Wechat built its initial traction by having voice memos as a first class interface element in its app. There are entire communities of illiterate Chinese communicating using purely voice memos/stickers.
Yes, but are voice memos indexable and searchable? Maybe they can be, but don't know what enterprising startup would find it profitable to create the machine learning algorithms necessary to do this successfully. As a well-funded nonprofit, maybe. Because of illiteracy, your metadata would also need to be a significant amount of audio, at least for the illiterate humans.