What is your idea?
We are enabling automated assessments of children's reading skills by ‘listening’ to them reading aloud and analysing their speech using our proprietary speech recognition.
These assessments are used to enable personalisation of an online reading lesson for each child by adjusting the child’s learning path in real time, maintaining just the right degree of difficulty and leaving no gaps in understanding.
What problem are you solving and what is innovative about your approach?
Assessments are important when children are learning to read, around age 4-7, or when learning a new language.
Assessments can identify how the individual is progressing and whether they need more challenging material or extra learning support.
For a teacher, sitting with each child and listening to them read is time-consuming, and that can render regular or formative assessments both expensive and impractical in large classes.
Our technology can automatically capture and analyse a person’s speech while reading a self-directed text.
The software can both personalise the material to match reading ability and provides a screening tool for the teacher to identify children who could benefit from learning intervention.
Our speech recognition technology is built using our unique, large database of captured reading sounds from young children all over the world in a range of real acoustic environments and using a range of devices.
Our technology is unlike anything currently available as we address sub-word units such as phonic sounds, blended sounds as well as words - which addresses the synthetic phonics approach to reading used most commonly in schools today.
The complexity of the collected data forms the basis for robust and accurate speech-recognition software that can be used even in a reasonably noisy environment such as a classroom.
What’s the backstory here and how did you get involved?
My background is in speech-recognition technology research, which was the subject of my PhD, and in machine learning and commercialisation with Bell Labs and IBM Research.
The inspiration for this technology came from helping my daughter learn to read. I saw that the apps and web services on offer to build reading skills were basic and lacked a robust capability to assess and personalise the learner’s path.
I took a novel approach to developing and incubating the idea. Working with the Learnovate Centre and supported by EI’s Commercialisation Fund, I became the first ’spin-in entrepreneur’ to Trinity College.
This unique model has enabled Soapbox Labs to develop the speech-recognition technology in an environment that is designed to promote new innovations in learning technology.
The software is undergoing trials in schools in Ireland and we are talking with and partnering with several SMEs and publishing organisations.
How is this idea commercially attractive?
Our goal is to license the speech-recognition technology to publishing and educational companies, so our technology can provide a ‘smart engine’ to underpin products across many education sectors.
The demand for reading assessment and lesson personalisation is large – each year in English-speaking countries alone more than 13 million learn the basics of reading in schools and nearly 3 million children in speech therapy would benefit from our speech-recognition technology.
The market for English language learners - who can also benefit from interventions and support services - is larger still, with an estimated 6.5 million school students aged 5-10 in the US for whom English is not the language spoken at home. The potential market for English-language learning for young children in countries such as China and India is in the hundreds of millions.
Our long-term goal is to apply our technology to languages other than English, and then the possibilities are endless.
What are you looking for at the Big Ideas event?
We are looking for investment to explore new markets and develop partnerships with potential licensees.