Use of noise dosimeter recordings in low sound level environments is often problematic due to the voice contribution from the carrier. The aim of the study was to find a microphone position with low speech contribution while still providing an accurate sound level recording. An experiment with different microphone positions was conducted in a low sound reverberation room. Two types of noise dosimeters were used (Brüel & Kjaer 4445 and Larson Davis Spark 706-Atex). In repeated measurements the carrier was instructed to read a text chapter during 60 seconds and at a speech of level about 60 dB(A), acquired by practice, with different background noise and noise levels. White noise and preschool noise was used at sound levels 50, 60, 70 and 80 dB(A). The voice contribution to the background noise level was then measured. Three microphone positions were tested; on the shoulder, above the ear and on the back of the head. The position with the microphone placed behind the head of the carrier had the lowest speech contribution, less than 2 dB(A) to the background noise at 70 dB(A). This compared to positions right above the ear 6 dB(A and on the shoulder 12 dB(A).