Fix ci

.github/workflows/speaker-diarization.yaml

@@ -25,6 +25,11 @@ jobs:
         with:
           fetch-depth: 0
 
+      - name: ccache
+        uses: hendrikmuhs/[email protected]
+        with:
+          key: ${{ matrix.os }}-speaker-diarization
+
       - name: Setup Python ${{ matrix.python-version }}
         uses: actions/setup-python@v5
         with:
@@ -35,11 +40,30 @@ jobs:
         run: |
           pip install pyannote.audio onnx onnxruntime
 
+      - name: Install sherpa-onnx from source
+        shell: bash
+        run: |
+          python3 -m pip install --upgrade pip
+          python3 -m pip install wheel twine setuptools
+
+          export CMAKE_CXX_COMPILER_LAUNCHER=ccache
+          export PATH="/usr/lib/ccache:/usr/local/opt/ccache/libexec:$PATH"
+
+          cat sherpa-onnx/python/sherpa_onnx/__init__.py
+
+          python3 setup.py bdist_wheel
+          ls -lh dist
+          pip install ./dist/*.whl
+
       - name: Run tests
         shell: bash
         run: |
           pushd scripts/pyannote/segmentation
 
+          python3 -c "import sherpa_onnx; print(sherpa_onnx.__file__)"
+          python3 -c "import sherpa_onnx; print(sherpa_onnx.__version__)"
+          python3 -c "import sherpa_onnx; print(dir(sherpa_onnx))"
+
           curl -SL -O https://huggingface.co/csukuangfj/pyannote-models/resolve/main/segmentation-3.0/pytorch_model.bin
 
           test_wavs=(
@@ -53,6 +77,8 @@ jobs:
             curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speaker-segmentation-models/$w
           done
 
+          soxi *.wav
+
           curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speaker-segmentation-models/sherpa-onnx-pyannote-segmentation-3-0.tar.bz2
           tar xvf sherpa-onnx-pyannote-segmentation-3-0.tar.bz2
           rm sherpa-onnx-pyannote-segmentation-3-0.tar.bz2

.gitignore

@@ -119,3 +119,4 @@ vits-melo-tts-zh_en
 *.ppu
 sherpa-onnx-online-punct-en-2024-08-06
 *.mp4
+*.mp3

scripts/pyannote/segmentation/speaker-diarization-onnx.py

@@ -1,6 +1,11 @@
 #!/usr/bin/env python3
 # Copyright    2024  Xiaomi Corp.        (authors: Fangjun Kuang)
 
+"""
+Please refer to
+https://github.com/k2-fsa/sherpa-onnx/blob/master/.github/workflows/speaker-diarization.yaml
+for usages.
+"""
 
 import argparse
 from datetime import timedelta
@@ -64,8 +69,6 @@ def merge_segment_list(in_out: List[Segment], min_duration_off: float):
             changed = True
             break
 
-    pass
-
 
 def get_args():
     parser = argparse.ArgumentParser()

scripts/pyannote/segmentation/speaker-diarization-torch.py

@@ -1,5 +1,11 @@
 #!/usr/bin/env python3
 
+"""
+Please refer to
+https://github.com/k2-fsa/sherpa-onnx/blob/master/.github/workflows/speaker-diarization.yaml
+for usages.
+"""
+
 """
 1. Go to https://huggingface.co/hbredin/wespeaker-voxceleb-resnet34-LM/tree/main
 wget https://huggingface.co/hbredin/wespeaker-voxceleb-resnet34-LM/resolve/main/speaker-embedding.onnx

sherpa-onnx/csrc/fast-clustering.cc

@@ -52,7 +52,7 @@ class FastClustering::Impl {
     std::vector<double> height(num_rows - 1);
 
     fastclustercpp::hclust_fast(num_rows, distance.data(),
-                                fastclustercpp::HCLUST_METHOD_SINGLE,
+                                fastclustercpp::HCLUST_METHOD_COMPLETE,
                                 merge.data(), height.data());
 
     std::vector<int32_t> labels(num_rows);