C3NL

vocalXplorer-01

Our team is currently developing a novel phone application – the vocalXplorer.

On this website, we want to keep you up to date on new developments of our app and showcase our latest prototype.

If you have any questions about the project or ideas on how to improve our app, please get in touch with us.

Enjoy!

 

Project summary

Driven by the latest neuroscientific findings, we are developing an interactive phone application – the vocalXplorer providing immediate and customised auditory feedback with the goal of improving native-like second language pronunciation.

blipsen 

Science background

This short clip gives you insights into our scientific findings that paved the way to the idea of our app:

And in case you want to do more background reading, here are the links to the publications featured in the video: Simmonds et al. 2014 and Simmonds 2015.

blipsen 

vocalXplorer

Our app is unique in two fundamental ways:

  1. Our app exclusively focuses on promoting native-like pronunciation of second language learners by using a novel interactive auditory feedback technique that is informed by latest neuroscience research. The online auditory feedback aims to optimize the user’s variability in pronunciation in order to engage the vocal learning pathway for longer, resulting in closer to native-like pronunciation than possible with any existing app or teaching method.
  2. The app will be automatically tailored to each user specifically using state-of-the art machine-learning techniques. This customisation ensures that each user is consistently presented with auditory feedback that best suits their optimal vocal learning strategy. This aspect is crucial to the success of the app, as we hope to engage with a highly heterogeneous group of users, each of which will come from a variety of distinct backgrounds and require a tailored exercise regime.

 

FigureWebsite-01

blipsen 

Technical details

App development framework

Our app is being developed using Unity 5. Unity is a flexible and powerful development platform for creating multi-platform (iOS, Android, Blackberry, OS X, Windows, Linux, …) interactive experiences. It is a complete ecosystem for anyone who aims to build a business on creating high-end content and connecting to their potential customers.

Machine-learning technique

The machine-learning part of our app is based on recent work by us (Lorenz et al. 2016, Lorenz et al. 2015), in which state-of-the-art Bayesian optimization methods have been employed to automatically design neuroimaging experiments in real-time. Bayesian optimization techniques have established themselves as a practical and efficient tool through which to efficiently optimize difficult functions. They are attractive due to their flexibility and as a result are now frequently employed in settings where traditional optimization methods are not feasible. We have leveraged Bayesian optimization in order to effectively learn the optimal auditory feedback which yields the largest variability of pronunciation for each user individually. This is fundamental, as users will come from a variety of distinct backgrounds and will therefore respond very differently to exercises.

blipsen 

Prototype

Please explore our latest app prototype. We are curious to hear your feedback!

Screenshot 2016-01-25 14.32.14

Screenshot from latest app prototype

A download version of our app prototype is available for OS X and Windows here:

blipsen 

Our team

Our team is composed of several mature researchers who have an extensive track record of successful collaboration:

Nips Photo

 

Ricardo Pio Monti is a PhD student within the Statistics Department of Imperial College London. He has extensive background in developing novel statistical methods for various applications. His current work focuses on machine learning, in particular Bayesian optimization techniques. Further, he is an advanced programmer. Get in touch!

 

 

RomyLorenz

Romy Lorenz is a PhD student within the Division of Brain Sciences and the Department of Biomedical Engineering at Imperial College London. Her research focuses on the development of novel closed-loop brain-computer interfaces using state-of-the art machine-learning techniques. She also holds a master’s in Human Factors, providing her with expertise in developing software with improved usability and high-quality User Experience – essential for the commercial success of any app. Get in touch!

 

 

C0086519 Anna Simmonds

 

Dr Anna Simmonds is currently a research associate within the Division of Brain Sciences at Imperial College London. Her experience as a languages teacher led to a greater interest in the neurological aspects of language processing. Her research investigates the neural mechanics of foreign language acquisition. Her scientific findings paved the way to the development of our app. Get in touch!

 

 

Leech_Picture

Dr Robert Leech is senior lecturer within the Division of Brain Sciences at Imperial College London. His research is inherently multi-disciplinary, integrating neuroscience, psychology and computer science to better understand the brain. One of his core interests is the development of computational models simulating neural dynamics (e.g. here, Firefox/Safari only) using the industry-leading cross-platform game engine Unity. Unity is also used for our app. Get in touch!