Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

Prepare ModelScope model card sync (#3383)

zhifu gaodd0b71b3fc43
diff --git a/docs/community_growth_20k.md b/docs/community_growth_20k.md
index 84363b13d..78dceefaa 100644
--- a/docs/community_growth_20k.md
+++ b/docs/community_growth_20k.md
@@ -361,7 +361,7 @@ Paste the Markdown output into launch notes, weekly community updates, or releas
- [x] Keep recent Python GitHub release pages linked to runtime assets. On 2026-07-23, `v1.3.19` through `v1.3.26` were verified to include a `Runtime downloads` section pointing to `runtime-llamacpp-v0.1.8`, including Linux Vulkan, Windows CUDA, all eight prebuilt llama.cpp / GGUF assets, and the matching PyPI install command. The `scripts/update_release_runtime_downloads.py` guard and `tests/test_release_runtime_downloads.py` cover future tag releases.
- [x] Update PyPI release description if needed. On 2026-07-23, the live PyPI `funasr 1.3.26` page was audited: its long description already comes from `README.md` and includes the current deployment-routing markers `OpenAI-compatible`, `/v1/audio/transcriptions`, `vllm`, `runtime-llamacpp-v0.1.8`, `GGUF`, and `funasr==1.3.26`. The package summary was prepared for the next PyPI publish by updating `setup.py` to `OpenAI-compatible speech recognition toolkit with WebSocket streaming, vLLM acceleration, and llama.cpp/GGUF edge runtime`, plus discovery keywords for `openai-compatible`, `websocket`, `vllm`, `gguf`, and `llama-cpp`; the currently published PyPI metadata remains `1.3.26` until the next weekly release upload.
- [ ] Update Hugging Face organization/model cards. The `FunAudioLLM/Fun-ASR-MLT-Nano-2512` README update is complete: HF PR [#5](https://huggingface.co/FunAudioLLM/Fun-ASR-MLT-Nano-2512/discussions/5) merged commit `3dd802bfb8a05dadc440c9d9fc84ecfc1e15c0ff`, and a 2026-07-23 live Hub download confirmed `main` now includes `python -m pip install -U "funasr>=1.3.26"` before optional demo-repo setup. Prepared README patches for `FunAudioLLM/Fun-ASR-Nano-2512` and `FunAudioLLM/SenseVoiceSmall` would add the same `funasr>=1.3.26` install floor, but the current `langgz` token still cannot write those model cards: direct upload fails with `403 Forbidden: pass create_pr=1`, and the Hub's suggested `--create-pr` path also fails at `preupload/main?create_pr=1` with `403 Forbidden: Authorization error`. Retry after model-card preupload/write permission is granted on those two repos.
-- [ ] Update ModelScope model cards and demos.
+- [ ] Update ModelScope model cards and demos. On 2026-07-23, `scripts/sync_modelscope_model_cards.py` was added to generate and optionally upload current ModelScope README updates for `FunAudioLLM/Fun-ASR-Nano-2512` (`README.md`, `README_zh.md`) and `iic/SenseVoiceSmall` (`README.md`). The prepared cards add `python -m pip install -U "funasr>=1.3.26" modelscope` and `torch.cuda.is_available()` CUDA-wheel verification guidance, matching the clean H100 quickstart evidence. Prepared files are under `/cpfs_speech/user/zhifu.gzf/.cache/funasr-ops/modelscope-card-sync-20260723/`. A live upload attempt with the existing 32-byte `.modelscope/credentials/session` failed: SDK uploads returned HTTP 400 from the ModelScope LFS batch API, and CLI upload reported `No credentials found`; the remaining blocker is a real ModelScope access token from `https://modelscope.cn/my/myaccesstoken`. Do not mark complete until those three ModelScope README files are uploaded and re-downloaded from the live model cards.
- [x] Update homepage hero and docs entry points. On 2026-07-23, the live homepage entry points were audited and added to the public website contract. [`https://www.funasr.com/`](https://www.funasr.com/) visibly leads with the Chinese industrial speech-recognition hero, links `/donors.html`, and exposes the OpenAI-compatible `/v1/audio/transcriptions` and `vLLM` deployment path. [`https://www.funasr.com/en/`](https://www.funasr.com/en/) exposes `Industrial Speech Recognition`, `OpenAI-compatible`, `/v1/audio/transcriptions`, and `vLLM`. `scripts/check_funasr_website_static.py` now patrols both homepages plus the ecosystem, donor, CLI tutorial, llama.cpp landing, and llama.cpp blog pages; the live check prints `funasr.com static page contract passed for 11 pages`.
### Week 4: Distribution and feedback
diff --git a/scripts/sync_modelscope_model_cards.py b/scripts/sync_modelscope_model_cards.py
new file mode 100755
index 000000000..a7e216ac0
--- /dev/null
+++ b/scripts/sync_modelscope_model_cards.py
@@ -0,0 +1,221 @@
+#!/usr/bin/env python3
+"""Prepare and optionally upload current ModelScope model-card guidance.
