An app for musicians that allows you to search for a song for free and watch the video with the transcription of the chords. It's called Yalp.io and it's a project developed by Nami Lab, a startup from Monza that deals with machine learning and artificial intelligence applied to creative contexts.
“Since we created the beta version of Yalp.io, our first product, growth has been exponential – explains Simone Geravini, 34, computer scientist and musician at the helm of the company – and we have had excellent feedback especially from abroad: 80% of our traffic comes from Japan”.
The new app does not violate copyrights, because "only the melody and words of a song are subject to copyright - Geravini underlines -, while the harmonic structure, i.e. the set of chords, is not subject to these rules , and for this reason we can publish the chord symbols of any song. For our project, therefore, we use information that is not subject to copyright, while the audio comes from YouTube's servers, so it is a very flexible and scalable system”.
Yalp.io is the only Italian project that uses this technology applied to music and worldwide only two other companies (one Israeli and one Dutch) operate in the same field.
“Our users are musicians or simple enthusiasts, amateurs and aspiring professionals: we set no limits – the founders of Nami Lab write in a note -. At the moment the technology we have developed is not 100% perfect, but it has a degree of approximation that is around 85%. We are trying to improve more and more, correcting machine errors with new artificial intelligence algorithms based on neural networks. The goal is to come close to perfection, that is to be able to have a 95% accuracy percentage of the transcripts.
On the financial side, in February Nami Lab closed a deal "for 250 thousand euros with the Geode Group, made up of Massimo Orlandi, Riccardo Bani and Egidio Ricciuti, founders of Sorgenia - continues Geravini -, who joined our company as financing partners. They believed in our project and, knowing that we deal with Neural Networks, Deep Learning, Topological Data Analysis and Machine Learning, they decided to trust us".
The next goal is to try to build a database of one million analyzed songs.
