Twi Machine Translation

French is a strategically and economically important language in the regions where African Twi spoken. However, only very small proportion of speakers Ghana speak French. The development Twi–French parallel corpus corresponding machine translation applications would provide various advantages, including stimulating trade job creation, supporting Ghanaian diaspora French-speaking nations, assisting tourists immigrants seeking medical care Ghana, facilitating numerous downstream natural processing tasks. Since there are hardly any systems or corpora between that cover modern versatile vocabulary, our goal was to extend Twi–English with develop French: Consequently, this paper, we present 10,708 sentences. Furthermore, describe experiments corpus. We investigated direct cascading use English as pivot language. Our best system state-of-the-art transformer-based achieves BLEU score 0.76. French–Twi system, which uses language, results 0.81. Both fine tuned corpus, even slightly outperforms Google Translate on test set by 7% relative.