Ziyouliangji seeks to utilize the AI music platform Hitto to make song creation accessible to everyone.

      As competition among large AI models gradually transitions from focusing on parameter optimization to real-world deployment capabilities, several Chinese AI startups concentrating on specific vertical scenarios are beginning to gain prominence. One such startup is Ziyouliangji Information Technology, which was established in 2023 and aims to transform how everyday individuals create music using AI.

      In contrast to many firms targeting general-purpose large models, Ziyouliangji has decided to specialize in the complex music sector. Its primary product, Hitto (YinChao), is an AI music creation platform driven by the company's own music foundation model. On Thursday, the company presented Hitto at the BEYOND Expo 2026, showcasing its latest AI-enhanced music creation functionalities.

      The platform allows users to generate complete songs by simply typing a sentence, uploading a photo, or conveying an emotion. Through this technology, the company aspires to eliminate obstacles to music creation, enabling even those without musical training to compose songs effortlessly.

      Ziyouliangji frequently speaks about the democratization of music. Jiang Tao, the company’s CTO, recounted his attempt eight years ago to compose a song for his wife as an anniversary gift, which left a lasting impression due to the complicated and costly traditional music production process. This experience influenced his decision to delve into the AI music domain.

      In 2024, as end-to-end music generation models advanced, Jiang gathered a diverse team that combined algorithmic expertise with musical knowledge to develop a proprietary music foundation model.

      AI music generation entails more complex technical challenges compared to text or image generation. It necessitates managing extensive context, melodic structures, and emotional conveyance, while Chinese songs also present unique linguistic characteristics such as tones and soft pronunciations. Ziyouliangji believes this complexity is a reason why international AI music models have struggled to fully penetrate the Chinese-language market.

      Technically, one of Hitto’s key strengths is its entirely self-developed pipeline. The team employed a hybrid AR+NAR architecture, which allows the model to maintain coherent song structures while also providing nuanced local details. The model is multimodal, able to comprehend text, images, audio, and even video inputs within a unified representative framework.

      With the recent Hitto V3.0 release, the team has made further enhancements to AI vocal performance quality. The model can now execute subtle singing techniques such as humming, vocal runs, and airy vocals, while also adjusting emotional expression in accordance with the lyrics.

      Simultaneously, the company is tackling a common challenge in AI music — producing songs that may sound polished but lack memorable hooks — by refining melody and arrangement generation to create more catchy and emotionally resonant songs.

      Currently, Ziyouliangji primarily serves everyday consumers. The platform has inspired numerous creative life examples: truck drivers turning poems from cigarette boxes into songs, families using photo-based song generation to capture their children's growth, and individuals expressing heartbreak musically. For many, music is increasingly seen not as a professional skill but as a medium for emotional expression.

      Beyond consumer entertainment, AI music is also entering commercial settings. Parts of the lyrics, composition, and vocals for "AI For Good," the English theme song of the 2025 World Artificial Intelligence Conference, were generated using the Hitto model. The company has begun partnerships with organizations in education, healthcare, and mental wellness to explore AI music's applications in therapeutic contexts and emotional support.

      As AI music advances into its next phase, industry competition is shifting from simply generating music to creating pieces that genuinely resonate with audiences. Ziyouliangji aims not only to develop a music generation tool but also to empower individuals to express themselves through music.

      In a time when AI is becoming ubiquitous, this focus on emotional connection and creative freedom marks a distinctive trend emerging among Chinese AI startups.

      Jessie Wu is a tech reporter based in Shanghai, covering consumer electronics, semiconductors, and the gaming sector for TechNode. You can reach her via e-mail at jessie.wu@technode.com.