Amharic is an important language of Ethiopia and is commonly written using the Ethiopic script. AI data projects should confirm the target script, text encoding, punctuation conventions, and normalization rules before creating or transforming text data.
Text handling
Spellings, punctuation, and tokenization choices can affect search, language modeling, and evaluation. Document the expected representation and retain source material where appropriate so normalization decisions can be traced. For annotation, provide examples of names, borrowed terms, and ambiguous cases relevant to the domain.
Speech data
A speech project should specify speaker criteria, recording conditions, consent, audio formats, transcript conventions, and the target use case. If regional speech differences matter, state how the sampling plan will represent them and how reviewers should label uncertain segments.
For Amharic language data collection or transcription, prospective buyers should include the task, script and audio requirements, sample size, domain, and quality criteria. Langigs can assess a proposed scope, but availability must be confirmed rather than assumed.
