Logo PTB

Better accessibility for the hearing-impaired in public transport areas by optimizing public address announcements


The transmission quality of public address systems is evaluated by means of the Speech Transmission Index (STI) that essentially describes the goodness of modulation transfer (DIN EN 60628-16). An STI value of 0 corresponds to a total loss of signal modulation, whereas an STI value of 1 represents ideal signal transmission. For emergency warning systems an STI value ≥ 0.5 is required, which results in a sentence intelligibility ≥ 95 % for the normal-hearing. The sentence intelligibility at a given STI value is, however, reduced by hearing impairment. The aim of this project was the determination of objective parameters to evaluate speech intelligibility for the hearing-impaired in public transport areas. Based on these parameters, modified requirements were supposed to be derived, to provide also the hearing-impaired with sufficient sentence intelligibility for emergency announcements. 

Speech intelligibility was measured in a synthetic sound field with a 9-channel setup (see Figure 1) using the Oldenburg Sentence Test (OLSA). Eight of the nine loudspeakers were arranged equidistantly around the test subject (loudspeakers 1-8 in Figure 1). The ninth loudspeaker was placed facing the test subject, at the head height and in the same plane as the loudspeakers of this side.

To simulate the acoustic characteristics of public transport areas, the speech material was modified by adding reverberation or noise, each resulting in the realization of ten different measurement conditions with STI values reaching from 0.48 to 0.75. In the third test setting, the STI was set to a fixed value of 0.5 with ten different combinations of reverberation time and noise level. The frequency spectrum of the noise corresponded to the average spectrum of passing passenger trains in railway stations. The presentation level (75 dB (A)) met the minimum requirement for emergency warning systems.


Figure 1: Schematic diagram of the measurement setup in the anechoic room. The loudspeakers are denoted by the red numbers. The test subject (VP) is located in the center of the setup.

Figure 2: Mean value and standard deviation of the sentence intelligibility for different STI values in reverberation for normal- hearing (left panel) and hearing-impaired test subjects (right panel). The dashed red line indicates the mean value reduced by the standard deviation.

The measurements were carried out with 27 normal-hearing (reference group) and 33 hearing-impaired test subjects, 14 of them hearing-impaired with low-degree hearing losses, at the point, when using a hearing aid is prescribed and 19 hearing aid wearers.

The results of the study showed that the speech intelligibility of the hearing-impaired is more strongly affected by reverberation than by noise.

An example of the results is given in Figure 2, showing the mean value und standard deviation of the sentence intelligibility for normal-hearing and hearing-impaired test subjects in reverberation. While the normal-hearing test subjects expectedly comprehended, on average, 95 % of the sentences at all tested STI values, the hearing-impaired test subjects only reached a sentence intelligibility of less than 80 % at an STI value of 0.51. A sentence intelligibility greater than 90 % was reached only at STI values ≥ 0.63.

In contrast, the mean sentence intelligibility of the hearing-impaired in noise was already 95 % from STI values from 0.66 onwards.

Despite the fact that a mean sentence intelligibility of 95 % could not be completely achieved by the hearing-impaired in reverberation, summarizing  the statistical analysis of the results showed no significant differences of the sentence intelligibility of the hearing-impaired, compared to the 27 normal-hearing test-persons, at STI values ≥ 0.69 (mean value: 93 %).

Therefore, the safety limit "STI ≥ 0.69" was derived as a modified requirement to provide the hearing-impaired with sufficient sentence intelligibility in public transport areas.


[1] Wagener K, Brand T, Kühnel V, Kollmeier B (1999): Entwicklung und Evaluation eines Satztests für die deutsche Sprache I–III: Design, Optimierung und Evaluation des Oldenburger Satztests . ZfA, 38 (1– 3) (in German)

Contact person:

Florian Kramer, Dept. 1.6, WG 1.61, e-mail: florian.kramer@ptb.de