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
fix: preserve scripts in fuzzy hotword correction (#3798)
zhifu gaodd00b0e42261
diff --git a/.github/workflows/test-text-postprocessing.yml b/.github/workflows/test-text-postprocessing.ymlnew file mode 100644index 000000000..e76358e8e--- /dev/null+++ b/.github/workflows/test-text-postprocessing.yml@@ -0,0 +1,63 @@+name: Validate text postprocessing++on:+ pull_request:+ paths:+ - "funasr/utils/postprocess_hotwords.py"+ - "funasr/utils/postprocess_utils.py"+ - "runtime/python/onnxruntime/funasr_onnx/utils/postprocess_utils.py"+ - "runtime/python/libtorch/funasr_torch/utils/postprocess_utils.py"+ - "tests/test_postprocess_hotwords.py"+ - "tests/test_rich_transcription_postprocess.py"+ - ".github/workflows/test-text-postprocessing.yml"+ push:+ branches: [main]+ paths:+ - "funasr/utils/postprocess_hotwords.py"+ - "funasr/utils/postprocess_utils.py"+ - "runtime/python/onnxruntime/funasr_onnx/utils/postprocess_utils.py"+ - "runtime/python/libtorch/funasr_torch/utils/postprocess_utils.py"+ - "tests/test_postprocess_hotwords.py"+ - "tests/test_rich_transcription_postprocess.py"+ - ".github/workflows/test-text-postprocessing.yml"++permissions:+ contents: read++jobs:+ text-postprocessing:+ runs-on: ubuntu-24.04+ timeout-minutes: 10+ strategy:+ fail-fast: false+ matrix:+ python: ["3.10", "3.12"]+ steps:+ - uses: actions/checkout@v4+ with:+ ref: ${{ github.event.pull_request.head.sha || github.sha }}+ - uses: actions/setup-python@v5+ with:+ python-version: ${{ matrix.python }}+ - name: Install text test dependencies+ run: python -m pip install pytest==8.1.1 pypinyin==0.55.0 rapidfuzz==3.14.5+ - name: Record source and dependencies+ run: |+ git rev-parse HEAD | tee source-commit.txt+ python -m pip freeze | tee resolved-dependencies.txt+ python -m pip check+ python -c "import pypinyin, rapidfuzz"+ - name: Check hotword and rich transcription behavior without model weights+ run: |+ python -m pytest -q \+ tests/test_postprocess_hotwords.py \+ tests/test_rich_transcription_postprocess.py \+ --junitxml=text-postprocessing-results.xml+ - uses: actions/upload-artifact@v4+ if: always()+ with:+ name: text-postprocessing-python-${{ matrix.python }}+ path: |+ source-commit.txt+ resolved-dependencies.txt+ text-postprocessing-results.xmldiff --git a/funasr/utils/postprocess_hotwords.py b/funasr/utils/postprocess_hotwords.pyindex 5ff6183fa..a86a2013e 100644--- a/funasr/utils/postprocess_hotwords.py+++ b/funasr/utils/postprocess_hotwords.py@@ -364,6 +364,11 @@ class PostprocessHotwordMatcher:# replace a Chinese character read "ai".if (segment_latin and not target_latin) or (segment_han and not target_han):continue+ # Cross-script correction needs a complete homophone:+ # "ai zhushou" can become "AI助手", but "duan" must+ # not gain the missing Latin prefix in "C端".+ if target_latin and not segment_latin and segment_py != target_py:+ continuescore = self._fuzz.ratio(segment_py, target_py) / 100.0if score >= self.threshold:candidates.append(diff --git a/tests/test_postprocess_hotwords.py b/tests/test_postprocess_hotwords.pyindex 330219d5e..6dea9e218 100644--- a/tests/test_postprocess_hotwords.py+++ b/tests/test_postprocess_hotwords.py@@ -138,6 +138,23 @@ class TestPostprocessHotwordMatcher(unittest.TestCase):matcher = PostprocessHotwordMatcher(fuzzy_targets=["AI助手"])self.assertEqual(matcher.apply_text("打开爱助手")[0], "打开AI助手")+ def test_fuzzy_mixed_target_requires_complete_cross_script_homophone(self):+ try:+ import pypinyin # noqa: F401+ import rapidfuzz # noqa: F401+ except ImportError:+ self.skipTest("pypinyin and rapidfuzz are required for fuzzy tests")++ matcher = PostprocessHotwordMatcher(fuzzy_targets=["C端"])+ for text in ["这个端口", "端", "端口2024", "C端产品"]:+ with self.subTest(text=text):+ self.assertEqual(matcher.apply_text(text), (text, []))+ self.assertEqual(matcher.apply_text("C段产品")[0], "C端产品")++ matcher = PostprocessHotwordMatcher(fuzzy_targets=["AI助手"])+ self.assertEqual(matcher.apply_text("打开爱助手")[0], "打开AI助手")+ self.assertEqual(matcher.apply_text("打开AI助守")[0], "打开AI助手")+def test_missing_fuzzy_dependency_raises(self):with mock.patch.object(postprocess_hotwords,