This is the runtime configuration, the main processing will be done in this class. More...

Public Member Functions
	adaptive_feedback_canceller_config (algo_comm_t &ac, const mhaconfig_t in_cfg, adaptive_feedback_canceller *afc)

	~adaptive_feedback_canceller_config ()

mha_wave_t *	process (mha_wave_t *MICsig)
	The process() method contains the actual AFC algorithm, the output is stored in LSsig_output. More...

void	insert ()
	Insert all AC-variables into the AC-space. More...

Private Attributes
const unsigned int	ntaps
	Length of the estimated filter. More...

const unsigned int	frames
	Length of a block in samples (fragsize) More...

const unsigned int	channels
	Number of channels. More...

const int	n_no_update_
	Number of blocks after startup where the feedback filter is not updated. More...

int	no_update_count
	Index counting no-update-blocks up to n_no_update_ to check in process() whether a filter update shall be performed or not. More...

const mha_real_t	fragsize
	Fragsize for computing delay_roundtrip and delay_update, defaults to the MHA's fragsize. More...

const mha_real_t	stepsize
	Normalized stepsize of the NLMS-Algorithm. More...

const mha_real_t	min_const
	Minimum constant to prevent division by zero in the NLMS-Algorithm. More...

MHASignal::waveform_t	forward_sig
	Copy of error signal that is channeled into the forward path processing. More...

MHASignal::waveform_t	LSsig_initializer
	Signal consisting of zeros, it is only used to initialize the loudspeaker signal (LSsig) for the first iteration (before forward_path_proc.process() is called for the first time). More...

mha_wave_t *	LSsig
	Pointer to the loudspeaker (LS) signal. More...

MHASignal::waveform_t	LSsig_output

MHASignal::delay_t	delay_forward_path
	Delay line for additional decorrelation in the forward path. More...

MHAParser::mhapluginloader_t &	forward_path_proc
	Pluginloader to load the plugins that represent the hearing aid processing in the forward path. More...

MHASignal::delay_t	delay_roundtrip
	Delay line equal to the measured roundtrip latency - fragsize. More...

MHASignal::delay_t	delay_update

std::vector< MHAFilter::filter_t >	FBfilter_estim

MHA_AC::waveform_t	FBfilter_estim_ac

MHASignal::waveform_t	FBsig_estim

MHASignal::waveform_t	ERRsig

MHA_AC::waveform_t	ERRsig_ac

bool	use_lpc_decorr

std::vector< MHAFilter::filter_t >	lpc_filter

MHASignal::waveform_t	white_LSsig
	Loudspeaker signal, whitened by filtering with LPC coefficients. More...

MHASignal::waveform_t	white_LSsig_smpl
	Single sample (for each channel) of white_LSsig that will be written to rb_white_LSsig next. More...

MHASignal::ringbuffer_t	rb_white_LSsig
	Ringbuffer containing the values used to determine the power of white_LSsig in each channel (see channels) and update FBfilter_estim. More...

std::vector< mha_real_t >	current_power

MHASignal::waveform_t	white_MICsig

MHASignal::waveform_t	white_FBsig_estim

MHASignal::waveform_t	white_ERRsig

bool	debug_mode
	When executing the code for debug purposes this flag can be set to true in order to capture variable states and provide them as AC variables to the user. More...

MHA_AC::waveform_t	current_power_ac

MHA_AC::waveform_t	estim_err_ac
	AC-variable publishing the state of 'estim_err' if debug_mode == true 'estim_err' is a variable local to this plugin's process() method. More...

Detailed Description

This is the runtime configuration, the main processing will be done in this class.

During runtime AC variables are published by this class, mainly for debbugging purposes.

Constructor & Destructor Documentation

◆ adaptive_feedback_canceller_config()

adaptive_feedback_canceller_config::adaptive_feedback_canceller_config	(	algo_comm_t &	ac,
		const mhaconfig_t	in_cfg,
		adaptive_feedback_canceller *	afc
	)

Parameters

ac	Algorithm communication variable space
in_cfg	MHA signal dimensions for this plugin
adaptive_feedback_canceller	Pointer to plugin interface instance, used to extract user configuration values.

