The long road of French to English with a Machine.
In 2016 TranslateMe LTD was born in Mauritius, the company positioned itself as a niche translation service provider for hotels and restaurants. The company created an instant translation solution for menus. www.translateme.menu
The concept a first of its kind used Google Translation API to convert menus uploaded by the client into multiple languages. The product was an instant success but was short-lived by the apparent costs levied by Google Translation Service. The company was forced to adjust pricing and penalise the customer for usage that exceeded the monthly subscription. Ninety percent of the restaurant segment dropped subscription because of the adjusted pricing structure.
The company had a unique product but the costs were simply too high to achieve the goals set out by the founders.
In 2017 TranslateMe Network was created, more of a concept than an entity but the birth was spurred by the simple question “ Why does machine translation cost so much”
Since then the company worked hard to find answers and bend the rules, many said it would be impossible to build something that would rival the great tech companies like Google and Microsoft. The last two years the company has formed a team of respected experts in this space, people who believed our goals where possible.
2020 the company has made a major break threw with a particular model focused on French to English as a proof of concept. The tested results show hard work paying off finally. The AI model instantly translates French to English. It's at an acceptable level to be applied to general conversation. The company plans to set up a live web-based user interface to allow our french community to test the current results. They claim by the end of March 2020 a basic API will be available for BETA testing. The team will actively pursue BETA partners to test our API. These platforms would benefit by implementing the API in key areas. This will accelerate testing.
The results are sooner than expected and because the results can simply be replicated with machine learning for other languages its likely new models will be trained on popular languages like German, Spanish, Portuguese and Italian. These languages would be obvious choices because of the abundant data available. These models will need tweaking, the team will work on using its community to assist in correcting results and allow the team to expand on the data used to achieve these results.
TranslateMe Network is now positioned to produce various models that can be commercially used for testing within six months. Pushing for a complete commercial offer by end of 2020.
To review the latest test results we uploaded a file showing the input French and the output English of our AI model plus a version of the actual French translated to English by human translators to help with comparison please visit:https://docs.google.com/spreadsheets/d/1Z80C5cmxaryS5EzncBt-n37X4bqF-zVK2-elz7gpwLo/edit?usp=sharing
Please visit www.translateme.network for more information about our project.