+
+The script keeps high-traffic ModelScope cards aligned with the FunASR GitHub
+README and PyPI release guidance. By default it writes patched README files to a
+local output directory. Use --upload with a ModelScope access token to publish
+only those README files back to their model repositories.
+"""
+
+from __future__ import annotations
+
+import argparse
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Iterable
+
+
+DEFAULT_SOURCE_ROOT = Path(
+ "/cpfs_speech/user/zhifu.gzf/.cache/funasr-ops/modelscope-card-audit-20260723/models"
+)
+DEFAULT_OUTPUT_ROOT = Path(
+ "/cpfs_speech/user/zhifu.gzf/.cache/funasr-ops/modelscope-card-sync-20260723"
+)
+
+
+@dataclass(frozen=True)
+class CardPatch:
+ repo_id: str
+ source: Path
+ path_in_repo: str
+ replacements: tuple[tuple[str, str], ...]
+ markers: tuple[str, ...]
+
+ def output_path(self, output_root: Path) -> Path:
+ return output_root / self.repo_id.replace("/", "__") / self.path_in_repo
+
+
+def _patches(source_root: Path) -> tuple[CardPatch, ...]:
+ nano_en_env_old = """# Environment Setup 🐍
+
+```shell
+git clone https://github.com/FunAudioLLM/Fun-ASR.git
+cd Fun-ASR
+pip install -r requirements.txt
+```"""
+ nano_en_env_new = """# Environment Setup 🐍
+
+Use FunASR v1.3.26 or newer for the current ModelScope hub path, OpenAI-compatible server, and vLLM fallback fixes:
+
+```shell
+python -m pip install -U "funasr>=1.3.26" modelscope
+```
+
+For GPU inference, install the PyTorch / torchaudio CUDA wheels that match your driver from [pytorch.org](https://pytorch.org/get-started/locally/) and verify CUDA before using `device="cuda"`:
+
+```python
+import torch
+print(torch.__version__, torch.version.cuda, torch.cuda.is_available())
+```
+
+Optional demo repository setup:
+
+```shell
+git clone https://github.com/FunAudioLLM/Fun-ASR.git
+cd Fun-ASR
+pip install -r requirements.txt
+```"""
+ nano_zh_env_old = """# 环境安装 🐍
+
+```shell
+git clone https://github.com/FunAudioLLM/Fun-ASR.git
+cd Fun-ASR
+pip install -r requirements.txt
+```"""
+ nano_zh_env_new = """# 环境安装 🐍
+
+推荐使用 FunASR v1.3.26 或更新版本,以获得当前 ModelScope hub 路径、OpenAI 兼容服务和 vLLM fallback 修复:
+
+```shell
+python -m pip install -U "funasr>=1.3.26" modelscope
+```
+
+GPU 推理前,请按 [PyTorch 官网](https://pytorch.org/get-started/locally/)安装与你驱动匹配的 PyTorch / torchaudio CUDA wheel,并先确认 CUDA 可见:
+
+```python
+import torch
+print(torch.__version__, torch.version.cuda, torch.cuda.is_available())
+```
+
+如需运行 demo 仓库,可再执行:
+
+```shell
+git clone https://github.com/FunAudioLLM/Fun-ASR.git
+cd Fun-ASR
+pip install -r requirements.txt
+```"""
+ sensevoice_env_old = """推理之前,请务必更新funasr与modelscope版本
+
+```shell
+pip install -U funasr modelscope
+```"""
+ sensevoice_env_new = """推理之前,请务必更新 funasr 与 modelscope 版本。推荐使用 FunASR v1.3.26 或更新版本,以获得当前命令行、OpenAI 兼容服务、字幕输出和 ModelScope hub 路径修复:
+
+```shell
+python -m pip install -U "funasr>=1.3.26" modelscope
+```
+
+GPU 推理前,请按 [PyTorch 官网](https://pytorch.org/get-started/locally/)安装与你驱动匹配的 PyTorch / torchaudio CUDA wheel,并先确认 CUDA 可见:
+
+```python
+import torch
+print(torch.__version__, torch.version.cuda, torch.cuda.is_available())
+```"""
+ return (
+ CardPatch(
+ repo_id="FunAudioLLM/Fun-ASR-Nano-2512",
+ source=source_root / "FunAudioLLM--Fun-ASR-Nano-2512/snapshots/master/README.md",
+ path_in_repo="README.md",
+ replacements=(
+ (nano_en_env_old, nano_en_env_new),
+ (
+ "## Quick start via ModelScope\n```shell\npip install -U funasr modelscope\n```",
+ "## Quick start via ModelScope\n```shell\npython -m pip install -U \"funasr>=1.3.26\" modelscope\n```",
+ ),
+ ),
+ markers=("funasr>=1.3.26", "torch.cuda.is_available()", "Quick start via ModelScope"),
+ ),
+ CardPatch(
+ repo_id="FunAudioLLM/Fun-ASR-Nano-2512",
+ source=source_root / "FunAudioLLM--Fun-ASR-Nano-2512/snapshots/master/README_zh.md",
+ path_in_repo="README_zh.md",
+ replacements=((nano_zh_env_old, nano_zh_env_new),),
+ markers=("funasr>=1.3.26", "torch.cuda.is_available()", "环境安装"),
+ ),
+ CardPatch(
+ repo_id="iic/SenseVoiceSmall",
+ source=source_root / "iic--SenseVoiceSmall/snapshots/master/README.md",
+ path_in_repo="README.md",
+ replacements=(
