Songscription will launch “Shazam for Sheet Music” with AI

A small company called a telephone song Last week it was launched on an AI model that automates the transcription of music, and converted the audio file of the song into sheet music within minutes. Operated on a freemium model, this product is aimed at both professional and hobby musicians.

“We hope to make music playing more enjoyable,” Andrew Carlins, CEO of Songscription and student at Stanford’s MBA/MA Education Program, told TechCrunch. “Imagine a future where students can get sheet music for the songs they want to play in rural Nebraska high school band teachers (Imagine a future where students can get sheet music for the songs they want to play.

At launch, SongScript can transfer music from several different instruments, but piano models are the most reliable. In the future, the company hopes to add a different transcription output (such as a guitar tab) and a full band arrangement in contrast to one instrument.

This kind of product can be useful for musicians who record the song they are working on and upload it to get the sheet music. This will allow you to skip the steps that require you to manually transfer the work. Songscript also generates piano rolls for those who cannot read and write sheet music. This shows a digital representation of the music being played on a virtual piano.

Users can also automate music transcription directly from YouTube links. To upload a file, you need to check the box to make sure that the user has the right to transcribe the file, but it’s easy to check the box to get free sheet music for copyrighted songs.

“For music learners (…) they are allowed to listen to songs, write down notes with their ears and perform on their piano at home (unless they charge performance), so it is not entirely clear that giving a head start using a tech-enabled platform is beyond legal boundaries.

However, the legality of how to engage with these creative AI tools is controversial. Recent Court decisions He seems to support high-tech companies more than artists. However, rather than creating new generated music, Songscription offers tools to speed up the process of creating your own guitar tabs and sheet music.

“Our platform allows users to edit scores, positioning themselves as an enhanced music notation software that helps people speed up the transcription process,” Carlins said.

The underlying architecture of SongScript’s AI model is based on paper Its co-founder Tim Beyer presented with researcher Angela Dye.

To obtain the training data needed to create this type of AI model, SongScript works with musicians who are willing to share or sell piano performances and sheet music. The company also used public domain sheet music, but the majority of the training data is synthetic, Carlin said. In that case, Songscription converts the score to audio, changes the file, simulates the actual conditions with background noise or reverb.

Just seven months after its establishment, Songscription raises pre-seed money from Reach Capital and joins Stanford’s StartX Accelerator.