HARK version 1.1.0 Document : ChannelSelector

6.7.1 ChannelSelector

Outline of the node

This node extracts data from specified channels, in specific order, from multichannel speech waveforms.

Necessary files

No files are required.

Usage

When to use

This node is used to delete unnecessary channels from multichannel speech input waveforms, changing the alignment of channels and copying channels.

Typical connection

$\includegraphics[]{fig/modules/ChannelSelector}$

Figure 6.75: Example of a typical connection of ChannelSelector

Figure 6.75 shows an example of a typical connection. Several, but not all, channels will be extracted from the multichannel audio files in this network. The main input sources are AudioStreamFromMic and AudioStreamFromWave , and the main output sources are SaveRawPCM , MultiFFT .

Input-output and property of the node

Input

INPUT: Matrix<float> type. Multichannel speech waveform data.

Output

OUTPUT: Matrix<float> type. Multichannel speech waveform data.

Parameter

Parameters of ChannelSelector

Parameter name	Type	Default value	Unit	Description
SELECTOR	`Object`	`Vector< int >`		Designate the number of channels to output

SELECTOR

Object type. No default value. Designate the numbers of the channels to be used. Channel numbers begin with 0. For example, to use only Channels 2, 3 and 4 of the five channels (0-4), use the values shown in (1) below; when swapping Channels 3 and 4, use the values shown in (2).

(1): <Vector<int> 2 3 4>
(2): <Vector<int> 2 4 3>

Details of the node

This node extracts only the speech waveform data from the channels designated in the $N\times M$ type matrix (Matrix ) of input and output data as a new $N’\times M$ type matrix. Here, $N$ indicates the number of input channels and $N’$ indicates the number of output channels.