Ok, imagine you are a sound designer (if you are not one already). You walk into your studio in the morning, prepare yourself a nice cup of steamy coffee and sit down in front of your computer screen. First task for today: The location recording of an outdoor dialog scene is useless, because nearby stationed jet fighters thought it would be a wonderful day to do some low-altitude flybys. No big deal, the re-recording engineer already got the dialog. It’s now your task to prepare the rest of the auditory scene. Some footsteps, some car sounds in the back, maybe a bit of bird singing will do.

You take a sip of your coffee and fire up your sound effect database. It holds over 2 terabytes of sounds, so you’re feeling more than prepared to do the job. You go for the footsteps first: You type “footstep”. Your database gives you close to “footstep tarmac“. Your database gives you 300 results. You type “footstep tarmac male”. “footstep tarmac male close” gives you 20 results. You stop typing, and start clicking the results one by one to listen to what your database thinks you should be using for this scene. After going through half of the results, you get bored. Of course, as you already listened to ten minutes of footstep recordings. All have, however, sounded too direct. You could add some effect to achieve the sound you have in mind. But you know that there was this one recording that you listened to the other week, which would just be perfect. So you go back to your search field, you delete the “close”. Back to 100 results. Some of which you already listened to. You’re adding “-close”. Wohooo! Logic Operators, but still 80 results. You go through some of them again, already leaving out a few in-between. You take one recording which you think suits so-so, and decide to fix the rest later on in the mix.

The car is next. You type “car”, you get more than 2000 results. You take the third recording out of that list, the one you always take if you need a car driving by because you just can’t be bothered going through the car-list. Moving on to the birds. Luckily you just bought another sound library of nature sounds. You insert the first DVD and copy the files to your hard disk, but already while the database is newly indexed you read that your newest weapon contains over 10.000 recordings of nearly every bird that mankind ever knew of. With fear in your eyes you decide go out for a while, it’s lunchtime anyway. You take a walk in the park and accidently record some little birds with your mobile phone. 64kBps mono MP3, but at least one unique sound. You spend the rest of the day wondering why there is just no way to ever listen to all that sounds in your database again. You’re actually so bored going through lists and lists of sounds every day, that you decide to record more sounds yourself, just when you need them. You decide to stop thinking about how much money you spent on your database.

If this sounds like the blueprint of your daily work, then you might want to take a look at SoundTorch: This revolutionary interface allows you to fly through your audio collection. You can lay out your entire audio collection in front of you, or just a small portion, for example 100 search results, or the content of a folder. All sounds you hover over with the SoundTorch are played back simultaniously on a surround sound system. You can easily determine which sound is which, as they sound from different directions, depending on their position on the screen and the position of the SoundTorch. All sounds have an individual icon, which spins when the sound is played back, just like a disk on a record player. The rotational speed shows whether you're listening to long recording, or just a one-shot sample. Further, the shape of the icon essentially is the sound's waveform - just circularly wrapped. The sounds are grouped by similarity. Not by the similarity of their description, however by how similar they sound. Essentially the same algorithm that SonoSketch uses to search through your database is used by SoundTorch to determine where the icon of a sound should be placed on the screen.

SoundTorch is an auditory interface. Yes, it does have nice visuals, but you essentially could use it with your eyes closed, you just have to listen. Just naturally our brains can process hundreds of sounds simultaniously, because that is what they essentially do all day. The one sound that you're interested in will automatically stick out, you won't be able to do anything against it. This psychological effect is called “Cocktail Party Effect”. The interface allows you to zoom closer into a group of sounds, and you can adjust the diameter of the torch, just like a real flashlight. Everythings seems just natural. Using SoundTroch, the process of listening to 100 search results is sped up to just a handful of seconds. You can finally listen to all the sounds in your collection again, and select the one you like best based on your creativity, not your boredom.

For further information, please watch this Google TechTalk, or this Youtube Video.