The thunderous roar of the T. Rex is one of cinema's most iconic sounds. It is also almost certainly wrong. Reconstructing the vocalisation of an extinct animal is one of palaeontology's most speculative undertakings — but the evidence we do have points consistently away from the Hollywood version.
Why They Probably Didn't Roar
Modern reptiles — crocodilians, lizards, turtles — don't roar. They hiss, grunt, bellow, and produce low-frequency vibrations. The roar is a mammalian specialisation, produced by a larynx with specific adaptations found only in cats and a few other species. Dinosaurs, as archosaurs, are more closely related to crocodilians and birds than to mammals.
Birds — the living dinosaurs — don't have a larynx. They vocalise using a structure called the syrinx, located deeper in the throat where the trachea branches. The syrinx is far more efficient than a mammalian larynx and capable of extraordinary complexity, but it doesn't produce mammalian roars.
Crocodilian evidence: Modern crocodilians produce deep, resonant bellows by vibrating water on their backs — a behaviour called the "water dance." These infrasonic rumbles can travel kilometres through water and ground. Large theropods may have produced similar low-frequency acoustic displays.
The Closed-Mouth Vocalisation Hypothesis
A 2016 study by Julia Clarke and colleagues proposed that many large dinosaurs may have used closed-mouth vocalisation — producing sound by inflating throat or body sacs and forcing air through them without opening the mouth, similar to the booming calls of modern emus, ostriches, and doves. This mechanism can produce extremely low-frequency sounds that travel long distances without requiring a complex larynx or syrinx.
The anatomy of several large theropods and sauropods is consistent with this hypothesis — inflatable throat structures leave bone traces that have been tentatively identified in some specimens.
Parasaurolophus: A Dinosaur We Can Simulate
The hollow crests of hadrosaurs like Parasaurolophus offer a rare opportunity for acoustic reconstruction. These crests were essentially biological wind instruments: curved tubes connecting the nostrils to the airways in a complex loop. By using CT scanning to map the interior of the crest and computational fluid dynamics, researchers have modelled what sound the structure would have produced when air passed through it.
The result, published by Diegert and Bakker in 1998 and later refined, was a deep, resonant, trombone-like tone — a low-frequency call that would have carried enormous distances through Cretaceous forests. This is almost certainly closer to the reality of hadrosaur communication than any roar.
What We'll Never Know For Certain
The honest answer is that direct vocalisation evidence is preserved only in exceptional circumstances — occasional soft tissue traces in exceptionally preserved specimens, and the structural evidence of resonating chambers. We can reconstruct the physics of potential sound production with reasonable confidence. The actual acoustic character, timing, and social context of dinosaur calls — whether they called at dawn, whether they had individual voices, whether their calls changed with age — remains, for now, beyond the reach of the fossil record.
What we can say with some confidence: dinosaurs were not silent. They almost certainly communicated acoustically, probably at low frequencies, probably without roaring. The real sound of a Cretaceous morning was probably stranger, and more beautiful, than anything the movies have imagined.