A startup that can ‘translate’ baby cries is now exploring early detection of autism in newborns

• Founded two years ago, Zoundream specializes in cataloging and translating baby cries.
• In its first round of funding last October, Zoundream raised just under $1 million.
• The company now wants to expand into early-stage detection of atypical developments in newborns.
• See more stories on Insider’s business page.

We all know children don’t come into the world with a “how-to” guide.

During the first months of a newborn’s life, it’s often a struggle for parents not just to meet their baby’s needs but simply to know what they are.

However much a new parent may want to soothe their baby’s endless desperate crying, it can be challenging without knowing what they want.

Many resign themselves to one of their first lessons as parents: they won’t always understand their children.

In the age of the Internet of Things, smartphones, and tablets, however, some are using tech to explore modern ways of working around age-old problems.

Ana Laguna, a 33-year-old scientist and expert in data management, gave birth to her first child in 2016.

After a few hours of crying, she had a thought – there had to be a way of translating a newborn’s cries.

The idea seemed such an obvious one that she assumed there must already have been a company that had successfully developed some kind of device or app, but the only thing she could find was a Korean application that was just about functioning.

Taken aback by what seemed somewhat of a technological oversight, her intuition soon turned into a project: she would record her own baby’s cry to look for patterns.

“Many projects come about by mistake or by necessity. Mine is one of the latter,” Laguna told Insider.

Over the years, Laguna’s project transformed into a fully-fledged company, Zoundream.

The company specializes in developing software to translate newborn babies’ cries, particularly those up to the age of six months.

After raising just under $1 million in its first round of funding in October 2020, Zoundream now wants its studies to help detect atypical developments in newborns at an early stage.

There have been several stages in Zoundream’s development to get it to where it is now.

Laguna’s first major concern was to find out whether babies from different countries cry differently.

If, for instance, the cry of a German baby were different from that of a Spanish baby, that would have significantly reduced the software’s potential audience – as well as the viability of the whole project.

After many hours of gathering information through scientific publications on the subject and analysis of sound samples, Zoundream came to a conclusion – although there were notable differences in the prosody of the cry, the content is always the same across languages.

In other words, though German and Spanish babies may sound different, they’re essentially trying to say the same thing.

The only thing left to do was to get the business going – that’s where Roberto Iannone, the company’s current CEO, comes in.

Hundreds of kilometers away from Laguna, almost at the same time as her, Iannone, an entrepreneur, had already had a similar idea. So, when a colleague told him about Laguna and her studies on crying patterns in newborns, which were already beginning to gain some traction in the press, Iannone knew what he needed to do.

Zoundream was born out of a single idea – while there are more than 7,000 languages in the world, the way newborn babies express their needs is universal.

Now the company translates babies’ cries into five types: hunger, sleep, pain, gas, and attachment or the desire to be held.

This classification method works best on infants up to the age of three months, when crying is more genuine. From this point on, according to Laguna, the baby’s brain synapses become more complex – they start to be able to learn at full speed.

As a result, babies start using certain strategies to get what they want.

In other words, human beings learn to lie before they learn to speak.

After the birth of her second child, Laguna decided she didn’t just want to record her own child’s cries; she wanted for other parents to be able to contribute to the project.

A newborn cries an average of two to three hours a day. Over time, Zoundream managed to collect thousands and thousands of hours of cries analyzed using spectrograms, from Europe, Asia, and the entire American continent.

This means tZoundream is already working on refining translations through devices that are still just prototypes.

Zoundream is building partnerships with companies that, in the coming years, will make it possible to integrate this technology in prams, bracelets, or even in surveillance cameras.

The one condition is that the system has to be automatic.

“When a child cries, their parents will go to attend to them and entirely forget about the mobile phone and everything,” says Laguna.

In addition to the audio, the company started to receive feedback.

“I remember, for example, a mother who said that her son wouldn’t stop crying. The recordings told us that he was hungry,” says Laguna. “It didn’t make sense to her, because she kept breastfeeding him. Eventually, she told us that we were right, that the doctor had detected a problem with his lingual frenulum and that he wasn’t feeding well.”

Cases like these have inspired Zoundream’s team to try to take the next leap and detect atypical developments through the way babies cry.

Some cases, she says, are obvious: “The cry of a child on the autistic spectrum is very characteristic, very hoarse. You can see it quite clearly on the spectrogram.”

By doing this, Laguna explains, the company hopes to help improve the early diagnosis possibilities, which can greatly improve quality of life.

“In cases of early diagnosis,” she says, “autism is detected at around the age of two. Imagine the improvement if it could be done before the age of six months.”