Microsoft's Songsmith ain't so bad after all

I’m am currently viewing Robert Scoble’s video of the two Microsoft Engineers who act in the Songsmith Commercial. Now for those of you who somehow missed this video, behold the wonders of Microsoft’s ad campaigns (who actually used a Mac covered with stickers. Go figure).

Ok, it looks bad, really bad, as if a smruf or a snorky produced it. Well first off, maybe I’m the exception but I can’t get it out of my head and have watched the ad more than once. They wanted it to go viral, and they succeeded marvelously.

My first feelings about the product were mixed – I found it to be a cool idea of course, imagining how complex the algorithms to make it work were, and I knew that yet again we had broken down another boundary towards our musical demise with robots generating music and all, but still, it looked so childish that I didn’t give it much thought.

Now I see things differently. Songsmith is on to something big but that stupid yet very effective ad doesn’t honor it’s capabilities. The video Scoble shot of those two engineers really grabbed my attention as they explain in much more detail how it works. Granted that commercials can’t last 20 minutes, but in those 4 very long minutes composing the ad, Microsoft could’ve demonstrated the good stuff instead of ridiculing themselves. Like the fact that the algorithms can easily scale the music from major to minor, or that you can actually alter the chord progressions and re-work the produced music. Also, the system pre-populates a media section where all you have to do to send your song to Youtube with whatever video or graphic display you want as a background is to click on a couple of buttons.

Actually I think this ‘childish like version’, which is probably going to be more popular at drunken parties than at pre-teen sleep-overs, is a spin-off to some more complex versions, and possibly some heavy-duty pro-audio software. If they can generate harmonies and melodies by analyzing ones voice, they can surely generate an instrument’s score for a mix. In a pro-audio context, they could let the producer tweak the sounds, import sound banks, link it up to virtual instruments etc. Re-tweak the results with Melodyne’s Celemony or Direct Note Access and you could really start getting creative. Another thing they don’t show you in the ad is that we can actually export a generated song’s MIDI files, tweak them in an audio host like a sequencer, and re-import them into Songsmith. Not sure many people are going to use this feature with such chipmunk-sounding instruments, but I appreciate the attention to detail, and in further releases this might seem more relevant.

I’ve been playing around with the demo version and I must say that from a musician’s point of view, there are some major downsides, like basically having no control over the instrumentation. We can change basic stuff like the chords, the swing, the tempo and the mood of the song, but I still felt pretty limited after 20 minutes – all styles only have one instrumentation preset making the experience kind of dull after a while (everything sounds the same). Then again this version was never meant for musicians, on the contrary, and in the end that’s a good thing. It opens up  music to the vast quantities of people who are absolutely positive they are tone deaf, kind of like that Beamz Music System we reviewed in September. Speaking of self-proclaimed tone-deaf people who could benefit from products like Songsmith, Andrew Dubber’s blog, ‘New Music Strategies’ has a very interesting post called ‘Amateur Nonsense‘ on the matter.

By youtubing ‘Songsmith’ you’ll encounter quite a few entertaining videos. Here are a couple I linked to my new Sqworl account (new bookmarking platform that enables users to create folders of bookmarks).

Enjoy.

About these ads