[AI Times] Emotionwave Introduces AI Music Data Generation and Processing Process

▲ Diagram of Emotionwave Dataset Construction and Processing Process and Artificial Intelligence Model YOURMT3+ (Photo=Emotionwave)

Emotionwave, a specialist in artificial intelligence (AI) media technology (CEO Rio Jang), announced on the 13th that it participated in the '2024 Data Promotion Week' held at the Fairmont Ambassador Seoul Grand Ballroom in Yeouido, unveiling the process and results of valuing AI music data, as well as the latest AI technologies and processes for data generation and processing.
The presentation given by Emotionwave CEO Rio Jang was selected as an excellent case study for data value assessment. CEO Jang explained the systematic creation process of the AI music data built by Emotionwave and the impact of this data on service and product innovation. In particular, he emphasized the importance of AI technology for data accuracy and refinement, presenting concrete ways to enhance the competitiveness of technology companies.
First, they revealed that they introduced 'Cross Dataset Stem Augmentation' and 'Intra-Stem Augmentation' technologies to improve data accuracy. They explained that this enables the integrated use of various music datasets and effectively solves issues of data imbalance and insufficient annotation.
They also reported enhancing the universality and accuracy of the AI model by implementing a 'real-time data augmentation technique,' such as pitch shifting in audio sources, to realize diverse data scenarios.
It was also mentioned that technology developed by Emotionwave Director Sang Sung-kyun (CMTO) (YourMT3+) was applied. This technology improves the analysis accuracy and efficiency of data through a Multi-T5 multi-channel decoder, making it possible to convert into single or multi-track sheet music based on PCM data. They reported efficiently processing complex music data and improving data quality.
It also explained that it applied this to the processing process for creating datasets. Through this, it conveyed that it has built a proprietary AI data and core data processing process for commercialization, and is being recognized for its technical value in the industry.
Rio Jang, CEO of Emotionwave, said, "The core competency that allows us to look further into the future—5 or 10 years as a technology company—lies in data," adding, "I want to emphasize that the beginning and end of information technology is data."
Original article: AI Times Reporter Jang Se-min semim99@aitimes.com