Did you know it? The birth rate in a country such as Senegal is 4 times higher than in a European country such as France.

With the increasing growth of mobile ownership, in such a country, artificial intelligence makes it easier to provide emergency assistance to pregnant women.

Umbaji with another start-up is developing a mobile app to respond to the need of pregnant women to get the proper advice at the right time.

Umbaji plays a core role in the deployment of this service by its intelligent translator which translates requests from local languages to the official language.

The rise of computer vision and deepLearning makes it possible to develop new models for the speech recognizer. The example below briefly explains a basic technic used to build the model.

Wave form- "No"  in Ewe

Wave form "No" in English

The concepts behind Umbaji'speech recognizer lies in the nature of sound itself. Samples from the dataset are used to generate a word-to-waveform mapping. At this stage, various techniques can be used to process the sound including using a DFT or a CFT  and then building a classifier. The classifier can be built using a CNN, and time-series for sentence recognition.
The accuracy of the model surpasses 90 % and is predicted to increase with the available data.