Single- and multi-speaker Tagalog and Cebuano voice data

The agreement followed an approximately six-month procurement, legal, security and sample-evaluation process. Silencio says the data could not have been assembled from public sources. Silencio also disclosed a separate, value-undisclosed licensing agreement covering Swahili, Amharic, Hausa, Wolof and Yoruba datasets for another unnamed AI developer. Its contributor network covers over 185 countries, with 250,000-plus licensing-ready hours. Corporate announcement. This is direct evidence that scarce linguistic and geographic representation converts into commercial value when the required data does not exist online.