It is widely accepted that, within the binaural spatialisation domain, the choice of the Head Related Transfer
Functions (HRTFs) can have an impact on localisation accuracy and, more in general, realism and sound sources
externalisation. The impact of the HRTF choice on speech-in-noise performances in cocktail party scenarios has
though not yet been investigated in depth. Within a binaurally-rendered virtual environment, Speech Reception
Thresholds (SRTs) with frontal target speaker and lateral noise maskers were measured for 22 subjects several
times across different sessions, and using different HRTFs. Results show that for the majority of the tested subjects, significant differences could be found between the SRTs measured using different HRTFs. Furthermore,
the HRTFs leading to better or worse SRT performances were not the same across the subjects, indicating that
the choice of the HRTF can indeed have an impact on speech-in-noise performances within the tested conditions. These results suggest that when testing speech-in-noise performances within binaurally-rendered virtual
environments, the choice of the HRTF should be carefully considered. Furthermore a recommendation should
be made for future modelling of the speech-in-noise perception mechanisms to include monoaural spectral cues
in addition to interaural differences