'data-{k}="([^"]*)"', blk) or [None, ""])[1]
lic = re.search(r'title="License: ([^"]+)"', blk)
sid, user = g("sound-id"), g("username")
rows.append({"key": f"fs:{sid}", "title": html.unescape(g("title")), "dur": f"{float(g('duration') or 0):.1f}s",
"author": user, "license": FS_LIC_SHORT.get(lic.group(1), lic.group(1)) if lic else "?", "preview": g("mp3").replace("-lq.mp3", "-hq.mp3"),
"url": g("mp3").replace("-lq.mp3", "-hq.mp3")})
show(rows[: a.n], a, lic=a.license != "cc0")
def k_num(t): # "2.4K" -> 2400
m = re.match(r"([\d.]+)\s*([KM]?)", t or "")
return int(float(m[1]) * {"": 1, "K": 1e3, "M": 1e6}[m[2]]) if m else 0
def pixabay(a):
# Pixabay can't sort by likes server-side, so pull several relevance pages, rank their cards by
# the rounded like counter, then open only the top n detail pages for exact numbers + mp3.
q = urllib.parse.quote(a.query.strip())
pages = range(a.page, a.page + a.pages)
with ThreadPoolExecutor(len(pages)) as ex:
htmls = list(ex.map(lambda n: fetch(f"https://pixabay.com/sound-effects/search/{q}/" + (f"?pagi={n}" if n > 1 else "")).decode(), pages))
seen, cards = set(), []
for h in htmls:
for blk in h.split('<div class="nameAndTitle')[1:]:
t = re.search(r'<a href="(/sound-effects/[^"]*?-(\d+)/)" class="title[^"]*">(.*?)</a><a href="/users/[^"]*" class="name[^"]*">(.*?)</a>', blk)
if t and t[2] not in seen:
seen.add(t[2])
cards.append((t[1], t[2], html.unescape(t[3]), html.unescape(t[4]), k_num((re.search(r'counter="([^"]*)"', blk) or [0, ""])[1])))
cards.sort(key=lambda c: -c[4])
def detail(c): # duration, exact likes/downloads and mp3 url only live in the detail page's JSON-LD
path, sid, title, user, likes = c
dl, dur, url = 0, "?", ""
try:
d = fetch("https://pixabay.com" + path).decode()
j = next(x for x in map(json.loads, re.findall(r'<script type="application/ld\+json">(.*?)</script>', d, re.S))
if x.get("@type") == "AudioObject")
st = {i["interactionType"].rsplit("/", 1)[-1]: i["userInteractionCount"] for i in j.get("interactionStatistic", [])}
likes, dl = st.get("LikeAction", likes), st.get("DownloadAction", 0)
m = re.match(r"PT(?:(\d+)M)?(\d+)S", j.get("duration", ""))
dur, url = f"{int(m[1] or 0) * 60 + int(m[2])}s" if m else "", j["contentUrl"]
except Exception:
pass
return {"key": f"px:{sid}", "title": title, "dur": dur, "author": user, "license": "Pixabay", "likes": likes,
"stats": f"♥{likes} ↓{dl}", "page": "https://pixabay.com" + path, "preview": url, "url": url}
with ThreadPoolExecutor(6) as ex:
rows = [r for r in ex.map(detail, cards[: a.n]) if r["url"]]
rows.sort(key=lambda r: -r["likes"])
show(rows, a, hint=" try a broader English query")
def kenney(a):
if not a.pack:
h = fetch("https://kenney.nl/assets/category:Audio").decode()
packs = sorted({p for p in re.findall(r"kenney\.nl/assets/([a-z0-9-]+)['\"]", h)})
print(" ".join(packs))
return
d = CACHE / "kenney" / a.pack
if not d.exists():
h = fetch(f"https://kenney.nl/assets/{a.pack}").decode()
z = re.search(r"https://kenney\.nl/media/[^\"']+\.zip", h)
if not z:
sys.exit(f"no zip found for pack '{a.pack}' (run `kenney` to list packs)")
zp = d.with_suffix(".zip")
zp.parent.mkdir(parents=True, exist_ok=True)
zp.write_bytes(fetch(z.group(0), timeout=300))
zipfile.ZipFile(zp).extractall(d)
files = sorted(p.relative_to(d).as_posix() for p in d.rglob("*") if p.suffix.lower() in (".ogg", ".wav", ".mp3"))
groups = {}
for f in files: # collapse click_001..click_005 into one line to keep output short
