archibate/dotfiles-opencode / skills/qwen-asr
qwen-asr
VerifiedTranscribe audio files using Qwen ASR. Use when the user sends voice messages and wants them converted to text.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 441 B |
| scripts/main.py | script | 3.3 KB |
scripts/main.py
View on GitHub# /// script
# requires-python = ">=3.11"
# dependencies = ["aiohttp", "argparse", "gradio_client"]
# ///
import os
import sys
import logging
import asyncio
import aiohttp
import argparse
from gradio_client import Client, handle_file
from contextlib import contextmanager, redirect_stdout
_LOGGER = logging.getLogger(__name__)
_LOGGER.setLevel(logging.INFO)
BASE_URL = os.getenv("QWEN_ASR_BASE_URL") or "https://qwen-qwen3-asr-demo.ms.show"
USER_AGENT = "Mozilla/5.0 AppleWebKit/537.36 Chrome/143 Safari/537"
SESSION: aiohttp.ClientSession | None = None
@contextmanager
def gradio_stdout():
with redirect_stdout(sys.stderr):
yield
async def transcribe(file, prompt="", lang="auto", itn=False):
audio_path = await upload_file(file)
if not audio_path:
exit(1)
audio_url = f"{BASE_URL}/gradio_api/file={audio_path}"
res = await SESSION.get(audio_url)
if res.status == 200:
_LOGGER.warning("Audio file: %s", audio_url)
else:
_LOGGER.warning("Audio file upload failed: %s", [audio_url, res.status, res.headers])
exit(1)
with gradio_stdout():
gradio = Client(BASE_URL)
result = gradio.predict(
audio_file=handle_file(audio_url),
context=prompt or "",
language=lang or "auto",
enable_itn=itn or False,
api_name="/asr_inference",
)
gradio.close()
_LOGGER.info("Gradio result: %s", result)
return result[0]
async def upload_file(audio):
file = None
if isinstance(audio, (bytes, bytearray)):
file = audio
elif audio in (None, "", "-"):
file = sys.stdin.buffer.read()
elif audio:
with open(audio, "rb") as f:
file = f.read()
if not file:
_LOGGER.warning("No file provided. %s", audio[:100])
return None
form = aiohttp.FormData()
form.add_field("files", file, filename="audio")
res = await api_request("/gradio_api/upload", data=form)
try:
src = (await res.json())[0]
if not src:
_LOGGER.warning("Upload failed: %s", [res.status, await res.text()])
except:
src = None
_LOGGER.error("Upload failed: %s", [res.status, await res.text()], exc_info=True)
return src
async def api_request(api, json=None, headers=None, **kwargs):
_LOGGER.info("%s: %s", api, json)
return await SESSION.post(
api,
json=json,
headers={
aiohttp.hdrs.USER_AGENT: USER_AGENT,
aiohttp.hdrs.REFERER: BASE_URL,
**(headers or {}),
},
**kwargs,
)
async def main():
global SESSION
async with aiohttp.ClientSession(base_url=BASE_URL) as SESSION:
parser = argparse.ArgumentParser()
parser.add_argument("-f", "--file", help="Audio file, - for stdio")
parser.add_argument("-p", "--prompt", help="Context")
parser.add_argument("--lang", help="Language: auto/zh/en/ja/ko/es/fr/de/ar/it/ru/pt")
parser.add_argument("--itn", help="Enable ITN", action="store_true")
args = parser.parse_args()
if args.file in (None, ""):
file = sys.stdin.buffer.read()
if file:
args.file = file
if args.file:
print(await transcribe(**vars(args)))
else:
parser.print_help()
if __name__ == "__main__":
asyncio.run(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: 197k | 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: 194k | 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: 194k | 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: 194k | Updated | Claude CodeCodex | Verified | Install |