SoundAnalyser Class Reference

Main class that sets up and runs a pipeline. More...

#include <soundanalyser.h>

Inheritance diagram for SoundAnalyser:

List of all members.

Public Member Functions
	SoundAnalyser (bool live=false)
	Sets up a new sound analyser.
	~SoundAnalyser ()
	Destructor.
void	resetExtractors ()
	Extractors management.
void	cleanExtractors ()
	Clean the data in the extractors.
void	registerExtractor (string name, FeatureExtractor *extr, bool used=true, bool drawLines=true)
	Adds a new extractor.
FeatureExtractor *	getExtractor (string name)
	Find an extractor.
bool	setupPipeline (string filename)
	Setup.
void	setNormalization (float min, float max)
	Set up unified bounds for all features.
bool	compute (string url)
	Compute the features of a given file.
void	waitComputed ()
	Wait up to the end of the computation.
unsigned int	nbFeatures ()
	Features management : sizes.
unsigned int	nbElems (unsigned int n)
	Return the number of elements of the nth feature.
unsigned int	nbSamples ()
	Return the number of sample we've been computing.
unsigned int	dimension ()
	Get the dimension : sum of all the nbElems(i)
unsigned int	realDimension ()
	Get the real dimension : sum of all the nbElems we care (the values which are actually used for detection)
double **	features (unsigned int n)
	Return the array of features for the nth sample.
void	clearFeatures ()
	Clears the features.
void	cleanOldFeatures (unsigned int newestCount)
	Removes the old features so that the newest remain.
bool	isUsed (unsigned int index)
	Features management : individual properties.
bool	isDrawnWithLines (unsigned int index)
	Should this feature be drawn as a lines or a shade of grey ?
string	name (unsigned int n)
	Get the name of the nth feature.
int	getFeatureByName (string name)
	Get the id of the feature named so (-1 if not found)
void	useBuffer ()
	Callbacks.
void	sequenceEnds ()
	End the computation.
virtual void	useFeatures ()
	Callback called when the buffer has been used (the features are ready to be used)
unsigned int	samplingFrequency ()
	Mirrors.

Additional Inherited Members
Protected Member Functions inherited from StreamPlayer
	StreamPlayer (bool live=false, bool verbose=true)
	Default constructor. Starts VLC instance.
	~StreamPlayer ()
	Default destructor. Frees memory and closes VLC.
string	url ()
	Returns URL of the currently playing stream.
void	setUrl (string url)
	Defines the URL of the stream to play.
void	setOverlapping (float factor)
	Turns on / off audio chunk overlapping.
void	setChunkSize (int size)
	Changes the audio chunk size.
int	chunkSize ()
	Get the current chunk size.
void	play ()
	Plays the media.
libvlc_time_t	playingTime ()
	Returns the playing time.
libvlc_time_t	totalTime ()
	Returns the total time (played + to be played)
void	stop ()
	Stops playing.
bool	isLive ()
	Are we playing a live stream ?
void	setVolume (int vol)
	Changes the volume (between 0 and 100)
int	volume ()
	Gets the current volume.
virtual void	sequenceStarts ()
	Callback called when we start playing a file.
void	watch ()
	Watching thread.
uint16_t	buffer (int i)
	Buffer acces (for the user)
int	bufferSize ()
	Current buffer size (internal)
void	fillBuffer (uint16_t value)
	Fill buffer with a new value (internal)
void	flushBuffer ()
	Flush buffer (just keep the overlapping part)
Static Protected Member Functions inherited from StreamPlayer
static uint16_t *	convert8to16 (const uint8_t *source, int size)
	Converts an array of uint8_t to another array of uint16_t (assuming the values are coded on two bytes) The returned pointer has to be deleted by the user.
static uint16_t *	average (uint16_t *source, int size, int passes, int scale=1)
	Computes the average value of the stream, on a given frequency The returned pointer has to be deleted by the user.
static void	reduce (uint16_t *source, uint16_t *dest, int size, int passes, int scale=1)
	Reduces the stream, keeping only the extreme values.
static void	addOffset (uint16_t *source, uint16_t *dest, int size, int offset)
	Adds an offset to each value of the array.
static int	pow2 (int n)
	Get 2^n.
Protected Attributes inherited from StreamPlayer
std::deque< uint16_t >	mBuffer
	Feature extraction : those variables need to be public (I know, I can write accessors...)
QMutex	mLock
char *	mAudioData
unsigned int	mAudioDataSize
unsigned int	mFrequency
bool	mVerbose

---

Detailed Description

Main class that sets up and runs a pipeline.

A pipeline is a network of features and filters connected to the input stream.

Some explanations about the behaviour of the class :

Features, elements

A feature is a set of values computed by an algorithm. The number of values it outputs on a single audio chunk is the number of elements.

Used features, Dimension vs real dimension

In the pipeline, some features are labelled as "used" : it means they are sent to the application to be used for classification, visualization, etc. The others are not outputted directly and can only be used as input for filters. The "real dimension" is the number of used features (x their number of elements). The "dimension" is the number of computed features and filters (again, x their number of elements).

Note : the feature named _spectrum is usually not labelled as "used" but is displayed by FeatureDrawer.

Normalization

Forces the outputs of the features (and filters) to be in a given range.

---

Member Function Documentation

void SoundAnalyser::cleanExtractors

(

)

Clean the data in the extractors.

Clears the data in the extractors.

double ** SoundAnalyser::features

(

unsigned int

n

)

Return the array of features for the nth sample.

Return the array of features at a given time.

bool SoundAnalyser::isUsed

(

unsigned int

index

)

Features management : individual properties.

Is this feature used for detection ?

unsigned int SoundAnalyser::nbFeatures ( )

inline

Features management : sizes.

Return the number of different features we computed

void SoundAnalyser::resetExtractors

(

)

Extractors management.

Unregisters all the extractors.

Unregisters all the extractors

unsigned int SoundAnalyser::samplingFrequency ( )

inline

Mirrors.

Get the sampling frequency

void SoundAnalyser::setNormalization	(	float	min,
		float	max
	)

Set up unified bounds for all features.

Enable feature normalization.

bool SoundAnalyser::setupPipeline

(

string

filename

)

Setup.

Set up from an XML file.

Set up the extractors from an XML file

Todo:

Coding style to be fixed (see tpop, 15 / 26)

Reimplemented in CorpusBuilder.

void SoundAnalyser::useBuffer ( )

virtual

Callbacks.

Handle audio chunks

Reimplemented from StreamPlayer.

---

The documentation for this class was generated from the following files:

• include/core/soundanalyser.h
• src/core/soundanalyser.cpp