edu.gtts.sautrela.sp
Class VoiceActivityDetector

java.lang.Object
  edu.gtts.sautrela.engine.AbstractDataProcessor
      edu.gtts.sautrela.sp.VoiceActivityDetector

All Implemented Interfaces:

DataProcessor, java.beans.BeanInfo, java.io.Serializable

public class VoiceActivityDetector
extends AbstractDataProcessor

An energy based Voice Activity Detector. Adds a voiced/unvoiced boolean mask to a DataStream. Frames with an energy level lower than max(energy)-threshold (in dB), are considered to be unvoiced. A boolean array representing the voice mask is added to the DataBeginStream by means of a propperty called VADFlag.

Author:

mpenagar

See Also:

DataStreamBegin.getProperty(String), Serialized Form

Field Summary

static double

DEFAULT_THRESHOLD default energy threshold = 30dB

Fields inherited from interface java.beans.BeanInfo

ICON_COLOR_16x16, ICON_COLOR_32x32, ICON_MONO_16x16, ICON_MONO_32x32

Constructor Summary

VoiceActivityDetector()
Creates a new instance of VoiceActivityDetector

Method Summary

void	editBeanInfo(java.beans.BeanInfo info)
double	getThreshold() Getter for property threshold.
static void	main(java.lang.String[] args)
void	process(Buffer in, Buffer out) Processes the input Data.
void	setThreshold(double threshold) Setter for property threshold.
static boolean[]	transformVADFlag(boolean[] flag, int windowSize, int shift, int leftSkip, int rightSkip, int minVoiceCount) Utility method for converting VAD flags.

Methods inherited from class edu.gtts.sautrela.engine.AbstractDataProcessor

getAdditionalBeanInfo, getBeanDescriptor, getDefaultEventIndex, getDefaultPropertyIndex, getEventSetDescriptors, getIcon, getMethodDescriptors, getName, getPropertyDescriptors, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

DEFAULT_THRESHOLD

public static final double DEFAULT_THRESHOLD

default energy threshold = 30dB

See Also:

Constant Field Values

Constructor Detail

VoiceActivityDetector

public VoiceActivityDetector()

Creates a new instance of VoiceActivityDetector

Method Detail

getThreshold

public double getThreshold()

Getter for property threshold.

Returns:

Value of property threshold.

setThreshold

public void setThreshold(double threshold)

Setter for property threshold.

Parameters:

threshold - New value of property threshold.

transformVADFlag

public static boolean[] transformVADFlag(boolean[] flag,
                                         int windowSize,
                                         int shift,
                                         int leftSkip,
                                         int rightSkip,
                                         int minVoiceCount)

Utility method for converting VAD flags. DataProcessors that break the synchronization between the VAD flags and stream data (for example Deltas and SDC) must transform the VADFlag too. This utility method performs a sliding windowing transformation of an existing VADflag array:

outFlag(i) = numberOfTrue(window(i)) >= minVoiceCount ? true : false
The returned array length is:

outFlag.length = (flag.length - leftSkip - rightSkip - windowSize + shift) / shift

Parameters:

flag - The source VADflag array.

windowSize - The sliding window size

shift - The sliding window shift

leftSkip - The number of values to skipt at the left

rightSkip - The number of values to skipt at the rigth

minVoiceCount - Minimum number of voiced values in the window to generate a voiced output

process

public void process(Buffer in,
                    Buffer out)
             throws DataProcessorException

Processes the input Data.

Parameters:

in - Input Buffer

out - Output Buffer

Throws:

DataProcessorException

editBeanInfo

public void editBeanInfo(java.beans.BeanInfo info)

Overrides:

editBeanInfo in class AbstractDataProcessor

main

public static void main(java.lang.String[] args)
                 throws java.lang.Exception

Throws:

java.lang.Exception