docs(ecosystem): add MOSS diarization deployment path

docs/deployment_matrix.md

@@ -15,6 +15,7 @@ Use this page to choose the shortest deployment path for a product, demo, benchm
 | Runtime WebSocket service | Live captions, meetings, call-center streams | [Runtime service docs](../runtime/readme.md) | Use when partial results, endpointing, or long-lived audio streams matter. |
 | ONNX/C++ runtime | High-concurrency CPU services or embedded realtime ASR | [ONNX runtime docs](../runtime/onnxruntime/readme.md) | Keep this path when latency/concurrency is already proven; add text post-processing for fixed business terms before moving to GPU LLMs. |
 | vLLM acceleration | Higher-throughput LLM-based ASR with Fun-ASR-Nano | [vLLM guide](./vllm_guide.md) | Use for LLM decoder throughput; does not apply to non-autoregressive Paraformer. |
+| MOSS-Transcribe-Diarize | Long-form multi-speaker transcription with timestamps and speaker labels | [Third-party MOSS deployment guide](./moss_transcribe_diarize.md) | OpenMOSS Apache-2.0 model served with vLLM or SGLang Omni; not a FunASR-owned model or `AutoModel` backend. |
 | MCP server | Claude/Cursor/desktop agent speech tools | [MCP example](../examples/mcp_server/) | Good when the ASR result should be exposed as a local tool. |
 | Subtitle generator | SRT/VTT from long audio or video | [Subtitle example](../examples/subtitle/) | Use verbose segments and speaker labels when readability matters. |
 | Batch ASR script | Archives, meetings, datasets, repeated offline runs | [Batch example](../examples/batch_asr_improved.py) | Add queueing, manifests, and retry logs for production use. |

docs/deployment_matrix_ja.md

@@ -13,6 +13,7 @@
 | Kubernetes API | Cluster service 向け internal speech API | [Kubernetes template](../examples/openai_api/kubernetes/) | private `ClusterIP` から開始。公開範囲を広げる前に auth、TLS、network policy、GPU scheduling を追加します。 |
 | Runtime WebSocket service | Live captions、meeting、call-center stream | [Runtime service docs](../runtime/readme.md) | partial result、endpointing、long-lived audio stream が重要な場合に使います。 |
 | vLLM acceleration | Fun-ASR-Nano の LLM-based ASR throughput 向上 | [vLLM guide](./vllm_guide.md) | LLM decoder throughput 向け。non-autoregressive Paraformer には適用しません。 |
+| MOSS-Transcribe-Diarize | 長時間の複数話者 transcription、timestamp、speaker label | [Third-party MOSS guide](./moss_transcribe_diarize.md) | OpenMOSS の Apache-2.0 model を vLLM / SGLang Omni で提供します。FunASR 所有 model や `AutoModel` backend ではありません。 |
 | MCP server | Claude/Cursor/desktop agent の speech tool | [MCP example](../examples/mcp_server/) | ASR 結果を local tool として Agent に渡したい場合に便利です。 |
 | Subtitle generator | 長時間 audio/video から SRT/VTT 作成 | [Subtitle example](../examples/subtitle/) | readability が重要な場合は verbose segment と speaker label を使います。 |
 | Batch ASR script | Archive、meeting、dataset、繰り返し offline run | [Batch example](../examples/batch_asr_improved.py) | production では queue、manifest、retry log を追加してください。 |

docs/deployment_matrix_ko.md

@@ -13,6 +13,7 @@
 | Kubernetes API | Cluster service용 internal speech API | [Kubernetes template](../examples/openai_api/kubernetes/) | private `ClusterIP`부터 시작합니다. 범위를 넓히기 전에 auth, TLS, network policy, GPU scheduling을 추가하세요. |
 | Runtime WebSocket service | Live captions, meeting, call-center stream | [Runtime service docs](../runtime/readme.md) | partial result, endpointing, long-lived audio stream이 중요할 때 사용합니다. |
 | vLLM acceleration | Fun-ASR-Nano의 LLM-based ASR throughput 향상 | [vLLM guide](./vllm_guide.md) | LLM decoder throughput용입니다. non-autoregressive Paraformer에는 적용되지 않습니다. |
+| MOSS-Transcribe-Diarize | 긴 다중 화자 transcription, timestamp, speaker label | [Third-party MOSS guide](./moss_transcribe_diarize.md) | OpenMOSS의 Apache-2.0 model을 vLLM / SGLang Omni로 제공합니다. FunASR 소유 model 또는 `AutoModel` backend가 아닙니다. |
 | MCP server | Claude/Cursor/desktop agent speech tool | [MCP example](../examples/mcp_server/) | ASR 결과를 local tool로 Agent에 전달할 때 유용합니다. |
 | Subtitle generator | 긴 audio/video에서 SRT/VTT 생성 | [Subtitle example](../examples/subtitle/) | readability가 중요하면 verbose segment와 speaker label을 사용합니다. |
 | Batch ASR script | Archive, meeting, dataset, 반복 offline run | [Batch example](../examples/batch_asr_improved.py) | production에서는 queue, manifest, retry log를 추가하세요. |

