Listening attempt

from Wikipedia, the free encyclopedia

Listening tests are experiments in which the acoustic perceptions of humans or animals are examined under defined conditions .

The aim of listening tests is a correlation between the physically occurring sound events and the physiologically - psychologically held auditory events to manufacture or to determine the acoustic expectations of customers while developing noise-generating products. Listening tests are the preferred form of psychoacoustic research .

The object of investigation of listening tests can be, for example

  • Properties of the auditory events such as timbre , tone , loudness , roughness
  • Spatial features such as distance, direction, extent
  • Trading attempts to explain the opposing effects of level difference and time difference on the ears.
  • Acceptance of the acoustic properties of a product (e.g. acoustic climate in a vehicle)

Query of perceptions

Since a person's perception cannot be measured directly, the results of listening tests depend on how well test subjects can describe their hearing events. Preliminary tests are particularly useful for inexperienced test subjects, in which the test subjects can learn to describe their perceptions or can even only become clear about the effect the sound being examined has on them.

There are different methods for querying perceptions

  • Perceptibility: Can a sound be perceived or not?
  • Identity: Do two sounds being presented sound the same or different?
  • Ranking: Sorting the sounds presented according to a given sorting criterion (louder / quieter, higher / lower, duller / sharper, further to the right / further to the left, etc.)
  • Intervals: Setting of the same intervals or the same ratios between the sounds presented according to a given criterion (twice as loud, twice as high, etc.)
  • acoustic assessment of products by potential customers or trained experts


Since many test subjects have difficulties in describing their perceptions and different people can perceive the same sound differently, the spread of the results of listening tests is often considerable. As an example: one test person perceives an increase in the sound level of 5 dB, while another perceives an increase of 15 dB as a doubling of loudness. When the experiment is repeated, the first test person reports a gain of 12 dB as a doubling of loudness, the second test person, on the other hand, now reports a noise that has been amplified by 8 dB, etc.

It is important to find out during the evaluation whether a trend can be derived from the results (e.g. a level increase of 10 dB ± 3 dB results in a doubling of loudness) or whether the test results are distributed so randomly that one must assume that the test subjects have just guessed. There are a number of statistical methods that can be used to clarify whether or not a reliable statement can be derived from the results. It can only be assumed that a certain physical sound situation is associated with a corresponding perception if there is a corresponding statistical significance.

The assessment also depends on the test environment. When evaluating noise climates in vehicles, For example, it makes sense to reproduce the interior of the vehicle as realistically as possible in the sound studio in order to achieve a relevant result.


Results of listening tests are e.g. B.

  • Perception thresholds such as the hearing threshold , lowest and highest perceptible frequency
  • Discrimination thresholds, e.g. B. smallest distinguishable pitch difference, smallest distinguishable difference in direction, smallest distinguishable time interval
  • Perception scales, e.g. B. Loudness , tone , lateralization . Scales for the strength of perception are derived from the measured values. Psychoacoustic units of measurement are defined for important scales , such as Sone for loudness, Mel for tone, etc.

The accuracy of the psychoacoustic measurement results obtained in this way does not approach the accuracy of physical measurements. Usually, in addition to the order of magnitude, an accuracy of only one decimal place can be achieved. An accuracy of two digits is a very good result here.