Alibaba Releases AI Music Generation Model HappyShrimp 1.0 in Beta Test

Alibaba has launched the beta version of HappyShrimp 1.0, an AI music generation model developed by its Alibaba Token Hub (ATH) business group, for users to turn vague ideas into a fully produced song.

HappyShrimp 1.0 allows users to generate complete tracks from a single prompt, whether based on an emotion, a story concept, or a target genre. The model can produce melody, arrangement, lyrics, and vocals without requiring technical music knowledge such as BPM, key signature, or instrumentation.

In its beta release, the model supports text-to-music (T2M) generation for both full vocal songs and instrumental compositions. Users can either create an entire track, including lyrics, from scratch, or provide their own lyrics and have the model generate the composition, arrangement, and vocal performance.

HappyShrimp 1.0 is built with broad musical knowledge to understand the creative aesthetics underlying various genres, regions, eras, and cultural contexts. It can interpret prompts such as “millennial Mandarin pop,” “the dawn-like sensibility of urban folk,” or “opera-style vocals in the chorus” and translate them into structured musical outputs. It demonstrates outstanding performance across genres including Chinese style (Zhongguo feng), pop, R&B/soul, hip hop, rock, funk, electronic, classical, and jazz.

Creator's Note: I simply asked for a K-pop girl-group track in the prompt, but HappyShrimp interpreted it well and independently crafted the melody, arrangement, vocals, and rap flows in the style of K-pop aesthetic. The melodies unfold smoothly, while the rap sections feature varied flows and deft rhythmic shifts in the bridge. Paired with an explosive drop, the arrangement instantly establishes a powerful rhythmic foundation.

Utilizing its world knowledge and music-domain reasoning, HappyShrimp 1.0 builds a structured representation of both the “grammar” of music — such as song structure, rhythmic development, and harmonic progression — and its “semantics,” including emotional tone, energy profile, and lyrical intent. This enables the model to fully comprehend user input and generate music that more accurately aligns with creative intent.

This gives the model precise controllability, enabling user-defined narrative flow, instrumentation, vocal style, and emotional dynamics to be accurately reflected in the generated tracks.

Creator's note: This Future Garage track is exceptionally refined in both timbral design and spatial treatment. The overall mix is clean and crystalline, with well-controlled dynamics and smooth, clearly defined transitions between sections. It not only showcases the futuristic, high-tech edge of electronic music, but also perfectly captures the cold, ethereal atmosphere of “glacial melting” described in the prompt. The piece has a strong documentary-like quality, making it especially well suited for videos featuring glaciers, snowfields, wildlife, nature exploration, or environmental themes, enhancing both immersion and epic scale.

HappyShrimp will collaborate with TAIHE MUSIC GROUP, a leading music service provider in China. Combining HappyShrimp's advanced AI technology with TAIHE's expertise in music ecosystem, the two will explore opportunities in initiatives such as artist co-creation, and high-quality content development.

HappyShrimp 1.0 can be accessed via official website https://www.happyshrimp.ai/

This article was originally published on Alizila written by Claire Mo

Community

Alibaba Releases AI Music Generation Model HappyShrimp 1.0 in Beta Test

Read previous post:

Read next post:

Alibaba Cloud Community

You may also like

Comments

Alibaba Cloud Community

Related Products

Alibaba Cloud Model Studio

Token Plan

Qwen

Alibaba Cloud for Generative AI