◆ ~adaptive_feedback_canceller_config()

adaptive_feedback_canceller_config::~adaptive_feedback_canceller_config ( )

Member Function Documentation

◆ process()

mha_wave_t * adaptive_feedback_canceller_config::process ( mha_wave_t * MICsig )

The process() method contains the actual AFC algorithm, the output is stored in LSsig_output.

Parameters

MICsig The microphone signal which contains the target signal + the feedback signal. The input signal is not altered in place.

Returns: A pointer to LSsig_output which contains the loudspeaker signal that is actually channeled to the output.

◆ insert()

void adaptive_feedback_canceller_config::insert ( )

Insert all AC-variables into the AC-space.

Member Data Documentation

◆ ntaps

const unsigned int adaptive_feedback_canceller_config::ntaps

private

Length of the estimated filter.

◆ frames

const unsigned int adaptive_feedback_canceller_config::frames

private

Length of a block in samples (fragsize)

◆ channels

const unsigned int adaptive_feedback_canceller_config::channels

private

Number of channels.

This plugin assumes that the number of input channels is equal to the number of output channels. Each input is paired with an output and one pair is called a channel. This is necessary because the AFC needs input and output information to work.

◆ n_no_update_

const int adaptive_feedback_canceller_config::n_no_update_

private

Number of blocks after startup where the feedback filter is not updated.

◆ no_update_count

int adaptive_feedback_canceller_config::no_update_count

private

Index counting no-update-blocks up to n_no_update_ to check in process() whether a filter update shall be performed or not.

◆ fragsize

const mha_real_t adaptive_feedback_canceller_config::fragsize

private

Fragsize for computing delay_roundtrip and delay_update, defaults to the MHA's fragsize.

It is assumed that the soundcard's and the MHA's fragsize are equal. In special cases, the user can set this variable to an individual value.

◆ stepsize

const mha_real_t adaptive_feedback_canceller_config::stepsize

private

Normalized stepsize of the NLMS-Algorithm.

◆ min_const

const mha_real_t adaptive_feedback_canceller_config::min_const

private

Minimum constant to prevent division by zero in the NLMS-Algorithm.

◆ forward_sig

MHASignal::waveform_t adaptive_feedback_canceller_config::forward_sig

private

Copy of error signal that is channeled into the forward path processing.

◆ LSsig_initializer

MHASignal::waveform_t adaptive_feedback_canceller_config::LSsig_initializer

private

Signal consisting of zeros, it is only used to initialize the loudspeaker signal (LSsig) for the first iteration (before forward_path_proc.process() is called for the first time).

◆ LSsig

mha_wave_t* adaptive_feedback_canceller_config::LSsig

private

Pointer to the loudspeaker (LS) signal.

The destination of this pointer will be altered by the plugin loaded by the pluginloader in the forward path processing.

◆ LSsig_output

MHASignal::waveform_t adaptive_feedback_canceller_config::LSsig_output

private

◆ delay_forward_path

MHASignal::delay_t adaptive_feedback_canceller_config::delay_forward_path

private

Delay line for additional decorrelation in the forward path.

◆ forward_path_proc

MHAParser::mhapluginloader_t& adaptive_feedback_canceller_config::forward_path_proc

private

Pluginloader to load the plugins that represent the hearing aid processing in the forward path.

◆ delay_roundtrip

MHASignal::delay_t adaptive_feedback_canceller_config::delay_roundtrip

private

Delay line equal to the measured roundtrip latency - fragsize.

The roundtrip latency describes the timespan between playing back and receiving the same signal, including delays induced by soundcard buffering and the transducers. Can be measured via jack_iodelay. It is used on the loudspeaker signal LSsig before the filtering with FBfilter_estim, to achieve about the same temporal alignment to MICsig as the alignment of the target signal and the true feedback signal.

