Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
A small company called Song Started with AI models last week, automate the music transcription and transform an audio file of a song into notes within a few minutes. The product that works on a freemium model is geared towards both professional and hobbyist musicians.
“We hope to make music players,” Andrew Carlins, CEO of Songscription and Stanfords MBA/MA, told Techcrunch in Stanfords MBA/MA. “We imagine a future [will be] able to get grades for the songs who want to play their students, [and] This music is specially arranged for the instruments of the band and offered at the individual game level of each student. “
At the start, the songs can transcribe music for various instruments, although the piano model is most reliable. In the future, the company hopes that various transcription outputs (e.g.
This type of product can be useful for a musician who records a song on which you work and then uploaded to get the grades. In this way you can skip the step to transcribe your work manually. And for those who cannot read and write grades, the songs will also create a piano role that shows a digital representation of the music that is played on a virtual piano.
Users can also automate the music transcription directly from YouTube links. If users upload a file, users must check a check box to confirm that they have the rights to transcribe the file. However, it would be easy to simply check the check box and to receive free grades for copyright protected songs.
“For music learners […] Since you listen to a song, write down the notes and list it on your home piano (as long as you do not calculate for a performance), it is not entirely clear that the use of a technically capable platform to offer you a lead exceeds a legal limit, although we understand that the field develops further and our application can enter a gray area, ”said Carlins.
The legality that we deal with these creative AI tools Recent court decisions seem to prefer technology companies to artists. However, Songscription does not create new music with a-generated music. It offers musicians a tool to accelerate the process to create their own guitar tabs or their grades.
“Since our platform enables users to edit the ratings, we position ourselves as extended music notation software that helps people accelerate the transcription process,” said Carlins.
The underlying architecture of the KI model from Songscription is based on a paper This co -founder Tim Beyer published together with researcher Angela Dai.
In order to receive the training data required for the creation of this type of AI model, the song script works with some musicians who were ready to share or sell their piano executions and notes. The company also used Public -Domain grades, although the majority of the training data is synthetic, said Carlins. In this case, Songscription would convert the grades into audio and then change the files to simulate real conditions with background noise or reverb.
Just seven months after it was founded, the songs of the song has collected prepared money from Reach Capital and will take part in Stanford’s Startx Accelerator.
