Many animals communicate by producing sounds, so-called vocalizations. Mice use many different kinds of vocalizations in different social contexts. During social interaction recognizing the partner's sex is important and female mice appear to know the difference between male and female vocalizations. However, previous research had suggested that male and female vocalizations are very similar. We here show for the first time that the emitter's sex can be guessed from the vocalization alone, even single ones. The full spectrogram was the best basis for this, while reduced representations (e.g. basic properties of the vocalization) were less informative. We therefore conclude that while the information about the emitter's sex is present in the vocalization, both mice and our analysis must rely on complex properties to determine it. This novel insight is enabled by the use of recent machine learning techniques. In contrast, we show directly that a number of more basic techniques fail in this challenge. In summary, differences in the vocalizations between male and female mice allow to guess the emitter's sex, which enables sexual recognition between mice and automated analysis. This is important in studying social interactions between mice and how speech is produced and analyzed in the brain.
展开▼