docs/deployment_matrix_zh.md

@@ -15,6 +15,7 @@
 | Runtime WebSocket 服务 | 实时字幕、会议、客服流式音频 | [Runtime 服务文档](../runtime/readme_cn.md) | 需要中间结果、断句或长连接音频流时选择。 |
 | ONNX/C++ Runtime | 高并发 CPU 服务或嵌入式实时 ASR | [ONNX Runtime 文档](../runtime/onnxruntime/readme.md) | 如果延迟和并发已经验证，不要轻易替换；固定业务词优先做文本后处理。 |
 | vLLM 加速 | Fun-ASR-Nano 等 LLM-based ASR 高吞吐 | [vLLM 指南](./vllm_guide.md) | 适合 LLM 解码吞吐；不适用于非自回归 Paraformer。 |
+| MOSS-Transcribe-Diarize | 长音频多人转写、时间戳和说话人标签 | [第三方 MOSS 部署指南](./moss_transcribe_diarize.md) | OpenMOSS 以 Apache-2.0 发布，通过 vLLM 或 SGLang Omni 服务；不是 FunASR 自有模型或 `AutoModel` 后端。 |
 | MCP 服务 | Claude/Cursor/桌面 Agent 语音工具 | [MCP 示例](../examples/mcp_server/) | 适合把 ASR 结果暴露成一个本地工具。 |
 | 字幕生成 | 从长音频或视频生成 SRT/VTT | [字幕示例](../examples/subtitle/) | 需要可读性时使用 verbose segments 和说话人标签。 |
 | 批处理脚本 | 录音归档、会议纪要、数据集处理 | [批处理示例](../examples/batch_asr_improved.py) | 生产使用时建议增加队列、manifest 和重试日志。 |

