Objective: Measurement of reference speech-recognition curves for a specific speech test in typical clinical testing environments and for different loudspeaker configurations.
Design: Speech-recognition scores were measured at four signal-to-noise ratios for five loudspeaker configurations in two anechoic rooms, and in four audiometric test rooms with low reverberation times.
Study sample: 240 young participants (aged 18-25 years) without hearing impairment participated in the measurements.
Results: Reference speech-recognition curves for speech and noise from the front (S0N0) were similar across rooms. Compared to S0N0, lower speech-recognition thresholds (SRTs) were observed for all other loudspeaker configurations in which speech and noise were spatially separated. This spatial release from masking was significantly reduced for the audiometric test rooms compared to the anechoic rooms. A binaural speech-intelligibility model was used to verify the influence of room acoustic properties and loudspeaker configuration on SRT.
Conclusions: Speech-recognition curves for spatially separated loudspeaker configurations depend on the room acoustic properties, even in audiometric test rooms with low reverberation times. This makes it more difficult to compare clinical measurements with reference speech-recognition curves, or even with data measured in a different test room. It is thus recommended to document the loudspeaker configuration and test room for each clinical measurement.
Keywords: Freiburg monosyllabic speech test; Speech recognition; binaural speech-intelligibility model; loudspeaker configuration; reference speech-recognition curve; room acoustic properties.