Amharic Texts 

We have collected about 1 million Amharic sentences that are used mainly to investigate unsupervised POS tagging process.

The datasets can be further be used to build small-scale word embeddings as well as a reference text for speech-to-text and MT experiments.

Since the texts are not yet publicly published, contact us to get the passwords to download the texts

1. Original texts split into sentences here
2. Normalized sentences based on basic Amharic character/Fidel normalization here