Music Analysis Technology
What is Music Analysis Technology?
This technology is used to analyze the elements of a song, by recognizing the beat positions, the chord progression, and the song structure (verses and chorus) in its audio data. Even if there is no sheet music or chord chart for a song, as long as you have the audio data you can use this technology to immediately create a chord chart. Since this technology allows the elements of a song to be grasped instantly, it also supports creative activities by providing a more efficient music production process.
Music analysis technology uses the following steps to analyze audio data.
1. Estimate beat positions in audio data
2. Use AI to analyze the elements of a song from features based on the names of musical notes, etc.
3. Estimate beat positions, chord progression, song structure, etc.
Applications
This technology employs advanced analysis algorithms based on cutting-edge machine learning to achieve high speed and high accuracy. In addition, it can be used in a variety of operating environments and on servers, and embedded in mobile applications. You can experience this technology with Chord Tracker.
Training and Analysis Mechanisms
This technology uses probabilistic models and deep learning models to analyze the elements of a song. In order to achieve a highly accurate analysis, the models must be trained on a large dataset. So, for models that analyze chord progressions, for example, in addition to the Yamaha chord label dataset, we also collect chord labels corrected by users on applications such as Chord Tracker to ensure a large dataset, which we use to train the models.
