chore: import upstream snapshot with attribution

2026-07-13 13:25:10 +08:00
commit c397331b1e
3684 changed files with 990993 additions and 0 deletions
@@ -0,0 +1,41 @@
+---
+name: ❓ Questions/Help
+about: Ask for help after checking existing issues and docs
+labels: 'question, needs triage'
+---
+
+Notice: In order to resolve issues efficiently, please follow the template.
+（注意：为了更加高效率解决您遇到的问题，请按照模板提问，补充细节。）
+
+## Before asking
+
+1. Search existing issues: https://github.com/modelscope/FunASR/issues
+2. Search the docs: https://modelscope.github.io/FunASR/
+3. Check the README quick start and deployment section.
+
+## Question
+
+<!-- What are you trying to do, and where are you blocked? -->
+
+## Code or command
+
+```bash
+
+```
+
+## What have you tried?
+
+<!-- Include links to docs or examples you followed, plus the exact error/result. -->
+
+## Environment
+
+- OS:
+- Python version:
+- FunASR version:
+- ModelScope version:
+- PyTorch / torchaudio version:
+- Install method (`pip`, source, Docker):
+- Device (`cuda`, `cpu`, `mps`):
+- GPU model:
+- CUDA/cuDNN version:
+- Docker image tag, if used:
@@ -0,0 +1,64 @@
+---
+name: 🐛 Bug Report
+about: Submit a reproducible bug report to help us improve FunASR
+labels: 'bug, needs triage'
+---
+
+Notice: In order to resolve issues efficiently, please follow the template and include reproducible details.
+（注意：为了更加高效率解决您遇到的问题，请按照模板提问，补充可复现细节。）
+
+## 🐛 Bug
+
+<!-- A clear and concise description of the bug. -->
+
+## To Reproduce
+
+Steps to reproduce the behavior. Always include the exact command you ran.
+
+1. Install with: `...`
+2. Run: `...`
+3. See error: `...`
+
+## Code sample
+
+<!-- Paste the smallest code sample that still reproduces the issue. -->
+
+```python
+
+```
+
+## Expected behavior
+
+<!-- A clear and concise description of what you expected to happen. -->
+
+## Error logs
+
+<!-- Paste the full traceback or relevant logs. Remove private tokens and paths. -->
+
+```text
+
+```
+
+## Environment
+
+- OS:
+- Python version:
+- FunASR version:
+- ModelScope version:
+- PyTorch / torchaudio version:
+- Install method (`pip`, source, Docker):
+- Device (`cuda`, `cpu`, `mps`):
+- GPU model:
+- CUDA/cuDNN version:
+- Docker image tag, if used:
+
+## Audio details
+
+If the audio cannot be shared, please describe:
+
+- Duration:
+- Sample rate:
+- Format:
+- Language/dialect:
+- Speaker count:
+- Background noise/music:
@@ -0,0 +1,11 @@
+blank_issues_enabled: false
+contact_links:
+  - name: FunASR documentation
+    url: https://modelscope.github.io/FunASR/
+    about: Read installation, tutorial, deployment, benchmark, and API docs.
+  - name: GitHub Discussions
+    url: https://github.com/modelscope/FunASR/discussions
+    about: Share ideas, showcase usage, and ask open-ended community questions.
+  - name: Hugging Face models
+    url: https://huggingface.co/funasr
+    about: Browse FunASR checkpoints on Hugging Face.
@@ -0,0 +1,45 @@
+---
+name: 🚀 Deployment Help
+about: Get help with API server, WebSocket, Docker, vLLM, Triton, Android, or agent integration
+labels: 'deployment, needs triage'
+---
+
+## Deployment target
+
+- [ ] OpenAI-compatible API server
+- [ ] WebSocket streaming service
+- [ ] Docker runtime
+- [ ] vLLM acceleration
+- [ ] Triton
+- [ ] Android / mobile
+- [ ] Browser / Web UI
+- [ ] MCP / agent integration
+- [ ] Other:
+
+## Goal
+
+<!-- What user-facing service or workflow are you trying to deploy? -->
+
+## Command or config
+
+```bash
+
+```
+
+## Error or unexpected behavior
+
+```text
+
+```
+
+## Environment
+
+- OS:
+- Python version:
+- FunASR version:
+- Install method (`pip`, source, Docker):
+- Device (`cuda`, `cpu`, `mps`):
+- GPU model and memory:
+- CUDA/cuDNN version:
+- Docker image tag, if used:
+- Network endpoint / port:
@@ -0,0 +1,15 @@
+---
+name: 📚 Documentation/Typos
+about: Report an issue related to documentation or a typo
+labels: 'documentation, needs triage'
+---
+
+## 📚 Documentation
+
+For typos and doc fixes, please go ahead and:
+
+1. Create an issue.
