HARK version 1.1.0 Document : Sound source separation transfer function for GHDSS

5.1.2 Sound source separation transfer function for GHDSS

The sound source separation transfer function, similar to the source localization transfer function, also indicates a transfer function from a sound source to a microphone array and is expressed as a file in HGTF format with GTF sub chunks. The sound source separation transfer function file is designated in TF_CONJ_FILENAME when the property TF_CONJ of GHDSS is true. For details of usage, see GHDSS . Table 5.1.2 shows the definition of the GTF sub chunk. The rows in memory of a multi-dimensional array are row major (same as in the C language).

Type	Value
char[4]	"GTF "
int	Size of chunk,
	8+12* $Ns$ +4+4* $Nf$ +8* $Ns$ * $nmic$ * $Nf$ [bytes]
short[2]	GTF version (major, minor)
int	Number of source positions (Ns)
Vector3[Ns]	Coordinate of source positions [m]
int	Number of frequency lines (Nf)
int[Nf]	Maps from an index to a frequency bin
`complex<float>` [Ns][nmic][Nf]	Transfer function

This file is created from a file in the impulse response format or from a microphone position and source location list file using harktool, which is a support tool. Details of the creation method are described in Section 7.1.