Researchers develop AI algorithm that trains violinists

A wearable can track and record the exact movements made by violinists and provide real-time feedback

Spotted: Researchers at Barcelona’s Universitat Pompeu Fabra have developed an AI algorithm that provides violin players with real-time feedback on their technique. They used the Myo gesture control armband (now discontinued) in their research. The armband was able to track and record the exact movements made by a professional violinist as she played. Audio of the performance was recorded at the same time.

Once the movements were synchronised to the audio data of seven different bowing techniques, the researchers trained an AI algorithm to identify which bowing techniques matched which sounds. In tests, the model was able to correctly identify the technique used to produce a given sound 94 percent of the time.

The study was conducted as part of the TELMI (Technology Enhanced Learning Performance of Musical Instrument) project. The goal of the EU-funded project is to develop a suite of technological innovations to enhance musical learning, including the creation of “interactive music learning prototypes”.

Takeaway: The study was conducted as part of the TELMI (Technology Enhanced Learning Performance of Musical Instrument) project. The goal of the EU-funded project is to develop a suite of technological innovations to enhance musical learning, including the creation of “interactive music learning prototypes”. Wearables have previously been used to aid in music creation and music mixing. However, the Barcelona researchers hope the wearable wristband can eventually help music students develop better technique by providing them with feedback in real time. The AI algorithm could also allow students to compare their performance with those of professional violinists. In the future, the researchers believe their technique could also be applies to other instruments.