docs/moss_transcribe_diarize.md

@@ -0,0 +1,153 @@
+# MOSS-Transcribe-Diarize in the FunASR ecosystem
+
+[中文](./moss_transcribe_diarize_zh.md)
+
+This guide connects the third-party
+[OpenMOSS/MOSS-Transcribe-Diarize](https://github.com/OpenMOSS/MOSS-Transcribe-Diarize)
+model to the FunASR deployment ecosystem. The model is published by OpenMOSS
+under Apache-2.0; it is not a FunASR model and is not registered in FunASR
+`AutoModel`.
+
+MOSS-Transcribe-Diarize jointly generates transcription, timestamps, and
+speaker labels such as `[S01]`. An application therefore does not need to
+assemble an external VAD, ASR, and diarization pipeline. This is a deployment
+property, not a claim that the model has no internal segmentation or chunking.
+
+## Pinned sources
+
+This guide was verified against:
+
+- source: `OpenMOSS/MOSS-Transcribe-Diarize@cb765f2b0fe6f7a298aa2002e2281ae693d1f3c3`
+- model: `OpenMOSS-Team/MOSS-Transcribe-Diarize@e8681d68e7042738ffca8ac8212bc8fcb1131ab8`
+- license: Apache-2.0 in the pinned upstream source and model metadata
+- vLLM nightly index: `68b4a1d582818e67adc903bf1b8fc5a5447da2fa`
+
+Pin all three revisions. The model uses `trust_remote_code`; do not execute a
+floating model revision in a production service.
+
+## Choose a serving path
+
+| Path | Environment | Response contract | Use when |
+|---|---|---|---|
+| vLLM | CUDA 12 (`cu129`) or CUDA 13 (`cu130`) | `response_format=json` returns raw `[start][Sxx]text[end]` text | You already operate vLLM or need its scheduler |
+| SGLang Omni | CUDA 13 in the current upstream guide | `response_format=verbose_json` returns parsed segments | You need structured speaker segments directly from the API |
+| Transformers | PyTorch process | Python objects and raw tagged text | Evaluation, debugging, or custom preprocessing |
+
+The two HTTP backends share `/v1/audio/transcriptions`, but their documented
+response formats are not interchangeable. Do not promise vLLM `segments`
+without testing the exact vLLM revision.
+
+## vLLM
+
+Create an isolated environment and install the upstream-pinned nightly build:
+
+```bash
+uv venv --python 3.12 .venv-moss
+uv pip install --python .venv-moss/bin/python -U 'vllm[audio]' \
+  --torch-backend=auto \
+  --extra-index-url https://wheels.vllm.ai/68b4a1d582818e67adc903bf1b8fc5a5447da2fa/cu129
+```
+
+The `audio` extra is required. A plain `vllm` install can start the server but
+returns HTTP 400 (`Invalid or unsupported audio file`) because no audio decoder
+is installed.
+
+Start the service with the immutable model revision:
+
+```bash
+CUDA_VISIBLE_DEVICES=0 .venv-moss/bin/vllm serve \
+  OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  --revision e8681d68e7042738ffca8ac8212bc8fcb1131ab8 \
+  --served-model-name moss-transcribe-diarize \
+  --trust-remote-code \
+  --host 127.0.0.1 \
+  --port 8898
+```
+
+Submit audio and validate the speaker-tagged text:
+
+```bash
+curl -fsS http://127.0.0.1:8898/v1/audio/transcriptions \
+  -F file=@audio.wav \
+  -F model=moss-transcribe-diarize \
+  -F response_format=json \
+  -F temperature=0 \
+  | tee moss-transcription.json
+
+python - <<'PY'
+import json
+
+with open("moss-transcription.json", encoding="utf-8") as stream:
+    payload = json.load(stream)
+text = payload.get("text", "")
+assert text.strip(), payload
+assert "[S01]" in text, text
+print(text)
+PY
+```
+
+For long recordings, test `max_completion_tokens` against the longest expected
+meeting. A larger limit can increase memory use and tail latency.
+
+### Reproduced vLLM contract
+
+The FunASR ecosystem validation used vLLM
+`0.23.1rc1.dev949+g68b4a1d58`, Torch `2.11.0+cu129`, and one H100 80GB:
+
+- the bundled 6.000-second sample
+  (`ea03e1f473ad1618a03da3327a545369cb8f6f06cb0f4115535e5a866167d47e`)
+  returned HTTP 200 and `[0.96][S01]... [5.94]`;
+- an A + 0.8-second silence + B + 0.8-second silence + A probe
+  (`dbb32bcfed2e8226bedf64248a9f4a44685b293a4696d18fb4cfa701b04db912`)
+  returned HTTP 200 with `S01 -> S02 -> S01` and timestamps through 19.08 seconds.
+
+This proves the pinned API and speaker-return contract on those inputs. It is
+not a diarization accuracy result, overlap test, throughput benchmark, or
+production capacity claim.
+
+## SGLang Omni
+
+Follow the pinned upstream SGLang Omni installation guide for CUDA 13, then
+download the immutable model snapshot and serve the local directory:
+
+```bash
+hf download OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  --revision e8681d68e7042738ffca8ac8212bc8fcb1131ab8 \
+  --local-dir .models/moss-transcribe-diarize
+
+sgl-omni serve \
+  --model-path .models/moss-transcribe-diarize \
+  --port 8898 \
+  --max-running-requests 16 \
+  --cuda-graph-max-bs 16 \
+  --mem-fraction-static 0.80
+```
+
+Request parsed segments:
+
+```bash
+curl -fsS http://127.0.0.1:8898/v1/audio/transcriptions \
+  -F file=@audio.wav \
+  -F model=OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  -F response_format=verbose_json
+```
+
+Verify every segment has start/end timing, text, and the expected speaker
+field before wiring the response into subtitles, meeting notes, or analytics.
+
+## Production validation
+
+Use real multi-speaker audio, not only a short single-speaker smoke sample:
+
+- measure speaker consistency across long turns and speaker returns;
+- test overlap, crosstalk, music, noise, and long silence;
+- verify timestamp monotonicity and full-audio coverage;
+- record GPU, CUDA, Torch, backend commit, model revision, audio duration,
+  generation limit, wall latency, and peak memory;
+- keep authentication, TLS, request limits, and retention policy at the API
+  gateway; bind the model worker to a private address.
+
+When reporting a problem to FunASR, state that this is the OpenMOSS third-party
+path and include the exact backend and upstream revisions. Model architecture or
+weight issues belong upstream; deployment-center contract and documentation
+issues can be filed in FunASR.

