ECHO Corpus: Cross-cultural database of 15000 nonverbal vocalisations from speakers of 22 languages
The ECHO corpus is a groundbreaking, open-access database comprising over 15,000 non-verbal vocalizations from nearly 1,050 speakers across 52 countries and 22 languages, designed to provide a representative foundation for investigating the cross-cultural mechanisms of human emotional vocal communication.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Human beings have a vast toolkit for communicating without using words. Long before we developed complex languages to describe the world, our ancestors relied on sounds like screams, cries, and laughter to signal danger, pain, or joy. These non-verbal vocalizations are not just random noises; they carry specific information about who is making them, how big they are, and what they are feeling in that moment. Scientists have long wondered if these sounds are universal, meaning a scream of fear sounds the same to a person in Tokyo as it does to someone in the Andes, or if culture shapes how we make and understand these sounds. While previous studies have looked at this question, they often relied on small groups of people from similar backgrounds, leaving a huge gap in our understanding of the full human experience.
A team of researchers has now filled that gap with a massive new collection of data called the ECHO Corpus. This project brings together nearly 1,050 people from 52 different countries, representing speakers of 22 distinct languages that span ten major language families. The goal was to record over 15,000 non-verbal sounds made by these individuals as they acted out specific emotional situations. The participants ranged in age from 18 to 84 and included people from diverse backgrounds, from urban centers to isolated nomadic communities in Mongolia who have never traveled outside their region. By gathering such a wide variety of voices, the researchers created a resource that allows scientists to finally test whether the way we express emotions through sound is a shared human trait or something learned from our specific culture.
To gather this data, the researchers asked each participant to imagine a specific scenario and then produce a sound to match it, without using any words. They were given situations ranging from negative experiences, such as being attacked by a wild animal or burning one's hand, to positive ones, like receiving an unexpected gift or feeling relieved after a close call. They also recorded sounds for neutral situations, like when someone has not heard what was just said. The instructions were translated carefully into each participant's native language to ensure everyone understood the scenarios in the same way. While most people recorded their voices online using their own microphones, the team also traveled to Japan and Mongolia to record people in person, ensuring high-quality audio from communities that are rarely included in scientific studies.
The resulting database is enormous, containing thousands of audio files alongside detailed information about each speaker, including their age, sex, and how much media they consume. The researchers analyzed the acoustic properties of these sounds, looking at features like pitch, loudness, and the roughness of the voice. They found that the sounds varied in predictable ways depending on the emotion being expressed. For example, sounds meant to signal alarm or fear tended to be higher in pitch, while sounds meant to signal aggression were lower and rougher. Crucially, the study also included experiments to see if people from different cultures could understand these sounds. The results showed that listeners were generally very good at recognizing the intended emotion, even when the sound was made by someone speaking a completely different language.
This work suggests that while culture might add some subtle flavor to how we make sounds, the core ability to produce and understand emotional vocalizations is deeply rooted in our biology. The researchers found that people could identify emotions like pain, fear, and amusement across cultural boundaries with a high degree of accuracy. This supports the idea that these sounds are part of a shared human system, likely evolved to help us survive by quickly communicating our needs and feelings to others. However, the study also noted that people were slightly better at recognizing emotions when the sound came from someone in their own cultural group, hinting that there is still some influence of local habits on how we speak and listen.
The ECHO Corpus is now available to scientists around the world, offering a foundation for future research into how humans communicate. By moving beyond the limited groups that have dominated previous studies, this project reveals the true diversity of the human voice. It shows that despite our many differences in language and custom, the fundamental sounds of our emotions remain a powerful, connecting thread across the globe. This large-scale effort confirms that the voice is a universal instrument, capable of telling a story of joy or terror that anyone, anywhere, can understand.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.