DiveDeepAI

Langalore

DiveDeepAI developed an MVP based system keyboard for a translation app. The app is a new idea of learning languages e.g., French through the translation of their in daily life word typing in different phone applications.

About Client

A stealth start-up that wanted that wanted to make a mobile app that a user can use to learn other languages.

The Challenge

The team initially started working on iOS but there were security concerns, so they had to shift on android and start again. As the app needs, a suggestion bar on type and team had to design a system keyboard so all this demanded a huge amount of manual effort and time of the developers. Each letter even space bars and punctuation symbols e.g., comma, periods, question mark etc. all had to be manually added from scratch and had to be made dynamic. Moreover, the keyboard must be able to effectively work in different screen sizes without any glitches.

The Solution

It is an efficient system keyboard that is designed to be usable for all the apps in the phone e.g., WhatsApp, Facebook messenger, messages, etc. It is a customized system keyboard that has a model deployed at the backend which translates a word the user types in their daily life conversations. Once the user types and click translate, the model picks a random word and shows the translation of the word in the suggestion bar and highlights the word for which the translation is displayed. If the user chooses to use the word, the model saves it and when next time the user types that word, the model reminds the user that you have already used the word now you tell the translation and when the user types the translation the model verifies it too. This helps the user learn daily life words of other languages easily.

The Impact

The application facilitates society to learn other languages through English translation from their daily life words.

HIGHLIGHTS

Project in Action

Selecting a Machine Learning Algorithm for Translation

Back-end API development

Android Keyboard Development

Deployments