Technical Report: What is Sound Quality Evaluation? 2

Additionally, this characteristic varies with sound pressure. At high sound pressure levels, the characteristics are relatively flat, while at low sound pressure levels, sensitivity to bass decreases. The frequency characteristics of human hearing are very complex. Such complex characteristics are taken into account in loudness calculations. Incidentally, the A-weighting curve uses a filter with characteristics similar to those of auditory frequency response. However, since it uses a filter corresponding to the equal loudness curve at medium sound pressure levels (the curve at 40 phons in the figure), it may differ from the perceived loudness when heard by humans.

Now, suppose you are listening to narrowband noise at 1 kHz. At this time, the shaded area in the graph above is masked by the 1 kHz narrowband noise. Even if new sounds are added in this range, the loudness will not increase. Furthermore, some sounds may be completely drowned out and not heard at all.

The shape of the spectral masking curve varies with frequency. Additionally, the shape of the curve changes with different sound pressure levels. Spectral masking is a nonlinear and complex phenomenon. Furthermore, the curve produced by masking resembles that of a bandpass filter. The range corresponding to the passband of this filter is called the critical band (unit: Bark).

The reason spectral masking occurs lies in the structure of the ear. The following diagram shows the structure of the ear. Sounds that enter through the ear canal travel through the outer ear canal, vibrate the eardrum and ossicles, and reach the cochlea. The cochlea is an organ that easily performs frequency decomposition of sound, and it has a tubular shape that is coiled like a snail's shell.

The interior of the cochlea is divided into two parts by a membrane called the basilar membrane. The vibrations that reach the cochlea cause this membrane to vibrate as they progress from the cochlear entrance (left side) to the deeper part (right side), exciting the nerve cells located on the basilar membrane. At this time, a specific part of the basilar membrane vibrates particularly strongly. The part that vibrates strongly varies depending on the frequency of the incoming sound; for high-frequency sounds, the strong vibrations occur near the entrance, while for low frequencies, they occur further in. Therefore, if the frequency of the sound differs, the excited nerves also differ, allowing us to perceive whether the sound is high or low. The basilar membrane does not vibrate at just one point on the membrane but vibrates across a width. Thus, even if one hears a sound at 1 kHz, the surrounding frequencies corresponding to that sound will also excite nerves. If, at this time, one hears a sound that is slightly higher than 1 kHz, the already excited nerve cells cannot become more excited, and the loudness of the sound does not seem to change significantly. This is the phenomenon of spectral masking.

The diagram at the top of the previous page shows the case where two sounds, A and B, are separated on the frequency axis. The rectangular areas represent each sound, and the lightly shaded areas indicate the parts that are masked. The loudness of the sounds is proportional to these areas, so when sound B is added, it feels about twice as loud compared to when only sound A is present. On the other hand, the diagram below shows the case where the frequencies of the two sounds are close together. The areas masked by the two sounds overlap. As a result, even when sound B is added, the area does not change significantly compared to when only sound A is present, and the loudness only increases slightly. Since the energy of these sounds is the same in both the upper and lower diagrams, when compared in terms of sound pressure level, they yield the same value.

To simulate frequency masking, a model like the one shown in Figure A is used. The frequency masking curve for higher frequencies is less steep than that for lower frequencies, which has a greater impact on the surrounding area; therefore, only the upper curve is considered. To determine loudness, a 1/3 octave analysis is first performed, and the results are recorded on a chart as shown in the figure. The area below the drawn curve is calculated, and the corresponding loudness is read from that area. Depending on the sound pressure and sound field conditions, ISO 532B provides 10 types of charts. The appropriate chart is used to determine loudness.

In the chart of 5.4, the auditory frequency characteristics and spectral masking mentioned so far are taken into account. Another factor to consider here is auditory frequency resolution. Auditory frequency resolution corresponds to approximately 1/3 octave band width (when discussing sound intensity; the resolution for pitch is finer). However, in the frequency range below 500 Hz, the band width is larger than 1/3 octave, resulting in coarser resolution. Therefore, the ISO 532B method performs summation to widen the band width of the low frequencies according to auditory resolution. This is why the spacing between vertical lines in the low frequency section of the ISO chart is narrower, with 4-band or 3-band being grouped together.

Another important element of loudness is temporal masking (also known as time masking or temporal masking). This refers to the masking that occurs over the time axis.

For example, when a sound is followed shortly by another sound, the latter can be masked and become inaudible. This occurs because the vibrations of the membrane in the ear do not stop immediately after the first sound, but rather diminish gradually. Additionally, the excitation of the nerves in contact with the membrane also decreases gradually. Even if the next sound occurs before the excitation has sufficiently diminished, it will not be heard.

By the way, the method for calculating loudness standardized by ISO does not take into account the effects of temporal masking. Therefore, the loudness according to the ISO method can only evaluate steady sounds.