pysilero-vad 2.0.1__tar.gz → 2.1.0__tar.gz

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: pysilero_vad
-Version: 2.0.1
+Version: 2.1.0
 Summary: Pre-packaged voice activity detector using silero-vad
 Author-email: Michael Hansen <mike@rhasspy.org>
 License: MIT
@@ -21,7 +21,8 @@ Requires-Python: >=3.8.0
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
 Requires-Dist: onnxruntime<2,>=1.18.0
-Requires-Dist: numpy<2
+Requires-Dist: numpy
+Dynamic: license-file
 
 # pySilero VAD
 

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name        = "pysilero_vad"
-version     = "2.0.1"
+version     = "2.1.0"
 license     = {text = "MIT"}
 description = "Pre-packaged voice activity detector using silero-vad"
 readme      = "README.md"
@@ -27,7 +27,7 @@ classifiers = [
 requires-python = ">=3.8.0"
 dependencies = [
     "onnxruntime>=1.18.0,<2",
-    "numpy<2"
+    "numpy",
 ]
 
 [project.urls]
@@ -41,3 +41,6 @@ include-package-data = true
 [tool.setuptools.packages.find]
 include = ["pysilero_vad"]
 exclude = ["tests", "tests.*"]
+
+[tool.setuptools.package-data]
+pysilero_vad = ["models/silero_vad.onnx"]

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/pysilero_vad/__init__.py

@@ -73,6 +73,17 @@ class SileroVoiceActivityDetector:
 
         audio_array = np.frombuffer(audio, dtype=np.int16).astype(np.float32) / _MAX_WAV
 
+        return self.process_array(audio_array)
+
+    def process_array(self, audio_array: np.ndarray) -> float:
+        """Return probability of speech [0-1] in a single audio chunk.
+
+        Audio *must* be 512 float samples [0-1] of 16Khz mono.
+        """
+        if len(audio_array) != _CHUNK_SAMPLES:
+            # Window size is fixed at 512 samples in v5
+            raise InvalidChunkSizeError
+
         # Add batch dimension and context
         audio_array = np.concatenate(
             (self._context, audio_array[np.newaxis, :]), axis=1

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/pysilero_vad.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: pysilero_vad
-Version: 2.0.1
+Version: 2.1.0
 Summary: Pre-packaged voice activity detector using silero-vad
 Author-email: Michael Hansen <mike@rhasspy.org>
 License: MIT
@@ -21,7 +21,8 @@ Requires-Python: >=3.8.0
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
 Requires-Dist: onnxruntime<2,>=1.18.0
-Requires-Dist: numpy<2
+Requires-Dist: numpy
+Dynamic: license-file
 
 # pySilero VAD
 

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/pysilero_vad.egg-info/requires.txt

@@ -1,2 +1,2 @@
 onnxruntime<2,>=1.18.0
-numpy<2
+numpy

{pysilero_vad-2.0.1 → pysilero_vad-2.1.0}/tests/test_vad.py

@@ -2,8 +2,10 @@ import wave
 from pathlib import Path
 from typing import Union
 
+import numpy as np
 import pytest
-from pysilero_vad import SileroVoiceActivityDetector, InvalidChunkSizeError
+
+from pysilero_vad import InvalidChunkSizeError, SileroVoiceActivityDetector
 
 _DIR = Path(__file__).parent
 
@@ -29,12 +31,16 @@ def test_speech() -> None:
     vad = SileroVoiceActivityDetector()
     assert any(p >= 0.5 for p in vad.process_chunks(_load_wav(_DIR / "speech.wav")))
 
+
 def test_invalid_chunk_size() -> None:
     """Test that chunk size must be 512 samples."""
     vad = SileroVoiceActivityDetector()
 
     # Should work
     vad(bytes(SileroVoiceActivityDetector.chunk_bytes()))
+    vad.process_array(
+        np.zeros(SileroVoiceActivityDetector.chunk_samples(), dtype=np.float32)
+    )
 
     # Should fail
     with pytest.raises(InvalidChunkSizeError):
@@ -42,3 +48,13 @@ def test_invalid_chunk_size() -> None:
 
     with pytest.raises(InvalidChunkSizeError):
         vad(bytes(SileroVoiceActivityDetector.chunk_bytes() // 2))
+
+    with pytest.raises(InvalidChunkSizeError):
+        vad.process_array(
+            np.zeros(SileroVoiceActivityDetector.chunk_samples() * 2, dtype=np.float32)
+        )
+
+    with pytest.raises(InvalidChunkSizeError):
+        vad.process_array(
+            np.zeros(SileroVoiceActivityDetector.chunk_samples() // 2, dtype=np.float32)
+        )