+2. Fix the typo.
+3. Submit a PR.
+
+Thanks!
@@ -0,0 +1,27 @@
+---
+name: ✨ Feature Request
+about: Suggest a model, language, runtime, deployment, benchmark, or documentation improvement
+labels: 'enhancement, needs triage'
+---
+
+## Use case
+
+<!-- What are you trying to build, deploy, evaluate, or research with FunASR? -->
+
+## Proposed solution
+
+<!-- Describe the desired model/API/runtime/doc/benchmark behavior. -->
+
+## Alternatives or workarounds
+
+<!-- What workaround are you using today? What other tools did you consider? -->
+
+## Expected impact
+
+<!-- Who benefits: new users, production deployers, researchers, agent builders, educators, etc.? -->
+
+## Willing to contribute?
+
+- [ ] I can submit a PR
+- [ ] I can help test
+- [ ] I can provide data, logs, or benchmark results
@@ -0,0 +1,71 @@
+---
+name: 📊 Migration Benchmark Report
+about: Share FunASR results when comparing against Whisper, OpenAI audio APIs, or cloud ASR
+labels: 'benchmark, showcase, needs triage'
+---
+
+Thanks for benchmarking FunASR on your own audio. Migration reports help new users decide whether FunASR fits their language, domain, hardware, and deployment constraints.
+
+If you need help debugging a failure, please use Bug Report or Deployment Help instead.
+
+## Summary
+
+<!-- One or two sentences: what did you compare, and what was the headline result? -->
+
+## Baseline
+
+- Baseline ASR (`Whisper`, `Whisper large-v3`, cloud provider, internal system, other):
+- Baseline runtime or API:
+- Baseline hardware or pricing tier:
+
+## FunASR setup
+
+- FunASR version:
+- Model(s):
+- Runtime path (`Python API`, `funasr-server`, `OpenAI API`, `Docker`, `WebSocket`, `vLLM`, other):
+- Device (`cuda`, `cpu`, `mps`):
+- GPU / CPU:
+- CUDA / PyTorch versions:
+- Command or script used:
+
+```bash
+
+```
+
+## Audio set
+
+- Number of files:
+- Total audio duration:
+- Language(s) / dialect(s):
+- Domain (`meeting`, `call-center`, `subtitle`, `lecture`, `media`, `noisy field audio`, other):
+- Speaker count range:
+- Sample rate / format:
+- Can any sample be shared publicly? yes/no
+
+## Results
+
+<!-- Paste the aggregate section from examples/migration/benchmark_funasr.py summary.md, or summarize your own measurement. -->
+
+```text
+
+```
+
+## Quality notes
+
+<!-- Share WER/CER if available, or human-review notes about names, numbers, punctuation, timestamps, speaker labels, and domain terms. -->
+
+## Operational notes
+
+- Model download / warmup time:
+- Steady-state throughput:
+- Memory usage:
+- Failed files or error rate:
+- Deployment blockers:
+
+## Links or artifacts
+
+<!-- Public repo, demo, blog, benchmark sheet, screenshot, anonymized transcript snippet, or architecture diagram. Do not include private audio, credentials, or customer data. -->
+
+## What should FunASR improve next?
+
+<!-- Missing docs, rough edges, model gaps, deployment friction, benchmark needs, etc. -->
@@ -0,0 +1,56 @@
+---
+name: 🎤 Showcase / Production Story
+about: Share how you use FunASR in a demo, product, benchmark, research project, or internal workflow
+labels: 'showcase, needs triage'
+---
+
+Thanks for sharing your FunASR use case. Concrete usage reports help new users choose the right path and help maintainers prioritize docs, examples, models, and deployment work.
+
+If this is a support question, please use Bug Report or Deployment Help instead.
