Hearing test simulation

from Wikipedia, the free encyclopedia

The hearing test simulation is the technical reproduction of subjectively performed hearing tests .

tasks and goals

With the help of the hearing test simulation, the relatively high effort for carrying out hearing tests is to be saved through objective measurement methods and the scattering of the assessment of a subject, which is dependent on the daily form, is avoided.

The aim is to always achieve the same measurement results (reproducible measurement values) regardless of the propagation delay and modulation of an audio system with the same signal content .

Possible uses

Hearing tests are required when information on the quality of an audio system can no longer be made using known quality criteria such as signal- to- noise ratio (SNR), distortion factor or frequency response . The quality of an audio system can be impaired by the following hardware and software components.

These components can produce subjectively perceptible disturbances such as stochastic amplitude fluctuations , phase jumps , non-linear residual echoes, short-term noise disturbances, echoes, spectral coloring or artifacts (musical tones) and cannot be detected with the above-mentioned quality criteria.

Subjective hearing tests

Hearing tests are carried out according to the recommendations of the ITU-T at a defined playback volume and ambient conditions.

The subjective quality assessment is severely impaired by hearing test parameters such as playback volume, bandwidth and acceptance threshold. Background noises are perceived as more annoying when the playback volume is louder than when it is quieter. Limiting the bandwidth always affects the perceived quality. The acceptance threshold is a measure of the maximum accepted loudness of disturbances and is the opposite of what is expected of the audio system to be assessed. The quality expectations of an old telephone are much lower (high acceptance) than of a high quality audio system (low acceptance).

During listening tests, different audio samples are played to a test person. The test person then rates the audio samples on a given scale. According to ITU-T P.800, a point scale with Excellent = 5, Good = 4, Fair = 3, Poor = 2 and Bad = 1 is suggested.

The results of the subjective assessment are often incomprehensible in practice. This has, among other things, to do with the personal well-being of the test subjects, who perceive and evaluate the exact same example differently when repeated.

For these reasons, many test persons have to carry out various hearing tests in order to form a representative mean value, the so-called Mean Opinion Score (MOS), which is reflected in the high level of effort and cost.

Objective measurement methods for hearing test simulation

Objective measuring methods simulate the hearing properties of the human ear and evaluate the quality of the audio systems to be examined based on the audible errors between its input and output signal. The simulation of the human hearing sensation is implemented today by means of arithmetic rules that can be derived from the research results of Eberhard Zwicker with digital signal processing systems. Well-known measurement methods include PESQ ( Perceptual Evaluation of Speech Quality ) and HASQUE ( Hearing Adequate Signal Quality Evaluation ).

PESQ is a standardized procedure according to ITUT-P.862, which works with fixed hearing test parameters. The bandwidth of the signals to be evaluated is limited by the use of IRS filters . Disturbances in speech pauses are not assessed.

The HASQUE method is characterized by the fact that the hearing test parameters playback volume, bandwidth and acceptance threshold can be set via the parameterization and thus hearing tests can be simulated for a wide variety of applications . HASQUE works without band limitation and also evaluates disturbances in speech pauses.

Hearing test simulation with band limitation

So- called IRS or MIRS filters ( Intermediate Reference System ) are used in some hearing test simulators to simulate telephone quality . These filters limit the available bandwidth at 8 kHz (16 kHz) sampling rate to 300 to 3400 Hz (100 to 7000 Hz), which means that not the entire available frequency range from <100 to <4000 Hz (<100 to <8000 Hz) is evaluated.

However, this band limitation has the effect of a clearly perceptible reduction in quality, especially in systems scanned at 8 kHz, since both the natural base frequency at approx. 100 Hz and sibilance, which can improve transparency and thus intelligibility, are suppressed.

Therefore, in hearing test simulators with IRS band limitation, no quality differences are measured in this area, even if there are subjectively clearly perceptible differences.

Areas of application

The hearing test simulation is particularly needed in the research, development and certification of audio and telecommunication systems.

With the aid of the hearing test simulation, audio systems can also be parameterized automatically. In the field of research, the optimally achievable quality of new processes can be determined much more efficiently than through manually performed iterative processes.

The adaptive parameterization of new methods for optimizing the signal quality (bandwidth expansion, noise reduction, echo compensation ...) places high demands on objective measurement methods. For these tasks, objective measurement methods without band limitation and with evaluation of the speech pauses are required.

Whether a process is suitable for the desired application depends on the requirements. For example, high levels of interference in pauses in speech can be generated by speech codecs which are not designed for the reproduction of noises, for example terrestrial trunked radio (TETRA). In reality, however, these disturbances occur due to ambient noise and then require a measuring method which also detects disturbances in pauses in speech if realistic ambient conditions are to be recorded during the measurement.

literature

  • E. Zwicker: Psychoacoustics. Springer, Berlin 1982, ISBN 3-540-11401-7 .
  • E. Zwicker, H. Fastl: Psychoacoustics. Springer, Berlin 1999, ISBN 3-540-65063-6 .
  • ITU-T P.800 Methods for subjective determination of transmission quality.
  • ITU-T P.830 Subjective performance assessment of telephone-band and wideband digital codecs.
  • ITU-T P.835 Subjective test methodology for evaluating speech communication systems that include noise suppression algorithm.
  • ITU-T P.862 Perceptual evaluation of speech quality (PESQ), an objective method for end-to-end speech quality assessment of narrow band telephone networks and speech codecs. 02/2001.