Full-duplex Datasets Company

Voice AI should be emotionally intelligent.

We believe voice AI is more than a tool for completing tasks.
It should be a partner that understands human emotions, sensitivities, and inner states —
one that truly supports the human mind.

Today, most AI systems are designed to solve problems and optimize efficiency.
But voice is different.
Voice alone can carry emotion, intention, silence, hesitation, and meaning beyond words.

In the film Her, the AI Samantha is portrayed as a psychological partner —
someone who listens, understands, and grows alongside a human.
We see this not as fiction, but as the ultimate destination of voice AI.

In the near future, voice AI will replace roles such as customer support, teaching, and voice acting.
But that is only the beginning.

The true potential of voice AI lies in a domain technology has barely touched:
the human emotional world.

We are currently focused on building the world's most emotionally rich voice AI
by collecting large-scale, real-world conversational datasets.

Just as data is fundamental for text-based AI,
real human voice is the most critical resource for voice AI.
Yet unlike text, voices disappear.
They are not archived by default.

Human voice is one of the rarest forms of human output —
ephemeral, emotional, and largely lost throughout history.

Today, only tens of thousands of hours of real conversational voice data exist globally.
We aim to expand this to tens of millions of hours.

This mission requires people from all over the world —
across languages, accents, professions, and cultural backgrounds.
No single company or country can achieve this alone.

Imagine a world where everyone has a companion
that grows with them from birth,
understands their emotions,
and supports them through both joy and suffering.

Such a presence could become a profound source of support —
not just for individuals, but for humanity as a whole.

Join us in building the emotional foundation of the next generation of AI.