Our goal is to build the largest and most robust open-source speech corpus for African languages. To achieve this it is important for OpenVoice to contain different speech patterns across formal, informal and conversational speech. While there are a ton of open-source text corpora for conversational (informal) speech, corpora for formal speech (e.g Finance, Law) are harder to obtain. For this reason we will be happy to accept dataset contributions from individuals or organizations that cater to these niche corpora.

Email us at openvoice@ant.africa if you have a dataset you want to contribute.

QAVoice