How does a voice get shaped by a source and a filter?
Think of a mammal's voice as a two-part system. The source is the vibration of the vocal folds in the larynx, which produces a buzzy sound with a certain pitch. The filter is the vocal tract — the throat, mouth, and nasal cavities — which acts like a set of pipes that amplify some frequencies and suppress others. The source-filter model says that the final sound you hear is the product of these two parts: the source provides the raw material, and the filter shapes its tone. This framework has been used to explain vocal signals across many mammal species, and it helps researchers predict how acoustic attributes vary and how listeners might discriminate between individuals [2].
In marmosets, researchers tested this model on whistle-like calls by analyzing the acoustics and simulating helium-modulated voices. The results showed that these calls are produced using mechanisms similar to human speech, supporting the idea that source-filter models apply broadly within primates [1]. In other words, even a small primate with a very different vocal anatomy from a human can still be understood through the same source-and-filter lens.
What can the filter tell a listener about identity?
The filter part of the model is especially useful for understanding individual differences. Because the vocal tract is shaped by each animal's anatomy, the filtering pattern can encode cues about who is calling. In fallow deer, researchers used the source-filter perspective to examine individual acoustic variation in common and harsh groans. They found that vocal production mechanisms contribute to individual identity in this species [5]. This means that the way a deer's vocal tract filters the source sound can make its groans acoustically distinct from those of other deer.
The same logic applies to domestic dogs. The source-filter model has been used as a theoretical and applied tool to explain why dog vocalizations sound the way they do, and it helps researchers understand how acoustic features vary in a companion animal's extended social environment [4]. For a general reader, the takeaway is that the filter is not just a passive tube: it carries information about the caller's size, shape, and identity, which other animals may use to recognize each other.
Where does the source-filter model fall short across mammals?
The model is general, but not universal in its basic form. One issue is anatomy: mammal species differ in vocal fold structure, and these differences can affect the acoustic signal in ways the standard source-filter model does not fully capture [3]. For example, the model assumes a relatively simple source, but if vocal folds vary in shape or stiffness across species, the source itself may produce a more complex sound than the model predicts.
Another limitation is that some acoustic features in animal signals are not included in the source-filter model at all, yet they still affect the frequency spectrum [3]. This means that while the framework is a powerful starting point for explaining many mammal vocalizations, researchers need to be careful about assuming it explains every detail. The model is best treated as a useful guide that highlights the two main stages of vocal production, rather than a complete description of every sound a mammal makes.
Sources used in this answer
The source-filter theory of whistle-like calls in marmosets: Acoustic analysis and simulation of helium-modulated voices
Marmoset whistle-like calls are produced using mechanisms similar to human speech, supporting the universality of source-filter models in primates.
The contribution of source–filter theory to mammal vocal communication research
The source-filter theory helps explain how vocal signals are produced and how acoustic attributes vary, aiding prediction of vocal variation and individual discrimination across mammal species.
Applying linear predictive coding (LPC) to frequency-spectrum analysis of animal acoustic signals
Differences among mammalian species in vocal fold anatomy, and acoustic features not included in the source-filter model, can affect frequency spectra in ways the basic model does not capture.
Auditory communication in domestic dogs: vocal signalling in the extended social environment of a companion animal
The source-filter model can be used as a theoretical and applied tool to explain vocal production in domestic dogs and why their vocalizations sound as they do.
Individual Acoustic Variation in Fallow Deer (Dama dama) Common and Harsh Groans: A Source‐Filter Theory Perspective
In fallow deer, vocal production mechanisms contribute to individual acoustic variation in common and harsh groans, supporting a role for source-filter mechanisms in individual identity.