Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
release: prepare FunASR 1.4.12
LauraGPTe3c4a8bea3ed
diff --git a/README.md b/README.mdindex 7ca938bd1..847fe0aba 100644--- a/README.md+++ b/README.md@@ -338,7 +338,7 @@ inference on physical Blackwell hardware.## What's new- **MOSS-Transcribe-Diarize** brings long-form ASR, timestamps, and anonymous speaker labels to FunASR services, Docker, Kubernetes, vLLM/SGLang workflows, and FunClip. [Deploy MOSS ->](./docs/moss_transcribe_diarize.md)-- **FunASR 1.4.11** is the current PyPI release, with more readable multilingual subtitles and preserved model timestamps. [Install or upgrade ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.11)+- **FunASR 1.4.12** stabilizes Fun-ASR-Nano vLLM decoding when audio compute uses FP16: the Qwen3 decoder uses BF16, while GPUs without BF16 can select FP32. Install with `python -m pip install -U "funasr==1.4.12"`. [Release ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.12)- **Production deployment** now includes faster, more resilient realtime serving and verified llama.cpp archives for ten Linux, macOS, and Windows targets. [GPU services ->](./docs/vllm_guide.md) · [CPU/edge packages ->](https://www.funasr.com/en/deploy/llama-cpp.html)> See [GitHub Releases](https://github.com/modelscope/FunASR/releases) for the complete changelog and downloadable assets.diff --git a/README_ja.md b/README_ja.mdindex ef80af4ca..6f8d3353b 100644--- a/README_ja.md+++ b/README_ja.md@@ -100,7 +100,7 @@ Whisper は単一モデルですが、**FunASR はツールキット**です—## 最新情報- **MOSS-Transcribe-Diarize** を FunASR service、Docker、Kubernetes、vLLM/SGLang workflow、FunClip に統合し、長時間 ASR、timestamp、匿名 speaker label を一度に処理できます。[MOSS をデプロイ ->](./docs/moss_transcribe_diarize.md)-- **FunASR 1.4.11** は現在の PyPI 安定版で、多言語字幕の可読性を改善しながらモデル本来の timestamp を保持します。[Install / upgrade ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.11)+- **FunASR 1.4.12** は、audio compute が FP16 のときの Fun-ASR-Nano vLLM 出力を安定化します。Qwen3 decoder は BF16 を使い、BF16 非対応 GPU では FP32 を選択できます。`python -m pip install -U "funasr==1.4.12"`。[Release ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.12)- **Production deployment** に、より高速で安定した realtime serving と、Linux、macOS、Windows の 10 target 向け llama.cpp package を追加しました。[GPU service ->](./docs/vllm_guide.md) · [CPU / edge package ->](https://www.funasr.com/en/deploy/llama-cpp.html)> 完全な変更履歴と download asset は [GitHub Releases](https://github.com/modelscope/FunASR/releases) を参照してください。diff --git a/README_ko.md b/README_ko.mdindex 8bd3724e4..dec05c761 100644--- a/README_ko.md+++ b/README_ko.md@@ -100,7 +100,7 @@ Whisper는 단일 모델이지만, **FunASR는 툴킷**입니다. 용도에 맞## 최신 소식- **MOSS-Transcribe-Diarize**를 FunASR service, Docker, Kubernetes, vLLM/SGLang workflow, FunClip에 통합해 긴 오디오 ASR, timestamp, 익명 speaker label을 한 번에 처리합니다. [MOSS 배포 ->](./docs/moss_transcribe_diarize.md)-- **FunASR 1.4.11**은 현재 PyPI 안정 버전으로, 모델의 실제 timestamp를 유지하면서 다국어 자막 가독성을 개선합니다. [Install / upgrade ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.11)+- **FunASR 1.4.12**는 audio compute가 FP16일 때 Fun-ASR-Nano vLLM 출력을 안정화합니다. Qwen3 decoder는 BF16을 사용하며 BF16 미지원 GPU에서는 FP32를 선택할 수 있습니다. `python -m pip install -U "funasr==1.4.12"`. [Release ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.12)- **Production deployment**에 더 빠르고 안정적인 realtime serving과 Linux, macOS, Windows 10개 target용 llama.cpp package를 추가했습니다. [GPU service ->](./docs/vllm_guide.md) · [CPU / edge package ->](https://www.funasr.com/en/deploy/llama-cpp.html)> 전체 변경 기록과 download asset은 [GitHub Releases](https://github.com/modelscope/FunASR/releases)에서 확인할 수 있습니다.diff --git a/README_zh.md b/README_zh.mdindex 4e8daf06a..8253d6634 100644--- a/README_zh.md+++ b/README_zh.md@@ -150,7 +150,7 @@ Whisper 是单个模型,**FunASR 是一个工具箱**——按场景挑模型## 最新动态- **MOSS-Transcribe-Diarize** 已接入 FunASR 服务、Docker、Kubernetes、vLLM/SGLang 工作流和 FunClip,一次完成长音频转写、时间戳与匿名说话人标注。[部署 MOSS ->](./docs/moss_transcribe_diarize_zh.md)-- **FunASR 1.4.11** 是当前 PyPI 稳定版,重点改善多语言可读字幕并保留模型真实时间戳。[安装或升级 ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.11)+- **FunASR 1.4.12** 修复 Fun-ASR-Nano 使用 FP16 音频计算时的 vLLM 异常输出:Qwen3 解码器改用 BF16,不支持 BF16 的 GPU 可显式选择 FP32。升级命令:`python -m pip install -U "funasr==1.4.12"`。[发布页 ->](https://github.com/modelscope/FunASR/releases/tag/v1.4.12)- **工业部署** 新增更快、更稳定的实时服务,以及覆盖 Linux、macOS、Windows 十种目标的 llama.cpp 预编译包。[GPU 服务 ->](./docs/vllm_guide_zh.md) · [CPU/端侧包 ->](https://www.funasr.com/deploy/llama-cpp.html)> 完整改动记录和可下载资产请查看 [GitHub Releases](https://github.com/modelscope/FunASR/releases)。diff --git a/funasr/version.txt b/funasr/version.txtindex 079d7f692..f86e02988 100644--- a/funasr/version.txt+++ b/funasr/version.txt@@ -1 +1 @@-1.4.11+1.4.12diff --git a/tests/test_release_version_contract.py b/tests/test_release_version_contract.pyindex d18db74a8..4e9c490b5 100644--- a/tests/test_release_version_contract.py+++ b/tests/test_release_version_contract.py@@ -2,7 +2,7 @@ from pathlib import PathROOT = Path(__file__).resolve().parents[1]-RELEASE_VERSION = "1.4.11"+RELEASE_VERSION = "1.4.12"README_FILES = ("README.md", "README_zh.md", "README_ja.md", "README_ko.md")