Tuesday, January 20, 2009

Microsoft Songsmith and the future of music

Over the last few days, Microsoft Research's promotional video for SongSmith has been almost-universally panned. Take this example from Videogum (h/t edythe):

...this REAL commercial for Microsoft's new Songsmith software (you sing at it and it creates horrible musak to accompany you) is completely insane. Not only is it apparently earnest and not a parody, self- or otherwise, it seems like it comes from a bizarro parallel universe where irony was never discovered.

Unfortunately, the video (which didn't realize demonstrate how Songsmith actually works) helped to result in some negative reaction for the real capabilities of SongSmith.

This thing sucks. Is this supposed to be the answer to Garage Band?
I know it's not quite like GarageBand, you sing into it and it plays along, but the sounds on this thing suck (at least from that Video, which is probably the best they have).

I wonder what the reaction would have been if, instead of producing that cheesy video, Microsoft would have instead gotten someone up and had them share the thought behind the research project:

SongSmith is a Microsoft research project designed to narrow down, as much as possible, the gap between writing songs and writing music. Developed by Dan Morris and Sumit Basu, primary researchers at Microsoft Research, SongSmith is now available both as a trial version for free and for purchase through the Microsoft Store, for just $29.95. The solution is capable of building the complete musical accompaniment around a song “performed” for SongSmith, based only on the voice of the users. According to Microsoft, the project is aimed at musical novices and aspiring songwriters, as professionals will undoubtedly already posses the skills that SongSmith is designed to compensate for.

“Sumit and I are hobbyist musicians, and although neither of us is going to write the next hit single, we both have fun working on new songs,” Morris explained. “We saw an opportunity to use our backgrounds in computer science to give everyone a first taste of songwriting and to let everyone have fun making original music.”

Basu was quoted on the implications of the software:

"I remember my sister and I being in the car. I would be singing something, and my parents would say, ‘What song is that?’ My sister would say, ‘Sumit is just making it up,’ and they’d say, ‘Well, you should sing a song that actually exists.’ Before we learn that we’re not supposed to create music or we’re not able to do that, it’s something that exists in our hearts.

"It’s not very pleasant to listen to someone sing in isolation, though. Having a backing track that fits with what they’re trying to sing makes for a better overall experience. That is going to be our biggest contribution, that empowerment to let people go from having a musical idea or thought or melody to something they’d actually want to show other people."

And, once they got over the promotional video, people have been experimenting with SongSmith, most notably by taking lead vocal tracks from popular songs, running them through SongSmith, and seeing what happens.

So far my favorite Songsmith-processed song is "Wonderwall" by Oasis. Basically, in this and other instances, the lead vocals are taken from a popular song and run through Songsmith, just to see what happens.

Of course, this is just version 1.0 of the product. (Or, since it's research, version 0.99.) What happens when Microsoft gets to version 3 of this product? What are the implications for the creation of music?
blog comments powered by Disqus