Dependent software¶

HARKTOOL5-GUI depends on following softwares. Please install them before you install HARKTOOL5-GUI.

• node.js (An application server)
  • Windowsユーザ: https://nodejs.org/en/download/releases/ からv10.xをダウンロードする。
  • Ubuntuユーザ: https://github.com/nodesource/distributions からv10.xをインストールする。
• Google Chrome

Install¶

Step 1: Open the HARKTOOL5-GUI¶

For Windows users¶

スタートメニューの HARK > HARKTool をクリックする。直後にChrome ブラウザが開く。

Step 1: The initial screen

Step 2: Create the microphone array coordinate file¶

If you want to use the old microphone location information file, see Migration of microphone location information file.

1. Click “Microphone array positions” > “Create”.
2. Select the coordinate model.
3. Input parameters. Distance unit is meter.

General parameters¶

Position Type

Fixed “microphone”

Output File

Specifies the Output File name

4. Clicking the “Create” button at the bottom of the screen creates a microphone array coordinate file and displays the 3D graph on the right.
5. Click on the “Go back to the top page” link or the “HARKTOOL5” title on the screen and return to the top screen.
6. The created microphone array coordinate file can be downloaded by clicking “Microphone array positions”> “Download”.

Step 2: After creating microphone array coordinate file

Step 3: Create sound source coordinate file¶

If you want to use the old TSP response list file, see Migration of TSP response list file.

Also, if you want to use the old impulse response list file, see Migration of impulse response list file.

1. Click “Sound source positions” > “Create”.
2. Select the coordinate model.
3. Input parameters.

Neighbors parameters¶

Algorithm

Algorithm type. Fixed “NearestNeighbor”

Maximum Number

Maximum number of neighbors each position can have

Threshold

Threshold for neighbor decision (unit m)

4. Clicking the “Create” button at the bottom of the screen creates a sound source coordinate file and displays the 3D graph on the right side.
5. Click on the “Go back to the top page” link or the “HARKTOOL5” title on the screen and return to the top screen.
6. The created sound source coordinate file can be downloaded by clicking “Sound source positions”> “Download”.

Step 3: After creating sound source coordinate file

Step 4: Create the transfer function file¶

When the microphone array coordinate file and the sound source coordinate file are prepared, a transfer function file can be created.

1. Click “Transfer function” > “Create”.
2. Input Config (General Parameters) when it is needed.

Sync. Add Num

The count of consecutive replays of TSP signal at the time of TSP signal recording

Original TSP Path

The filepath of the TSP signal of one cycle used for recording

TSP Offset

Start position to use the recorded file for transfer function calculation (unit sample 0 origin)

Peak Search From

Start position to search for a peak when calculating transfer function (unit sample 1 origin)

Peak Search To

End position to search for a peak when calculating transfer function (unit sample 1 origin)

FFT Length

The number of bins in the discrete frequency expression of the transfer function

Sampling Rate

Sampling frequency of the transfer function

Signal Max

The value of the maximum amplitude

TSP Length

The length of one recorded TSP signal

The following figure shows the correspondence between the waveform of Typical TSP recording (17 times continuous TSP playback and its synchronized recording by wios) and the Config parameter.

3. Input Transfer Function Parameters when it is needed.

Output File Name (zip)

Specifies the output file name of the Transfer Function

TF Calculation Type

Select method to calculate transfer function (Depending on the Position Type chosen in Step 3, some options may not be displayed.)

• Use Impulse Response Files: Calculate transfer function from impulse response file
• Use TSP Record Files: Calculate transfer function from TSP recorded file
• Use Mouth TSP Record Files: Calculate transfer function from TSP recorded file by Transfer Function Estimation Using Complex Regression Model (New feature of HARKTOOL5)
• Use Geometric Calculation: Calculate transfer function by geometric calculation

TF Type

Whether to include for localization or separation for transfer function file

Direct Sound Length

Number of samples consider as direct sound

Reverb Sound Length

Number of samples consider as reverberation

Normalize SRC axis

Whether to normalize the sound source axis of the transfer function

Normalize MIC axis

Whether to normalize the microphone channel axis of the transfer function

Normalize FREQ axis

Whether to normalize the frequency axis of the transfer function

Reset Mic Channels

Whether or not to reset the transfer functions microphone channel selected in “Create sound source coordinate file”. It will reset the channels selected starting from 0.

4. Clicking the “Create” button at the bottom of the screen creates a transfer function file.
5. Click on the “Go back to the top page” link or the “HARKTOOL5” title on the screen and return to the top screen.

Step 4: After creating transfer function file

6. By clicking “Transfer function” > “Visualize”, the created transfer function file can be confirmed in 3D graph.

TF type

Select transfer function(localization/separation) to be displayed

Domain

Select X axis (frequency/time) of graph

Z Axis

Select the type of Z axis (only in frequency domain)

• Amplitude: Display amplitude spectrum
• Amplitude in dB: Display amplitude spectrum in decibels
• Phase: Display phase spectrum

Display transfer function

7. By clicking “Transfer function” > “Download”, the created transfer function file can be download.

Algorithm¶

The input signal S, the impulse response H, and the output signal X are expressed in the frequency domain by the following numerical formula:

Since it is the same sound field for different time frame \(t = 1 \cdots N\), it is expressed by the following determinant:

Migration of microphone location information file¶

1. Execute python conv4mic.py <microphone location information file> <output file>.

e.g.) python conv4mic.py miclist.xml microphones.xml

2. The output file can be loaded from “Microphone array positions” > “Upload” of HARKTOOL5-GUI.

Migration of TSP response list file¶

1. Execute python conv4src.py <TSP response list file> <output file>.

e.g.) python conv4src.py tsplist.xml source.xml

2. The output file can be loaded from “Sound source positions”> “Upload” of HARKTOOL5-GUI.

Migration of impulse response list file¶

1. Execute python conv4src.py <impulse response list file> <output file>.

e.g.) python conv4src.py implist.xml source.xml

2. The output file can be loaded from “Sound source positions”> “Upload” of HARKTOOL5-GUI.
3. Since the `` flt2multiwav.txt`` file is also output, the flt file is converted to a wav file by executing the command output in the file.