Silero VAD Plugin

Plugin implementation for Voice Activity Detection using Silero VAD with SQLite result caching.

Configuration

The SileroVADConfig dataclass defines all configurable parameters for the VAD model.

---

SileroVADConfig

def SileroVADConfig(
    threshold:float=0.5, min_speech_duration_ms:int=250, min_silence_duration_ms:int=100, speech_pad_ms:int=30,
    sampling_rate:int=16000, use_onnx:bool=True
)->None:

Configuration for Silero VAD parameters.

Plugin Implementation

The SileroVADPlugin implements MediaAnalysisPlugin to provide voice activity detection.

---

SileroVADPlugin

def SileroVADPlugin(
    
):

Voice Activity Detection plugin using Silero VAD.

Testing the Plugin

# Test basic functionality
plugin = SileroVADPlugin()

# Check availability
print(f"Silero VAD available: {plugin.is_available()}")
print(f"Plugin name: {plugin.name}")
print(f"Plugin version: {plugin.version}")
print(f"Supported media types: {plugin.supported_media_types}")
print(f"Config class: {plugin.config_class.__name__}")

Silero VAD available: True
Plugin name: silero-vad
Plugin version: 1.0.0
Supported media types: ['audio', 'video']
Config class: SileroVADConfig

# Test configuration dataclass
from dataclasses import fields

print("Configuration fields:")
for f in fields(SileroVADConfig):
    title = f.metadata.get(SCHEMA_TITLE, f.name)
    default = f.default
    print(f"  {title}: {default}")

Configuration fields:
  Threshold: 0.5
  Min Speech Duration (ms): 250
  Min Silence Duration (ms): 100
  Speech Padding (ms): 30
  Sampling Rate: 16000
  Use ONNX: True

# Test configuration validation
test_configs = [
    ({"threshold": 0.5}, "Valid config"),
    ({"threshold": 1.5}, "Threshold out of range"),
    ({"sampling_rate": 44100}, "Invalid sampling rate"),
]

for config, description in test_configs:
    try:
        test_cfg = dict_to_config(SileroVADConfig, config, validate=True)
        print(f"{description}: Valid=True")
    except ValueError as e:
        print(f"{description}: Valid=False")
        print(f"  Error: {str(e)[:80]}")

Valid config: Valid=True
Threshold out of range: Valid=False
  Error: threshold: 1.5 is greater than maximum 1.0
Invalid sampling rate: Valid=False
  Error: sampling_rate: 44100 is not one of [8000, 16000]

# Test get_config_schema for UI generation
import json

schema = plugin.get_config_schema()
print("JSON Schema for SileroVADConfig:")
print(f"  Name: {schema['name']}")
print(f"  Properties count: {len(schema['properties'])}")
print(f"\nSample properties:")
print(json.dumps({k: v for k, v in list(schema['properties'].items())[:3]}, indent=2))

JSON Schema for SileroVADConfig:
  Name: SileroVADConfig
  Properties count: 6

Sample properties:
{
  "threshold": {
    "type": "number",
    "title": "Threshold",
    "description": "Speech probability threshold (0.0 - 1.0). Higher values reduce false positives.",
    "minimum": 0.0,
    "maximum": 1.0,
    "default": 0.5
  },
  "min_speech_duration_ms": {
    "type": "integer",
    "title": "Min Speech Duration (ms)",
    "description": "Segments shorter than this will be ignored.",
    "minimum": 0,
    "default": 250
  },
  "min_silence_duration_ms": {
    "type": "integer",
    "title": "Min Silence Duration (ms)",
    "description": "Silence shorter than this will not split segments.",
    "minimum": 0,
    "default": 100
  }
}