Add CI workflow for testing and update model registration

Author: lifeiteng

.github/workflows/test.yml

@@ -0,0 +1,48 @@
+name: Run Tests
+
+on:
+  push:
+    branches: [ main, dev, develop ]
+  pull_request:
+    branches: [ main, dev, develop ]
+  workflow_dispatch:
+
+jobs:
+  test:
+    name: Test on ${{ matrix.os }} with Python ${{ matrix.python-version }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: ['3.10', '3.11', '3.12']
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Cache pip dependencies
+        uses: actions/cache@v4
+        with:
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-${{ hashFiles('setup.py') }}
+          restore-keys: |
+            ${{ runner.os }}-pip-
+
+      - name: Install & Model Test
+        run: |
+          python -m pip install --upgrade pip setuptools wheel
+          pip install -e .
+          python tests/model_test.py
+
+
+      - name: Install & Model/funasr Test
+        run: |
+          pip install -e .[test]
+          python tests/funasr_model_test.py

README.md

@@ -61,7 +61,7 @@ omnisense benchmark -s --num-workers 4 --device-id 0 --batch-size 16 --textnorm
 Set up pre-commit hooks:
 
 ```
-pip install pre-commit==3.6.0
+pip install pre-commit==4.2.0
 pre-commit install
 ```
 

setup.py

@@ -20,14 +20,17 @@
     readme="README.md",
     python_requires=">=3.8",
     install_requires=[
+        "kaldi-native-fbank",
         "modelscope>=1.18.0",
         "funasr>=1.1.6",
         "lhotse>=1.24.2",
         "kaldialign",
         "torch",
+        "torchaudio",
     ],
     extras_require={
         "test": [
+            "onnxscript",
             "funasr_onnx==0.4.1",
         ],
     },
@@ -42,6 +45,7 @@
         "Programming Language :: Python :: 3.11",
         "Programming Language :: Python :: 3.12",
         "Programming Language :: Python :: 3.13",
+        "Programming Language :: Python :: 3.14",
         "Intended Audience :: Science/Research",
         "Operating System :: POSIX :: Linux",
         "Operating System :: MacOS :: MacOS X",

tests/funasr_model_test.py

@@ -0,0 +1,44 @@
+import time
+
+import torch
+from funasr_onnx import SenseVoiceSmall
+from funasr_onnx.utils.postprocess_utils import rich_transcription_postprocess
+
+from omnisense import OmniSenseVoiceSmall
+
+model_dir = "iic/SenseVoiceSmall"
+model = SenseVoiceSmall(model_dir, batch_size=10, quantize=False, device_id=1)
+
+# inference
+wav_or_scp = ["tests/data/example.wav"]
+for textnorm in ["woitn", "withitn"]:
+    print(f"\n====== Text normalization: {textnorm} ======")
+    start_time = time.time()
+    res = model(wav_or_scp, language="auto", textnorm=textnorm)
+    print(f"Time cost: {time.time() - start_time:.2f}s")
+
+    print(res)
+    print([rich_transcription_postprocess(i) for i in res])
+
+
+model_dir = "iic/SenseVoiceSmall"
+model = OmniSenseVoiceSmall(model_dir, quantize=False, device_id=0 if torch.cuda.is_available() else -1)
+
+
+# inference
+wav_or_scp = ["tests/data/Laughter.wav", "tests/data/Cry.wav", "tests/data/Applause.wav"]
+for textnorm in ["woitn", "withitn"]:
+    print(f"\n====== Text normalization: {textnorm} ======")
+    start_time = time.time()
+    res = model.transcribe(wav_or_scp, language="auto", textnorm=textnorm, timestamps=True)
+    print(f"Time cost: {time.time() - start_time:.2f}s")
+    print(res)
+
+
+wav_or_scp = ["tests/data/example.wav"]
+for textnorm in ["woitn", "withitn"]:
+    print(f"\n====== Text normalization: {textnorm} ======")
+    start_time = time.time()
+    res = model.transcribe(wav_or_scp, language="auto", textnorm=textnorm, timestamps=True)
+    print(f"Time cost: {time.time() - start_time:.2f}s")
+    print(res)

tests/model_test.py

@@ -1,25 +1,8 @@
 import time
 
 import torch
-from funasr_onnx import SenseVoiceSmall
-from funasr_onnx.utils.postprocess_utils import rich_transcription_postprocess
-
-from omnisense import OmniSenseVoiceSmall, OmniTranscription
-
-model_dir = "iic/SenseVoiceSmall"
-model = SenseVoiceSmall(model_dir, batch_size=10, quantize=False, device_id=1)
-
-# inference
-wav_or_scp = ["tests/data/example.wav"]
-for textnorm in ["woitn", "withitn"]:
-    print(f"\n====== Text normalization: {textnorm} ======")
-    start_time = time.time()
-    res = model(wav_or_scp, language="auto", textnorm=textnorm)
-    print(f"Time cost: {time.time() - start_time:.2f}s")
-
-    print(res)
-    print([rich_transcription_postprocess(i) for i in res])
 
+from omnisense import OmniSenseVoiceSmall
 
 model_dir = "iic/SenseVoiceSmall"
 model = OmniSenseVoiceSmall(model_dir, quantize=False, device_id=0 if torch.cuda.is_available() else -1)
@@ -42,4 +25,3 @@
     res = model.transcribe(wav_or_scp, language="auto", textnorm=textnorm, timestamps=True)
     print(f"Time cost: {time.time() - start_time:.2f}s")
     print(res)
-