[Back to top of RWCP Sound Scene Database]

RWCP Sound Scene Database 
HRTF of interactive robot

Tetsunori Kobayashi Laboratory, School of Science and Engineering, Waseda University
http://www.tk.elec.waseda.ac.jp/

The latest information on this page is located at 
http://www.tk.elec.waseda.ac.jp/robot-hrtf/  

1. Outlines of measurement

We measured an HRTF of interactive robot ROBITA developed in our laboratory.  In the robot, two microphones can be installed on both side faces of the head. Measurement was carried out for both directional microphone (ATM-15a) and non-directional  microphone (ATM14a.) The major sizes of the ROBITA head are presented in Figs. 3 and 4. The latest information on ROBITA may be obtained from Kobayashi Tetsunori Laboratory, Department of Science and Technology, Waseda University(http://www.tk.elec.waseda.ac.jp.)

Fig. 1. Interactive robot ROBITA
ROBITA
Fig. 2. Interactive robot ROBITA
 (head part)
head part of ROBITA
Fig. 3. Major sizes of ROBITA head (frontal view).
major sizes of ROBITA head
	  (frontal view)
Fig. 4. Major sizes of ROBITA head (side view).
major sizes of ROBITA head
	  (side view)

2.Recording condition

A speaker was located at a distance of 3 [m] from a center of the robot and in each 10-degree interval in a range from a left 90-degree direction to a right 90-degree direction to measure HRTF in each position. For measurement of impulse response, a TSP signal was used. Arrangement of the robot head and the TSP sound source is shown in Fig. 5 and Fig. 6, respectively. Table 1 shows various conditions of measurement.

Table 1. Various conditions of HRTF measurement.
Binaural device Directional microphone Non-directional microphone
Date of recording Nov 19, 1999 Oct 28, 1998
Place of recording Anechoic room (volume 1000m3)
Sampling and quantization 48kHz 16bit
Number of TSP 131072 65536
Synchronous addition times 8 30
A/ D input stage filter Absent LPF (cutoff freq. 16 kHz)
Model of microphone Audio-technica ATM-15a Audio-technica ATM-14a
Directionality Unidirectional Omnidirectional
Frequency characteristic 100-15000Hz 30-20000Hz
Sensitivity (0 dB = 1 V/ 1 Pa) -43.5dB -44.5dB

arrangement of ROBITA head and the TSP source (top view)
Fig. 5. Arrangement of the head of the interactive robot ROBITA and the TSP
 sound source (top view). 

arrangement of ROBITA head and the TSP source (Side View)
Fig. 6. Arrangement of the head of the interactive
 robot ROBITA and the TSP sound source (side view). Table 1 shows various
 conditions of measurement.

3. Result of measurement

The frequency characteristic of the HRTF measured will be shown. A blue and green series are HRTF of a left ear and a right ear, respectively, and a red series is a differential of the left and right HRTF. For figures of directions other than 90 degrees, refer to the link thereof.

HRTF in the -90-degree (left 90-degree) direction
 (non-directional microphone)
Fig. 7. HRTF in the -90-degree (left 90-degree) direction
(non-directional microphone)


HRTF in the -90-degree (left 90-degree) direction
 (directional microphone)
Fig. 8. HRTF   in the -90-degree (left 90-degree) direction (directional
 microphone)

-90 degree (L) Non-directional -90 degree (L) Directional 90 degree (R) Non-directional 90 degree (R) Directional
-80 degree (L) Non-directional -80 degree (L) Directional 80 degree (R) Non-directional 80 degree (R) Directional
-70 degree (L) Non-directional -70 degree (L) Directional 70 degree (R) Non-directional 70 degree (R) Directional
-60 degree (L) Non-directional -60 degree (L) Directional 60 degree (R) Non-directional 60 degree (R) Directional
-50 degree (L) Non-directional -50 degree (L) Directional 50 degree (R) Non-directional 50 degree (R) Directional
-40 degree (L) Non-directional -40 degree (L) Directional 40 degree (R) Non-directional 40 degree (R) Directional
-30 degree (L) Non-directional -30 degree (L) Directional 30 degree (R) Non-directional 30 degree (R) Directional
-20 degree (L) Non-directional -20 degree (L) Directional 20 degree (R) Non-directional 20 degree (R) Directional
-10 degree (L) Non-directional -10 degree (L) Directional 10 degree (R) Non-directional 10 degree (R) Directional
0 degree (Front) Non-directional 0 degree (Front) Directional

4. A list of data recorded

The data recorded is the impulse response of HRTF  of the robot, measured using the directional microphone and the non-directional microphone.  Data format is the raw format of 48 kHz, 32-bit float, and little endian.

/robot/
 +---/data
      +---/non ....... The impulse response of the non-directional microphone.
      |    +---m90r.dat    -90-degree right ear
      |    +---m90l.dat    -90-degree left ear
      |    :
      |    +---m10r.dat    -10-degree right ear
      |    +---m10l.dat    -10-degree left ear
      |    +---000r.dat      0-degree right ear
      |    +---000l.dat      0-degree left ear
      |    +---p10r.dat    +10-degree right ear
      |    +---p10l.dat    +10-degree left ear
      |    :
      |    +---p90r.dat    +90-degree right ear
      |    +---p90l.dat    +90-degree left ear
      |
      +---/dir ....... The impulse response of the directional microphone    
           +---m90r.dat    -90-degree right ear
           :
           +---p90l.dat    +90-degree left ear

Supplement:
Concerning synchronous addition, materials preparation did not allow synchronization of a master clock of a TSP radiation system and a TSP response recording system. Therefore, synchronous addition was carried out as follows.

  1. Synchronizing with inputs of both microphones, the waveform of their TSP
     radiated was recorded and used as the standard signal.
  2. Matching of respective phase was carried out for a sampling clock unit.
  3. Each waveform has a temporal difference of less than 1 clock and thus a linear phase difference occurs between respective waveforms. Gradient of the phase difference to the frequency is calculated by the least square method.
  4. A phase compensation is carried out on the basis of gradient of the phase difference calculated for the waveform of an actually received sound in the above section.

[Back to top of RWCP Sound Scene Database]


RWCP Sound Scene Database in Real Acoustical Environments
Copyright (c) 1998-2001 Tetsunori Kobayashi, Waseda University