Humans almost certainly began singing long before they made any musical instruments, which places the origins of song somewhere hundreds of thousands of years ago. The earliest known flutes, carved from bone, date back roughly 40,000 to 60,000 years, but singing requires no tools at all. Because voices don’t fossilize, pinpointing an exact date is impossible, but converging lines of evidence from anatomy, genetics, archaeology, and primate behavior point to vocal musicality emerging gradually in our hominin ancestors well before modern humans even existed as a species.
What the Archaeological Record Shows
The oldest physical evidence of music comes from a bone flute found at the Divje Babe cave site in Slovenia, dated to roughly 60,000 years ago. What makes this artifact remarkable is that it was made by Neanderthals, not modern humans. The flute has two well-preserved holes and two damaged ones, and analyses have ruled out the possibility that the holes were caused by animal bites or coincidence. It was found near a hearth in a layer of sediment dated by electron spin resonance on bear teeth.
The next oldest flutes, found at sites in Germany, were made by anatomically modern humans around 40,000 years ago. These are complex instruments capable of producing differentiated pitches. But a flute is already a sophisticated piece of technology. You don’t carve a bone flute unless you already have a rich understanding of melody and pitch, which means vocal music must have been well established by the time anyone thought to build an instrument. The flutes are a floor, not a ceiling, for the age of music.
Singing Before Language
One of the most influential ideas in evolutionary musicology is that singing and speaking share a common ancestor. Rather than music being a byproduct of language, both may have descended from an older form of vocal communication that blended melody and meaning. Steven Brown’s “musilanguage” model describes this precursor as a spectrum: purely emotional expression on one end and purely referential meaning on the other, with early hominin vocalizations sitting somewhere in between.
In this model, the earliest form of group vocalization looked nothing like a choir. Multiple individuals would produce brief calls simultaneously, each at their own pitch and timing, creating a jumbled, overlapping sound. There was no harmony, no shared beat, no melody in the modern sense. The point wasn’t musical coordination. It was each individual expressing the same emotional state while remaining distinctly identifiable within the group. Over vast stretches of time, this raw vocal behavior gradually split into two branches: one becoming more structured and rhythmic (music), the other becoming more precise and symbolic (language).
Clues From Other Primates
Gibbons offer a fascinating window into what pre-human singing might have looked like. Unlike great apes and most other primates, every gibbon species produces elaborate, species-specific songs. Mated pairs in most species combine their vocalizations into coordinated duets, with males and females contributing distinct, sex-specific parts. Research on gibbon song evolution suggests that duetting was present in the last common ancestor of all living gibbons and that it evolved from a song pattern originally shared by both sexes, which later split into separate male and female parts.
Gibbons diverged from the human lineage roughly 20 million years ago, so their singing isn’t a direct model for ours. But their behavior demonstrates that complex, structured vocalization can evolve for social and territorial purposes in primates without anything resembling language. It suggests the raw capacity for song-like vocal behavior has deep roots in the primate family tree.
The Anatomy That Makes Singing Possible
Singing depends on precise control of the vocal folds, breathing muscles, and the small bone at the base of the tongue called the hyoid. When you sing higher notes, your larynx shifts upward and tilts forward, stretching the vocal folds to change their tension and length. This requires fine motor coordination that goes well beyond what’s needed for basic primate calls.
The Kebara 2 Neanderthal skeleton, found in Israel and dated to about 60,000 years ago, preserved a hyoid bone virtually identical in shape to that of modern humans. Computer modeling of the Neanderthal vocal tract, using this hyoid as a reference, found that Neanderthals could produce vowel sounds with frequency ranges close to those of modern humans for most vowels, though they may have struggled with the full range of the “ah” sound. This doesn’t prove Neanderthals sang, but it shows their anatomy wouldn’t have prevented it.
The FOXP2 gene provides another piece of the puzzle. Mutations in this gene cause severe difficulties with the complex, rapid mouth movements needed for speech. The gene is active across vertebrates and appears to influence the brain circuits responsible for sensory processing, sensorimotor integration, and skilled coordinated movements. The human version of FOXP2, with its specific mutations, was likely in place by the time modern humans and Neanderthals shared a common ancestor, perhaps 500,000 or more years ago, giving both species the genetic toolkit for sophisticated vocal control.
Why Early Humans Sang
The most compelling evolutionary explanation for singing centers on social bonding. Physical grooming, the way most primates maintain social relationships, is limited to one partner at a time. As hominin groups grew larger, they needed behaviors that could bond many individuals simultaneously. Group singing fits this role precisely. Research on modern singers has shown that coordinated group music-making triggers the release of endorphins and elevates pain thresholds, essentially producing a communal version of the feel-good effect of being groomed. This allows cohesive groups to expand well beyond the size that one-on-one grooming could sustain.
Hunter-gatherer groups that periodically assembled into larger mega-bands relied heavily on group rituals involving singing and dancing to create and maintain social ties among people who didn’t interact daily. Singing served as a kind of social glue that could hold together communities of dozens or even hundreds of individuals.
Another likely origin runs even deeper: the bond between mothers and infants. Human babies are born ready to respond to the exaggerated pitch contours, rhythmic repetition, and melodic quality of a caregiver’s voice. This sing-song style of communication, sometimes called motherese, appears across virtually all cultures. Evolutionary psychologist Ellen Dissanayake has argued that this ritualized vocal interaction between ancestral mothers and infants was itself an adaptation, one that later gave rise to music and dance as tools for broader group bonding. If this is correct, the earliest “singing” may have been a mother soothing her baby with patterned, melodic vocalizations, a behavior that could predate our species entirely.
Putting the Timeline Together
No single discovery can tell us the exact moment singing began, but the evidence converges on a rough picture. The genetic foundations for fine vocal control were likely in place at least 500,000 years ago, shared by the common ancestor of humans and Neanderthals. Neanderthals had the vocal anatomy for song-like sounds and, if the Divje Babe flute is genuine, were making music at least 60,000 years ago. Modern human flutes appear around 40,000 years ago. The musilanguage hypothesis pushes the origins of group vocalization back even further, potentially to early members of the genus Homo over a million years ago.
The honest answer is that singing probably didn’t “start” at any single point. It emerged gradually, from simple emotional group calls, to mother-infant melodic bonding, to the structured vocal traditions that eventually became recognizable as song. By the time humans were carving flutes from cave bear bones, singing was already ancient.