+
+## Summary
+
+<!-- What did you build or evaluate with FunASR? One or two sentences is enough. -->
+
+## Scenario
+
+- [ ] Local transcription demo
+- [ ] Private OpenAI-compatible speech API
+- [ ] Agent / MCP / voice-input workflow
+- [ ] Streaming ASR / live captions
+- [ ] Subtitle generation
+- [ ] Batch transcription / archive processing
+- [ ] Benchmark or migration from Whisper/cloud ASR
+- [ ] Research / education
+- [ ] Production deployment
+- [ ] Other:
+
+## Deployment details
+
+- FunASR version:
+- Model(s):
+- Device (`cuda`, `cpu`, `mps`):
+- GPU / CPU:
+- Runtime path (`Python API`, `funasr-server`, `WebSocket`, `Docker`, `vLLM`, `MCP`, other):
+- Audio language/domain:
+- Approximate audio duration or traffic:
+
+## Results
+
+<!-- Share speed, accuracy/CER/WER, latency, cost, CPU/GPU usage, or qualitative results if available. -->
+
+```text
+
+```
+
+## Links or screenshots
+
+<!-- Public demo, article, repo, benchmark report, screenshot, or architecture diagram. Do not include private audio, credentials, or customer data. -->
+
+## What was easy?
+
+<!-- Which docs, examples, or APIs helped most? -->
+
+## What should be improved?
+
+<!-- Missing docs, rough edges, model gaps, deployment friction, benchmark needs, etc. -->
@@ -0,0 +1,28 @@
+## Summary
+
+<!-- What changed and why? -->
+
+## Type of change
+
+- [ ] Bug fix
+- [ ] Documentation
+- [ ] Example or demo
+- [ ] Runtime or deployment
+- [ ] Benchmark or evaluation
+- [ ] Model/training change
+
+## Validation
+
+<!-- Commands run, logs, screenshots, benchmark results, or why validation is not applicable. -->
+
+- [ ] `python -m compileall funasr examples tests`
+- [ ] Docs or links checked
+- [ ] Runtime/deployment command tested
+
+## User impact
+
+<!-- Who benefits from this change? New users, production deployers, researchers, agent builders, etc. -->
+
+## Notes for reviewers
+
+<!-- Compatibility notes, risks, follow-up work, or review focus. -->
@@ -0,0 +1,135 @@
+name: Build llama.cpp runtime binaries
+
+# Cross-platform prebuilt binaries for the FunASR llama.cpp / GGUF runtime,
+# like whisper.cpp's whisper-bin-*. Push a `runtime-llamacpp-v*` tag to publish
+# a GitHub Release with the binaries; or run manually (workflow_dispatch) to test
+# the build without creating a release.
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - 'runtime-llamacpp-v*'
+
+permissions:
+  contents: write
+
+jobs:
+  build:
+    name: build-${{ matrix.name }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: ubuntu-latest
+            name: linux-x64
+            cmake_flags: >-
+              -DGGML_NATIVE=OFF
+              -DGGML_AVX=OFF
+              -DGGML_AVX2=OFF
+              -DGGML_AVX_VNNI=OFF
+              -DGGML_AVX512=OFF
+              -DGGML_AVX512_VBMI=OFF
+              -DGGML_AVX512_VNNI=OFF
+              -DGGML_AVX512_BF16=OFF
+              -DGGML_FMA=OFF
+              -DGGML_F16C=OFF
+              -DGGML_BMI2=OFF
+          - os: ubuntu-latest
+            name: linux-x64-avx2
+            cmake_flags: >-
+              -DGGML_NATIVE=OFF
+              -DGGML_AVX=ON
+              -DGGML_AVX2=ON
+              -DGGML_AVX_VNNI=OFF
+              -DGGML_AVX512=OFF
+              -DGGML_AVX512_VBMI=OFF
+              -DGGML_AVX512_VNNI=OFF
+              -DGGML_AVX512_BF16=OFF
+              -DGGML_FMA=ON
+              -DGGML_F16C=ON
+              -DGGML_BMI2=ON
+          - os: ubuntu-24.04-arm
+            name: linux-arm64
+            cmake_flags: -DGGML_NATIVE=OFF
+          - os: macos-latest
+            name: macos-arm64
+            cmake_flags: -DGGML_NATIVE=OFF
+          - os: windows-latest
+            name: windows-x64
+            cmake_flags: >-
+              -DGGML_NATIVE=OFF
+              -DGGML_AVX=OFF
+              -DGGML_AVX2=OFF
+              -DGGML_AVX_VNNI=OFF
+              -DGGML_AVX512=OFF
+              -DGGML_AVX512_VBMI=OFF
+              -DGGML_AVX512_VNNI=OFF
+              -DGGML_AVX512_BF16=OFF
+              -DGGML_FMA=OFF
+              -DGGML_F16C=OFF
+              -DGGML_BMI2=OFF
+          - os: windows-latest
+            name: windows-x64-avx2
+            cmake_flags: >-
+              -DGGML_NATIVE=OFF
+              -DGGML_AVX=ON
+              -DGGML_AVX2=ON
+              -DGGML_AVX_VNNI=OFF
+              -DGGML_AVX512=OFF
+              -DGGML_AVX512_VBMI=OFF
+              -DGGML_AVX512_VNNI=OFF
+              -DGGML_AVX512_BF16=OFF
+              -DGGML_FMA=ON
+              -DGGML_F16C=ON
+              -DGGML_BMI2=ON
+    runs-on: ${{ matrix.os }}
+    defaults:
+      run:
+        shell: bash
+        working-directory: runtime/llama.cpp
+    steps:
+      - uses: actions/checkout@v4
+      - name: Configure and build
+        run: |
+          # Release binaries must be portable across user machines, not tuned to
+          # the ephemeral CI runner CPU. A native ggml build may emit AVX512 or
+          # AVX-VNNI instructions and then crash with SIGILL on ordinary CPUs.