+ (sensevoice_env_old, sensevoice_env_new),
+ (
+ "#安装ModelScope\npip install modelscope",
+ "# 安装 ModelScope 与 FunASR\npython -m pip install -U \"funasr>=1.3.26\" modelscope",
+ ),
+ ),
+ markers=("funasr>=1.3.26", "torch.cuda.is_available()", "OpenAI 兼容"),
+ ),
+ )
+
+
+def prepare_cards(source_root: Path, output_root: Path) -> list[CardPatch]:
+ prepared: list[CardPatch] = []
+ for patch in _patches(source_root):
+ if not patch.source.exists():
+ raise FileNotFoundError(f"missing source card: {patch.source}")
+ text = patch.source.read_text()
+ for old, new in patch.replacements:
+ if old not in text:
+ raise ValueError(f"expected text not found in {patch.source}: {old[:80]!r}")
+ text = text.replace(old, new)
+ missing = [marker for marker in patch.markers if marker not in text]
+ if missing:
+ raise ValueError(f"prepared {patch.repo_id}:{patch.path_in_repo} missing {missing}")
+ destination = patch.output_path(output_root)
+ destination.parent.mkdir(parents=True, exist_ok=True)
+ destination.write_text(text)
+ prepared.append(patch)
+ return prepared
+
+
+def upload_cards(prepared: Iterable[CardPatch], output_root: Path, token: str) -> None:
+ try:
+ from modelscope.hub.api import HubApi
+ except ImportError as exc: # pragma: no cover - depends on ops environment
+ raise SystemExit("Install modelscope before using --upload") from exc
+
+ api = HubApi()
+ for patch in prepared:
+ local_path = patch.output_path(output_root)
+ print(f"uploading {patch.repo_id}:{patch.path_in_repo}")
+ api.upload_file(
+ repo_id=patch.repo_id,
+ path_or_fileobj=str(local_path),
+ path_in_repo=patch.path_in_repo,
+ repo_type="model",
+ token=token,
+ commit_message="Update FunASR quickstart guidance",
+ commit_description=(
+ "Sync model-card install guidance to funasr>=1.3.26 and add "
+ "CUDA wheel verification notes."
+ ),
+ revision="master",
+ disable_tqdm=True,
+ )
+
+
+def parse_args() -> argparse.Namespace:
+ parser = argparse.ArgumentParser(description=__doc__)
+ parser.add_argument("--source-root", type=Path, default=DEFAULT_SOURCE_ROOT)
+ parser.add_argument("--output-root", type=Path, default=DEFAULT_OUTPUT_ROOT)
+ parser.add_argument("--upload", action="store_true", help="upload prepared README files to ModelScope")
+ parser.add_argument("--token-file", type=Path, help="ModelScope access token file for --upload")
+ parser.add_argument("--token", help="ModelScope access token for --upload")
+ return parser.parse_args()
+
+
+def main() -> int:
+ args = parse_args()
+ prepared = prepare_cards(args.source_root, args.output_root)
+ for patch in prepared:
+ print(f"prepared {patch.repo_id}:{patch.path_in_repo} -> {patch.output_path(args.output_root)}")
+ if args.upload:
+ token = args.token or (args.token_file.read_text().strip() if args.token_file else "")
+ if not token:
+ raise SystemExit("--upload requires --token or --token-file")
+ upload_cards(prepared, args.output_root, token)
+ return 0
+
+
+if __name__ == "__main__":
+ raise SystemExit(main())
diff --git a/tests/test_growth_plan_doc.py b/tests/test_growth_plan_doc.py
index 165d0cd10..6db39833d 100644
--- a/tests/test_growth_plan_doc.py
+++ b/tests/test_growth_plan_doc.py
@@ -208,6 +208,28 @@ def test_growth_plan_records_sharp_github_release_focus():
assert "[ ] Create a GitHub release focused on one sharp value proposition" not in text
+def test_growth_plan_records_modelscope_card_sync_blocker():
+ text = PLAN.read_text()
+
+ required_markers = [
+ "[ ] Update ModelScope model cards and demos",
+ "`scripts/sync_modelscope_model_cards.py`",
+ "`FunAudioLLM/Fun-ASR-Nano-2512`",
+ "`README.md`, `README_zh.md`",
+ "`iic/SenseVoiceSmall`",
+ "`python -m pip install -U \"funasr>=1.3.26\" modelscope`",
+ "`torch.cuda.is_available()`",
+ "modelscope-card-sync-20260723",
+ "HTTP 400 from the ModelScope LFS batch API",
+ "`No credentials found`",
+ "https://modelscope.cn/my/myaccesstoken",
+ ]
+ for marker in required_markers:
+ assert marker in text
+
+ assert (ROOT / "scripts" / "sync_modelscope_model_cards.py").exists()
+
+
def test_growth_plan_records_pypi_description_audit():
text = PLAN.read_text()