Lars Bramsløw

Senior Scientist

> Learn more

The most common speech-intelligibility tests are not very representative of typical real-life situations. The resulting speech reception thresholds (SRTs) are often unrealistically low. To increase real-life resemblance, speech masking and target location uncertainty was added to the Danish Dantale II test. The new version showed higher SRTs in a test with 16 hearing-aid users.

Test set-up

The 16 test subjects were tested in four conditions. Three of these were based on a traditional loudspeaker set-up with the target speech presented from the front (0°), and with masking signals (stationary noise, speech-modulated noise, and real running speech, respectively) presented from the left and right sides (±30°). In the fourth (new) condition, the target speech was randomly presented from three different locations (0° and ±45°), and the masking signal was real running speech presented from the sides (±30°).

  • Speech-in-speech intelligibility test with target location uncertainty fig1
    Figure 1. Loudspeaker set-up with indication of target (T) and masker (M) loudspeakers. The test was performed in Eriksholm’s anechoic chamber.


Whereas no difference in mean SRT was observed between the stationary and modulated noise masking conditions, the introduction of real-speech (and thereby informational) masking increased the mean SRT by around 1.5 dB. The introduction of target location uncertainty further increased the mean SRT by around 1 dB. Thus, both elements contributed to an increase of the ecological validity of the test, as measured by the SRT. This conclusion was supported by subjective statements made by test subjects who found the new version truer to the problematic situations they experience in everyday life.

  • Speech-in-speech intelligibility test with target location uncertainty fig 2
    Figure 2. Speech recognition results in each of the test conditions, indicated as the target-to-masker ratio corresponding to 60%-correct speech intelligibility (TMR60), averaged across participants. The 95% confidence intervals are also indicated.


The results of this study suggest that further work with the test could be interesting. For example, other target and masker speech signals could be used and/or the test could be implemented in another type of loudspeaker setup. In addition, further validation studies need to be completed.

Target location uncertainty is included as an option in the Spatial Fixed-SNR (SFS) test. In that implementation of target location uncertainty, the loudspeaker set-up is changed such that the angular separation between the three target locations and their nearest masker is the same, as opposed to the approach followed above. This change in realisation of target location uncertainty helped to clarify that there are two different effects of presenting the target off to the side:

1. An increases in speech recognition brought about by an improvement in the SNR at the ‘better ear’ due the head’s acoustic baffle effect.

2. A decrease in speech recognition, which is probably due to an attention bias towards the front loudspeaker.

Further reading

Jensen NS, Johannesson RB, Laugesen S, Hietkamp RK (2012). Measuring speech-in-speech intelligibility with target location uncertainty. In: Dau T, et al. (Eds.) Speech perception and auditory disorders. Proceedings of the International Symposium on Audiological and Auditory Research (ISAAR), Denmark.

Learn more

  • learnmore-poster


    Measuring speech-in-speech intelligibility
  • learnmore-poster


    Validation of a spatial speech-in-speech test that takes signal-to-noise ratio (SNR) into account