◆ delay_update

MHASignal::delay_t adaptive_feedback_canceller_config::delay_update

private

◆ FBfilter_estim

std::vector<MHAFilter::filter_t> adaptive_feedback_canceller_config::FBfilter_estim

private

◆ FBfilter_estim_ac

MHA_AC::waveform_t adaptive_feedback_canceller_config::FBfilter_estim_ac

private

◆ FBsig_estim

MHASignal::waveform_t adaptive_feedback_canceller_config::FBsig_estim

private

◆ ERRsig

MHASignal::waveform_t adaptive_feedback_canceller_config::ERRsig

private

◆ ERRsig_ac

MHA_AC::waveform_t adaptive_feedback_canceller_config::ERRsig_ac

private

◆ use_lpc_decorr

bool adaptive_feedback_canceller_config::use_lpc_decorr

private

◆ lpc_filter

std::vector<MHAFilter::filter_t> adaptive_feedback_canceller_config::lpc_filter

private

◆ white_LSsig

MHASignal::waveform_t adaptive_feedback_canceller_config::white_LSsig

private

Loudspeaker signal, whitened by filtering with LPC coefficients.

If use_lpc_decorr is false then this is just a copy of LSsig before the delays.

◆ white_LSsig_smpl

MHASignal::waveform_t adaptive_feedback_canceller_config::white_LSsig_smpl

private

Single sample (for each channel) of white_LSsig that will be written to rb_white_LSsig next.

For the filter estimation we have to use a ringbuffer with the size of filter_length to buffer white_LSsig. For each filter tap we have to update the buffer and it is only possible to update a ringbuffer with a whole waveform_t. That is why we have to use this single sample variable.

◆ rb_white_LSsig

MHASignal::ringbuffer_t adaptive_feedback_canceller_config::rb_white_LSsig

private

Ringbuffer containing the values used to determine the power of white_LSsig in each channel (see channels) and update FBfilter_estim.

The ringbuffer has a size equal to filter_length.

◆ current_power

std::vector<mha_real_t> adaptive_feedback_canceller_config::current_power

private

◆ white_MICsig

MHASignal::waveform_t adaptive_feedback_canceller_config::white_MICsig

private

◆ white_FBsig_estim

MHASignal::waveform_t adaptive_feedback_canceller_config::white_FBsig_estim

private

◆ white_ERRsig

MHASignal::waveform_t adaptive_feedback_canceller_config::white_ERRsig

private

◆ debug_mode

bool adaptive_feedback_canceller_config::debug_mode

private

When executing the code for debug purposes this flag can be set to true in order to capture variable states and provide them as AC variables to the user.

The following variables are captured: FBfilter_estim_ac, ERRsig_ac, current_power_ac, estim_err_ac

◆ current_power_ac

MHA_AC::waveform_t adaptive_feedback_canceller_config::current_power_ac

private

◆ estim_err_ac

MHA_AC::waveform_t adaptive_feedback_canceller_config::estim_err_ac

private

AC-variable publishing the state of 'estim_err' if debug_mode == true 'estim_err' is a variable local to this plugin's process() method.

It represents a part of the NLMS equation, namely the stepsize normalized by current_power times ERRsig.

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

Public Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ adaptive_feedback_canceller_config()

◆ ~adaptive_feedback_canceller_config()

Member Function Documentation

◆ process()

◆ insert()

Member Data Documentation

◆ ntaps

◆ frames

◆ channels

◆ n_no_update_

◆ no_update_count

◆ fragsize

◆ stepsize

◆ min_const

◆ forward_sig

◆ LSsig_initializer

◆ LSsig

◆ LSsig_output

◆ delay_forward_path

◆ forward_path_proc

◆ delay_roundtrip

◆ delay_update

◆ FBfilter_estim

◆ FBfilter_estim_ac

◆ FBsig_estim

◆ ERRsig

◆ ERRsig_ac

◆ use_lpc_decorr

◆ lpc_filter

◆ white_LSsig

◆ white_LSsig_smpl

◆ rb_white_LSsig

◆ current_power

◆ white_MICsig

◆ white_FBsig_estim

◆ white_ERRsig

◆ debug_mode

◆ current_power_ac

◆ estim_err_ac