Corporate
This technology analyzes a music audio signal to extract its beat positions, chord progression, and music structure, such as verse and chorus.
The analysis algorithms use cutting-edge machine learning to achieve speed and accuracy.
This technology runs on a variety of environments, ranging from cloud servers to embedded systems.
You can experience the technology in several of Yamaha’s apps, such as Chord Tracker.
To recognize the elements of music, the technology uses probabilistic models and deep learning models. In these kinds of models, highly accurate analysis is possible thanks to the acquisition of a large dataset for training. For example, to train the chord recognizer we use not only an in-house dataset of chord labels, but also those corrected by users (with the user’s consent).
