Sometimes a simple “wah” is enough, but other times babies might need a little extra help communicating.
According to Switzerland-based startup Zoundream, that extra help can come in the form of artificial intelligence (AI) and a proprietary device that translates what they call the “universal language” of babies.
“Every time a baby cries, they are asking for something,” said Zoundream CEO Roberto Iannone in an interview with IE at the Mobile World Congress in Barcelona. The company’s idea stems from the principle that “babies cry more or less the same way for the same needs,” he said.
Get more updates on this story and more with The Blueprintour daily newsletter: Sign up here for free.
Zoundream collected thousands of hours of crying baby data
Iannone did stress that Zoundream’s device is by no means a replacement for that strong parental intuition. Instead, it acts as an aid that enhances the parent’s own abilities. “Our device is absolutely not made to replace the role of the parents in understanding their own babies,” said Zoundream CEO Roberto Iannone in an interview with IE at the Mobile World Congress in Barcelona. “It’s actually a way to give them confidence, and to bring more attention to how the baby cries and what they need.”
Zoundream used machine learning software to analyze thousands of hours of international baby cries. They then classified these cries into four different categories: hunger, pain, gas, and wants hug. The company was co-founded by Iannone alongside data scientist Ana Laguna, who Iannone approached when he discovered she was collecting data on baby cries and defining patterns in these cries.
“When we first started,” Iannone explained, “we bought some cheap recorders and we asked some parents, and even paid some of them, to keep a recorder close to their baby.” This, he said, was a very time-consuming manual process, but at least it was a start. Then, “as soon as it became possible, we developed our own machine, which was essentially a recorder that filtered out all of the sounds that weren’t baby cries.” This, he explained, was good for privacy, and also meant they no longer had to spend many hours manually filtering the recording to verify whether they had picked up baby cries, or other sounds. “That really started giving us some data,” he said.
A baby’s “own language”
The company’s CEO highlights the role of prosody in the early development of babies, which is essentially their ability to recognize and communicate via melodic intonations rather than speech. This is ingrained in them from their very early development, while they are in the womb. It is a baby’s “own language,” Iannone said, and it is central to the company’s research.
The next step will be hitting the market, which Iannone told us Zoundream aims to do by the summer of this year. Firstly, they aim to provide their third-party machine learning software to third parties in Europe and Asia via already established partnerships with organizations and brands. “Over the last three years, we’ve been giving free devices that conduct cry detection and cry translation to whoever wants them,” Iannone said. “They get it for free, they can keep it for as long as they want, and we keep the data for research purposes.”
And what of baby data privacy? Are parents happy with Zoundream’s work so far? So far, “parents love the technology,” the company’s CEO told us, also explaining that the reason why they use it differs depending on the parent’s needs. “So they all like it for different reasons,” he said. “So, generally speaking, it’s not that they have no clue what the baby wants most of the time. Sometimes it is like that, particularly for [parents having] the first child. But in other situations, it’s, ‘I think I knew what the baby wanted, but I still like to get that confirmation.'”
“So it adds that reassurance. Because the first months are tough, they’re really tough” Iannone said, explaining that he is a father himself.