The Viterbi’s algorithm allows to estimate latent time seriesaccording to observations in a hidden Markov model. This algorithm canbe used to merge estimations from different modalities as proposed inthis paper. Such a multi-modal estimation is more efficient than mono-modal estimations when the modalities are subject to independent noises.In this paper, this improvement is evaluated in function of noise level ofmodalities. Experiences on toy data and actual signals to estimate the fetalheart rate show that merging modalities will provide better estimations onaverage than using the modalities separately.