+          # Keep the default x64 package conservative, and publish explicit
+          # x64-avx2 assets for machines that support AVX2/FMA/F16C/BMI2.
+          cmake -B build -DCMAKE_BUILD_TYPE=Release ${{ matrix.cmake_flags }}
+          cmake --build build --config Release -j 2
+      - name: Package
+        run: |
+          mkdir -p pkg
+          cp build/bin/llama-funasr-* pkg/ 2>/dev/null || true
+          cp build/bin/Release/llama-funasr-* pkg/ 2>/dev/null || true
+          cp README.md download-funasr-model.sh pkg/ 2>/dev/null || true
+          echo "--- packaged binaries ---"; ls -la pkg
+          if [ "${{ runner.os }}" = "Windows" ]; then
+            7z a "funasr-llamacpp-${{ matrix.name }}.zip" ./pkg/*
+          else
+            tar czf "funasr-llamacpp-${{ matrix.name }}.tar.gz" -C pkg .
+          fi
+      - uses: actions/upload-artifact@v4
+        with:
+          name: funasr-llamacpp-${{ matrix.name }}
+          path: runtime/llama.cpp/funasr-llamacpp-${{ matrix.name }}.*
+          if-no-files-found: error
+
+  release:
+    needs: build
+    if: startsWith(github.ref, 'refs/tags/')
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          path: dist
+          merge-multiple: true
+      - name: Create GitHub Release
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          gh release create "${{ github.ref_name }}" dist/* \
+            --repo "${{ github.repository }}" \
+            --title "FunASR llama.cpp runtime ${{ github.ref_name }}" \
+            --notes "Prebuilt self-contained binaries for the FunASR llama.cpp / GGUF runtime — SenseVoice, Paraformer and Fun-ASR-Nano with built-in FSMN-VAD (a whisper.cpp-style on-device ASR, strong on Chinese). Get a model with \`bash download-funasr-model.sh <sensevoice|paraformer|nano>\`, then run \`llama-funasr-cli\` / \`llama-funasr-sensevoice\` / \`llama-funasr-paraformer\`. Use the default x64 asset for maximum CPU compatibility; use the x64-avx2 asset on CPUs with AVX2/FMA/F16C/BMI2 for higher throughput. No Python, no build. Docs: runtime/llama.cpp/README.md"
@@ -0,0 +1,21 @@
+name: Release on tag
+
+on:
+  push:
+    tags:
+      - "v*"
+
+permissions:
+  contents: write
+
+jobs:
+  release:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Create GitHub Release with auto-generated notes
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: gh release create "${{ github.ref_name }}" --generate-notes --verify-tag --title "${{ github.ref_name }}"
@@ -0,0 +1,33 @@
+name: Update API Documentation
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - 'funasr/**/*.py'
+      - 'scripts/gen_api_docs.py'
+      - 'training.html'
+      - '.github/workflows/update-api-docs.yml'
+
+jobs:
+  build-api-docs:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Generate API docs
+        run: python3 scripts/gen_api_docs.py
+
+      - name: Deploy to gh-pages
+        uses: peaceiris/actions-gh-pages@v3
+        with:
+          github_token: ${{ secrets.GITHUB_TOKEN }}
+          publish_dir: ./gh-pages-output
+          publish_branch: gh-pages
+          keep_files: true