docs/moss_transcribe_diarize_zh.md

@@ -0,0 +1,141 @@
+# 在 FunASR 生态中部署 MOSS-Transcribe-Diarize
+
+[English](./moss_transcribe_diarize.md)
+
+本文把第三方
+[OpenMOSS/MOSS-Transcribe-Diarize](https://github.com/OpenMOSS/MOSS-Transcribe-Diarize)
+模型接入 FunASR 部署生态。模型由 OpenMOSS 以 Apache-2.0 发布，不是 FunASR
+自有模型，也没有注册到 FunASR `AutoModel`。
+
+MOSS-Transcribe-Diarize 会联合生成转写、时间戳和 `[S01]` 等说话人标签，应用侧
+不必再拼接外部 VAD、ASR 和 diarization 管线。这里描述的是部署形态，不表示模型
+内部没有分块或分段。
+
+## 固定上游版本
+
+本文按以下不可变版本验证：
+
+- 源码：`OpenMOSS/MOSS-Transcribe-Diarize@cb765f2b0fe6f7a298aa2002e2281ae693d1f3c3`
+- 模型：`OpenMOSS-Team/MOSS-Transcribe-Diarize@e8681d68e7042738ffca8ac8212bc8fcb1131ab8`
+- 许可证：固定上游源码和模型元数据中的 Apache-2.0
+- vLLM nightly 索引：`68b4a1d582818e67adc903bf1b8fc5a5447da2fa`
+
+三者都要固定。模型依赖 `trust_remote_code`，生产服务不要执行浮动的模型 revision。
+
+## 选择服务后端
+
+| 路径 | 环境 | 响应契约 | 适用场景 |
+|---|---|---|---|
+| vLLM | CUDA 12（`cu129`）或 CUDA 13（`cu130`） | `response_format=json` 返回原始 `[start][Sxx]text[end]` 文本 | 已运行 vLLM，或需要其调度能力 |
+| SGLang Omni | 当前上游指南为 CUDA 13 | `response_format=verbose_json` 返回解析后的 segments | API 需要直接返回结构化说话人分段 |
+| Transformers | PyTorch 进程 | Python 对象和原始标签文本 | 评测、排障或自定义预处理 |
+
+两个 HTTP 后端都使用 `/v1/audio/transcriptions`，但官方文档中的响应格式不能混为
+一谈。没有对 exact vLLM revision 做验证时，不要承诺 vLLM 返回 `segments`。
+
+## vLLM
+
+创建隔离环境并安装上游固定的 nightly build：
+
+```bash
+uv venv --python 3.12 .venv-moss
+uv pip install --python .venv-moss/bin/python -U 'vllm[audio]' \
+  --torch-backend=auto \
+  --extra-index-url https://wheels.vllm.ai/68b4a1d582818e67adc903bf1b8fc5a5447da2fa/cu129
+```
+
+必须安装 `audio` extra。只安装 plain `vllm` 时服务可以启动，但因为没有音频解码器，
+请求会返回 HTTP 400（`Invalid or unsupported audio file`）。
+
+使用不可变模型 revision 启动：
+
+```bash
+CUDA_VISIBLE_DEVICES=0 .venv-moss/bin/vllm serve \
+  OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  --revision e8681d68e7042738ffca8ac8212bc8fcb1131ab8 \
+  --served-model-name moss-transcribe-diarize \
+  --trust-remote-code \
+  --host 127.0.0.1 \
+  --port 8898
+```
+
+提交音频并验证说话人标签文本：
+
+```bash
+curl -fsS http://127.0.0.1:8898/v1/audio/transcriptions \
+  -F file=@audio.wav \
+  -F model=moss-transcribe-diarize \
+  -F response_format=json \
+  -F temperature=0 \
+  | tee moss-transcription.json
+
+python - <<'PY'
+import json
+
+with open("moss-transcription.json", encoding="utf-8") as stream:
+    payload = json.load(stream)
+text = payload.get("text", "")
+assert text.strip(), payload
+assert "[S01]" in text, text
+print(text)
+PY
+```
+
+长音频需要按业务最长会议验证 `max_completion_tokens`。增大上限也会影响显存和尾延迟。
+
+### 已复现的 vLLM 契约
+
+FunASR 生态验证使用 vLLM `0.23.1rc1.dev949+g68b4a1d58`、Torch
+`2.11.0+cu129` 和一张 H100 80GB：
+
+- 仓库内 6.000 秒样例
+  （`ea03e1f473ad1618a03da3327a545369cb8f6f06cb0f4115535e5a866167d47e`）
+  返回 HTTP 200 和 `[0.96][S01]... [5.94]`；
+- A + 0.8 秒静音 + B + 0.8 秒静音 + A 的拼接样例
+  （`dbb32bcfed2e8226bedf64248a9f4a44685b293a4696d18fb4cfa701b04db912`）
+  返回 HTTP 200、`S01 -> S02 -> S01`，时间戳覆盖到 19.08 秒。
+
+这证明固定版本对这些输入满足 API 和说话人返回契约，不是 diarization 精度、重叠语音、
+吞吐或生产容量结论。
+
+## SGLang Omni
+
+按固定上游 SGLang Omni 安装指南准备 CUDA 13 环境，然后下载不可变模型快照并从本地目录启动：
+
+```bash
+hf download OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  --revision e8681d68e7042738ffca8ac8212bc8fcb1131ab8 \
+  --local-dir .models/moss-transcribe-diarize
+
+sgl-omni serve \
+  --model-path .models/moss-transcribe-diarize \
+  --port 8898 \
+  --max-running-requests 16 \
+  --cuda-graph-max-bs 16 \
+  --mem-fraction-static 0.80
+```
+
+请求结构化分段：
+
+```bash
+curl -fsS http://127.0.0.1:8898/v1/audio/transcriptions \
+  -F file=@audio.wav \
+  -F model=OpenMOSS-Team/MOSS-Transcribe-Diarize \
+  -F response_format=verbose_json
+```
+
+接入字幕、会议纪要或分析系统前，验证每个 segment 都包含起止时间、文本和预期的 speaker 字段。
+
+## 上线前验证
+
+不要只用短单人音频，应使用真实多人长音频：
+
+- 检查长轮次及说话人再次出现时的身份一致性；
+- 覆盖重叠语音、串音、音乐、噪声和长静音；
+- 验证时间戳单调递增并覆盖完整音频；
+- 记录 GPU、CUDA、Torch、后端 commit、模型 revision、音频时长、生成上限、
+  墙钟延迟和峰值显存；
+- 在 API 网关完成认证、TLS、请求限制和数据留存策略，模型 worker 只绑定内网。
+
+向 FunASR 报告问题时，请明确这是 OpenMOSS 第三方路径，并附上后端与上游 exact
+revision。模型架构或权重问题应反馈给上游；部署中心的契约或文档问题可以在 FunASR 提交。

