This project is our attempt to solve one of the United Nation's 17 Sustainable Development Goals using Google technologies, targetting quality education.
In 2015, the United Nations created 17 Sustainable Development Goals and aimed to achieve them by 2030. All 193 United Nations Member States agreed on these 17 goals to end poverty, ensure prosperity, and protect the planet.
When it comes to learning a new language, there are already plenty of material out there to learn about vocabulary and grammar, and to improve on listening, reading and writing.
However, when it comes to pronunciation, it gets rather difficult to learn and improve, especially if you're learning by yourself.
- A one-on-one tutoring is expensive.
- Getting a native to practise your pronunciation with may not be available on a frequent basis — more so if you're learning in a country that does not natively speak the language you're learning.
- Watching and listening to movies and other media is one option of self-learning pronunciation. However, this depends solely on the learner's ability to successfully distinguish the differences in their pronunciation from the source material. This task is rather difficult, since the learner's ears are adjusted to their own native language, and there is a high chance that they will recognise the sounds as the same, even though if heard by a native speaker of the target language, they might sound very different.
We would like to come up with a solution that helps users effectively learn and improve their pronunciation by themselves.
- Inform the users on the overall accuracy of their pronunciation in percentage.
- Evaluate all aspects of pronunciation (e.g. phonetics, stress, fluency, etc.).
- Identify which parts of which words were mispronounced.
- Point the users to (possibly external) material on how they can improve on each aspect.
Our plan is to develop an AI using supervised machine learning technology, that can evaluate the speech of a given text compared to how a native would speak.