The study and production of chorus effect has been a research and development issue during the last decades. In this paper we present a preliminary study of accuracy in choral intonation intervals. Six semi- professional male singers from a medieval liturgical chant choir are recorded
individually in an anechoic chamber singing two songs belonging to the Mozarabic repertoire. Although recordings are independent, synchronization is achieved by allowing singers to listen to a rehearsal of the song and to watch the choir director on a screen while they sing. Pitch is automatically extracted from the recordings and the stable parts of the notes are marked
manually. Mean pitch is calculated for each singer’s stable note and the pitch change, or interval, between consecutive notes is computed in cents. This interval is compared with the theoretical one indicated by the music sheet and the error (actual interval – theoretical interval) is obtained.
Statistical analyses show that all the singers present the same median, around 0 cents, but there are significant differences in error dispersion. These results provide a new insight into the characteristics that make up the chorus effect.