tests/test_moss_transcribe_diarize_docs.py

@@ -0,0 +1,50 @@
+from pathlib import Path
+
+import pytest
+
+
+ROOT = Path(__file__).resolve().parents[1]
+GUIDES = (
+    ROOT / "docs" / "moss_transcribe_diarize.md",
+    ROOT / "docs" / "moss_transcribe_diarize_zh.md",
+)
+MATRICES = (
+    ROOT / "docs" / "deployment_matrix.md",
+    ROOT / "docs" / "deployment_matrix_zh.md",
+    ROOT / "docs" / "deployment_matrix_ja.md",
+    ROOT / "docs" / "deployment_matrix_ko.md",
+)
+
+
+@pytest.mark.parametrize("guide", GUIDES)
+def test_moss_guides_pin_upstream_and_separate_serving_contracts(guide: Path) -> None:
+    text = guide.read_text(encoding="utf-8")
+
+    for marker in (
+        "OpenMOSS/MOSS-Transcribe-Diarize",
+        "cb765f2b0fe6f7a298aa2002e2281ae693d1f3c3",
+        "OpenMOSS-Team/MOSS-Transcribe-Diarize",
+        "e8681d68e7042738ffca8ac8212bc8fcb1131ab8",
+        "68b4a1d582818e67adc903bf1b8fc5a5447da2fa",
+        "vllm[audio]",
+        "0.23.1rc1.dev949+g68b4a1d58",
+        "dbb32bcfed2e8226bedf64248a9f4a44685b293a4696d18fb4cfa701b04db912",
+        "S01 -> S02 -> S01",
+        "/v1/audio/transcriptions",
+        "response_format=json",
+        "response_format=verbose_json",
+        "[S01]",
+        "Apache-2.0",
+    ):
+        assert marker in text, f"{guide.name} is missing {marker}"
+
+    assert "AutoModel" in text
+    assert "third-party" in text.lower() or "第三方" in text
+
+
+@pytest.mark.parametrize("matrix", MATRICES)
+def test_all_deployment_matrices_link_moss_guide(matrix: Path) -> None:
+    text = matrix.read_text(encoding="utf-8")
+
+    assert "MOSS-Transcribe-Diarize" in text
+    assert "moss_transcribe_diarize.md" in text