Hitto Tests China’s Vertical AI Push In Music Generation

Ziyouliangji Information Technology is using its Hitto platform to test a narrower question in China's generative AI market: can a vertical foundation model create music that ordinary users want to keep using?

The startup, founded in 2023, showed Hitto at BEYOND Expo 2026 as an AI music creation platform built on its own music foundation model.

The product lets users generate songs by entering a sentence, uploading a photo or describing an emotion.

The Product Move

Hitto is not positioned as a general chatbot or image tool.

Ziyouliangji has chosen music, a field where language, melody, rhythm and emotion have to work together.

The company argues that the market is moving away from competition based only on model size and toward products that solve specific creative problems.

That focus gives Hitto a clearer use case than many general AI applications.

The goal is to lower the cost and skill barrier for people who want to turn personal memories, messages or feelings into songs.

Technology Promise

The company says Hitto is built on a self-developed pipeline using a hybrid AR+NAR architecture.

The aim is to keep a coherent song structure while improving local details such as phrasing and vocal expression.

The model is also described as multimodal, with the ability to understand text, images, audio and video inputs in a shared representation space.

In Hitto V3.0, Ziyouliangji says it improved AI vocal quality, including humming, vocal runs and breathy vocals.

It also worked on melody and arrangement generation to avoid a common weakness in AI music: tracks that sound polished but fail to create memorable hooks.

User And Commercial Signals

The platform's current user base is mainly consumers.

TechNode described examples such as truck drivers turning poems into songs, families using photo-based song generation for children's memories and users converting heartbreak into music.

Those cases show why Ziyouliangji is framing music as emotional expression rather than only entertainment.

There are also early commercial tests.

The English theme song AI For Good for the 2025 World Artificial Intelligence Conference partly used the Hitto model for lyrics, composition and vocals.

The company has also started exploring education, healthcare and mental wellness partnerships.

What To Watch

The next test is retention.

Many generative AI products can impress users once, but music tools need repeat use, rights clarity and songs that listeners actually replay.

For Hitto, the signal to watch is whether its vertical model can turn personal expression into durable user behavior, not just one-time novelty.