Uploaded September 2026 | Updated November 2026, 2 days ago
Voice AI creates a demanding data infrastructure problem: models need enormous volumes of audio data while capturing context, emotion, acoustic signals, and the nuances of human conversation.
In this Backblaze Investor Day fireside chat, Backblaze CRO Anuj Kumar sits down with Olya Ossipova from Hume AI to discuss Hume’s work advancing voice AI and the data infrastructure required to build and improve multimodal models.
The conversation explores why voice presents a richer data challenge than text alone, and how Hume collects, processes, consolidates, and uses large datasets throughout model development. The discussion offers a practical look at how AI companies are managing the growing volume and complexity of multimodal data.
Voice AI creates a demanding data infrastructure problem: models need enormous volumes of audio data while capturing context, emotion, acoustic signals, and the nuances of human conversation.
In this Backblaze Investor Day fireside chat, Backblaze CRO Anuj Kumar sits down with Olya Ossipova from Hume AI to discuss Hume’s work advancing voice AI and the data infrastructure required to build and improve multimodal models.
The conversation explores why voice presents a richer data challenge than text alone, and how Hume collects, processes, consolidates, and uses large datasets throughout model development. The discussion offers a practical look at how AI companies are managing the growing volume and complexity of multimodal data.










