Sound calibration: Difference between revisions

From biophysics
Jump to navigation Jump to search
No edit summary
 
Line 1: Line 1:
==Introduction==
==Introduction==
% todo
For continuous sounds the dBspl is normally determined by the rms of the signal. Because of the silence periods in speech there is a special way to determine the dBspl.
 
==Speech==
The standard for speech is called Active Speech Level. It determines the dBspl by removing the silent parts from the speech.
 
There are two ways to do this:
*More precise: Sum the rms only over the parts above a certain power threshold.
*More easy:    Take the rms over the total sound and estimate the percentage of silence in the speech and correct for it.
 
ITU-T P.56 (Active Speech Level): This is the gold standard within speech science and telecommunications. This algorithm utilizes a dynamic threshold. Silences and background noise falling below this threshold are filtered out of the calculation. The $\text{dB SPL}$ is then calculated exclusively over the periods where speech energy is actually present.


==Sensitivity of speakers==
==Sensitivity of speakers==

Latest revision as of 09:39, 8 June 2026

Introduction

For continuous sounds the dBspl is normally determined by the rms of the signal. Because of the silence periods in speech there is a special way to determine the dBspl.

Speech

The standard for speech is called Active Speech Level. It determines the dBspl by removing the silent parts from the speech.

There are two ways to do this:

  • More precise: Sum the rms only over the parts above a certain power threshold.
  • More easy: Take the rms over the total sound and estimate the percentage of silence in the speech and correct for it.

ITU-T P.56 (Active Speech Level): This is the gold standard within speech science and telecommunications. This algorithm utilizes a dynamic threshold. Silences and background noise falling below this threshold are filtered out of the calculation. The $\text{dB SPL}$ is then calculated exclusively over the periods where speech energy is actually present.

Sensitivity of speakers

  • In all our labs we use the Cambridge Audio MINX MIN12 speaker.
  • The speaker has a specified sensitivity of 86 dB SPL (@2.83 Vrms input and @ 1 kHz)
  • The calibration is calculated @ 1Vrms input and depends one over the square of the distance of the speaker to the center of the subjects head.
   Calibration = mag2db(db2mag(Sens@1V)/Distance^2)

Table: calibration (@ 1Vrms and 1kHz) in different labs.

LAB Distance (m) Calibration(dB SPL) Measured (dB SPL)
TEST LAB 1.00 80 ?
AUDITORY PERCEPTION LAB 1.05 76.1 ?
AUDITORY SPHERE LAB ? ? ?
AUDITORY MOTION LAB ? ? ?
NIRS EEG LAB ? ? ?

Sensitivity of headsets

A commercial headset normally have a higher impedance than an audiological headset and therefore a higher sensitivity.

Table: impedances and sensitivity @ 1mW and @ 1V for Headphones measured @ 1 kHz

Model Impedance (Ohm) Sens@1mW (dB SPL) Sens@1V (dB SPL) Difference (dB) Audiological
AKG_K271_MKII 55 91 104 12.5 N
SENNHEISER_HD600 300 (92) 97 5.2 N
SENNHEISER_HD100 26 (94) 110 15.8 N
BEYERDYNAMIC_DT_770_PRO 80 96 (107) 11.0 N
SENNHEISER_HDA280 37 (103) 117 14.3 Y
SENNHEISER_HDA300 23 (108) 124 16.4 Y
RADIOEAR_P4492 10 (107) 127 20 Y
TELEPHONICS_TDH-39P 10 (108) 128 20 Y
  • The numbers are from the specs on the internet.
  • The numbers between brackets are calculated from the difference between the 1mW and 1V sensitivity by: Sens(1V) = Sens(1mW) + 10*log10(R/1000)