m = re.match(r"(.*?)(\d+)(\.\w+)$", f)
groups.setdefault((m[1], m[3]) if m else (f, ""), []).append(m[2] if m else "")
top = {f.split("/")[0] for f in files}
pre = next(iter(top)) + "/" if len(top) == 1 and all("/" in f for f in files) else ""
exts = {e for _, e in groups}
ext = exts.pop() if len(exts) == 1 else "" # print a shared extension once
print(f"key kn:{a.pack}/{pre}<name>{ext}")
print(" ".join(stem[len(pre):] + (f"{{{n[0]}..{n[-1]}}}" if len(n) > 1 else n[0]) + (e if not ext else "")
for (stem, e), n in groups.items()))
def get(a):
out = Path(a.out_dir)
out.mkdir(parents=True, exist_ok=True)
idx = load_index()
def one(key):
src, _, ref = key.partition(":")
if src == "kn":
p = CACHE / "kenney" / ref
if not p.exists():
return key, None, "not cached — run `kenney <pack>` first"
dst = out / f"kn_{ref.replace('/', '_')}"
shutil.copy(p, dst)
return key, dst, None
r = idx.get(key)
if not r and src == "px":
return key, None, "unknown key — run `pixabay <query>` first"
if not r and src == "mk":
r = {"title": "", "url": f"https://assets.mixkit.co/active_storage/sfx/{ref}/{ref}.wav"}
if not r:
return key, None, "unknown key — search first"
slug = re.sub(r"[^\w]+", "-", r["title"]).strip("-")[:40]
dst = out / f"{src}_{ref}{'_' + slug if slug else ''}{Path(urllib.parse.urlparse(r['url']).path).suffix}"
if not dst.exists():
tmp = dst.with_suffix(dst.suffix + ".part")
tmp.write_bytes(fetch(r["url"], timeout=300))
tmp.rename(dst)
lic = r.get("license", "CC0")
return key, dst, f"credit {r.get('author', '?')}" if src == "fs" and lic != "CC0" else None
def safe(key):
try:
return one(key)
except Exception as e: # one bad file shouldn't sink the batch
return key, None, repr(e)
# Freesound caps ~80 KB/s per connection, so parallelism is the only speed lever.
with ThreadPoolExecutor(6) as ex:
for key, dst, note in ex.map(safe, a.keys):
print(f"FAIL {key}: {note}" if not dst else f"{dst.name} [{note}]" if note else dst.name)
def main():
ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
sp = ap.add_subparsers(dest="cmd", required=True)
m = sp.add_parser("mixkit"); m.add_argument("keyword"); m.add_argument("-n", type=int, default=10); m.add_argument("--urls", action="store_true"); m.set_defaults(fn=mixkit)
k = sp.add_parser("kenney"); k.add_argument("pack", nargs="?"); k.set_defaults(fn=kenney)
f = sp.add_parser("freesound"); f.add_argument("query"); f.add_argument("--license", choices=FS_LICENSE, default="any")
f.add_argument("--sort", choices=FS_SORT, default="downloads"); f.add_argument("-n", type=int, default=8); f.add_argument("--urls", action="store_true")
f.add_argument("-p", "--page", type=int, default=1); f.set_defaults(fn=freesound)
x = sp.add_parser("pixabay"); x.add_argument("query"); x.add_argument("-n", type=int, default=8); x.add_argument("--urls", action="store_true")
x.add_argument("-p", "--page", type=int, default=1)
x.add_argument("--pages", type=int, default=3, help="relevance pages (20 each) to rank by likes"); x.set_defaults(fn=pixabay)
g = sp.add_parser("get"); g.add_argument("out_dir"); g.add_argument("keys", nargs="+"); g.set_defaults(fn=get)
a = ap.parse_args()
if a.cmd == "pixabay" or (a.cmd == "get" and any(k.startswith("px:") for k in a.keys)):
need_cffi()
try:
a.fn(a)
except OSError as e:
sys.exit(f"network error: {getattr(e, 'reason', e)}")
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |