Mulbert
Enhanced AI Music Generation Model Delivers 25% Boost
Media / Entertainment / Gaming
Highlights
CloudWerx elevated Customer by modernizing their AI music generation using Google's newly released MusicLM. Through a Vertex landing zone, CloudWerx crafted a custom NN-based music generation model, outperforming a competitor's model in an A/B test. The deployed MusicGen model showcased a 25% improvement in generation speed and customizability.
The Challenge
Customer struggled with slow generation and lack of customization on their current ML model. They wanted to improve their music generations in terms of time taken for generation and customizability.
The Solution
CloudWerx enabled the customer with a Vertex landing zone to build a POC for a NN based music generation model both to train on the customer’s data and also build an inference pipeline. CloudWerx crafted a custom NN-based music generation model, outperforming a competitor's model in an A/B test.
The Implementation Process
CloudWerx transformed Customer's music generation with Google's MusicLM, achieving a 25% boost in speed and customizability.