Gerard & Anton winner Whispp wants to give people with speech impediments a strong new voice

Worldwide, 300 million people suffer from a voice impairment or severe stuttering. This has a significant impact on their daily lives, work and attitude to life. They also cannot make understandable and relaxed phone and video calls. With Whispp, they can. The startup’s assistive real-time speech technology and calling app converts whispered speech (people who severely stutter and then speak in a relaxed manner) and impaired speech (laryngeal cancer, vocal cord paralysis, spasmodic dysphonia) into a person’s clear, natural voice.

Joris Castermans – founder and CEO of Whispp – suffered from moderate to severe stuttering during his childhood. In high school, he was particularly pained by not being able to express himself. The idea for Whispp was born from two main findings. When they whisper, people with severe stuttering speak very fluently and relaxed. They also really dislike using the phone. So Whispp developed a speech technology and a phone app that converts this whispered speech into a person’s clear and natural voice in real time, i.e. without any noticeable delay. With LUMO Labs as one of their main investors, the Brainport innovation ecosystem opened up to the startup. We met Castermans to talk about ambitions, success and challenges.

What is your biggest goal or ambition?

“As our lives become more digital, there are more and more voice interfaces. Usually these are designed for healthy voices and are therefore not inclusive. For a large group of people (with voice disorders), communication is not always accessible. With our assistive technology, we want to advance the goal of ‘No voice left behind’. Our big, bold dream is to make Whispp’s assistive speech technology available on every smartphone and laptop worldwide to create a more inclusive world.”

And how do you want to achieve that?

“Whispp operates in the assistive technology space and positions itself as “Assistive Voice Technology”. Large technology and voice assistance companies predominantly focus on automatic speech recognition (ASR), also known as speech-to-text (STT) for non-standard speech. The disadvantage of this approach is the high latency of 2 to 5 seconds. This creates obstacles to natural conversation, as each time a sentence must be spoken and then text recognized. If the STT model makes mistakes, the wrong sentence is generated. Another disadvantage of the TTS model is that the intended intonation, pauses, emphasis on the words and emotions cannot be controlled by text alone.

In this scenario, current AI speech technology solutions are unable to provide an adequate solution for people with vocal impairments who have lost their voice but still have good articulation. With our real-time audio-to-audio based speech technology, Whispp has created a new product category and is filling this gap to improve the lives of a currently underserved group of 300 million people worldwide.”

In our latest magazine – IO Next: Gerard & Anton Awards – you’ll read about the 10 winners of the 10th edition of our very own awards ceremony. We look back at the 90 winners before them and look ahead to the impact these 100 winners will collectively have on the world.

How do you see the next five to ten years for Whispp?

“Speech AI technology improves communication in general, bridging the gap by making speech more accessible and understandable. Advances in speech recognition (and translation) will improve for low-resource languages ​​and dialects (by low-resource we mean languages ​​for which there is relatively little data available to train AI models). This will improve inclusion of people with different accents and various speech impairments. Advances in smaller, more optimized models will enable the integration of these systems into smartphones and edge devices, making speech technology more accessible than ever before.”

What has been Whispp’s biggest success so far?

“The fact that we are making a positive impact on people’s lives. In May we attended a symposium of the US patient organization Dysphonia International in Florida. There we met a group of 150 people suffering from various voice impairments. In an open mic session, many of them told their personal stories and several people were already using Whispp in their daily lives!”

…and the biggest challenge?

“We started our AI developments in early 2019, ahead of recent major AI developments. Whispp uses a different approach for its real-time audio-to-audio based AI; it first breaks audio down into different (deep) components and then combines them to create audio with different properties. This approach is inspired by the source filter theory of human speech production.

Although this approach is novel, for many years we were not sure whether this research approach would produce fruitful results. Thanks to the persistence of our creative AI team and our belief in the cause, we have managed to develop a revolutionary AI technology with a very low conversion latency.”

Gerard & Anton winner Dembrane aims to make democratic processes accessible to all types of organisations at all levels

Whether it’s a company, a neighborhood, or a sports club, knowing what’s going on is important to be innovative and grow.

The Gerard & Anton Awards are supported by EY, Rabobank, VO Patents & Trademarks, TWICE, Kadans Science Partner, Braventure, Lumo Labs, Gemeente Eindhoven, High Tech Campus, Philips, Goevaers & Znn. BV and DeepTechXL.