Ten teachers made recordings during one normal working day using a portable DAT recorder and a head-mounted microphone. In addition, the subjects filled in a questionnaire of signs of vocal fatigue. The speech samples were selected from the first and last lesson from three points representing the beginning, middle and end part of the lesson, respectively. To standardize the samples, 30 [a] vowels from stressed syllables were chosen for spectral analysis. The level of the fundamental and second formant regions (L1), the level of frequency ranges 2-5 kHz (L2) and 5-10 kHz (L5) were measured. From these measurements the parameters L1-L0, L1-L2 and L1-L5 were formed and used in the analyses as well as the energy levels below and above 1 kHz (alpha). Statistically significant changes were observed in the following parameters: the L1-L2 and L1-L5 differences, and the alpha ratio. In general, there was an increase in the energy content of the high frequency components due to vocal loading. The subjective reports revealed a statistically significant relationship with the spectral characteristics.