Core Bounty/Contest: $150USD for FFT Digital Vinyl Macro

So it's about time I put my money where my mouth is, it's not a lot of money but it's more than anyone else is offering for a community effort and it's a start.

Here's the deal: As you guys may have seen I recently uploaded 'ammobox' Reaktor's first digital vinyl reader to the user library, it's crude but it works -- for documentation on how it works you can check out my wiki entry on it at http://www.remix.net/wiki/ammobox

This is where you guys come in (If you're up to the challenge)..

RULES:

• Create a digital vinyl reader that uses FFTs to give a more statistically reliable read of time-code.
  
• The read data need only be record direction and record speed, absolute positioning isn't necessary to earn the bounty.
  
• It must be written entirely in Core
  
• Running CPU usage must not exceed 35% on a 2.4ghz dual-core processor. I will allow a bit of leeway here, but not much.
  
• Latency must be low enough such that 'scratching' is possible, somewhere on the order of 4-6ms
  
• It must have 'universal' time-code support, this is not a difficult requirement as stated in my wiki article.
  
• It must have error correction such that if you raise the needle off the record, the ambient vibration will not supply bogus speed data.
  
• Output of the macro must be essentially identical to 'HangTheDJ', ~1.0 for 33rpm forward, ~-1.0 for 33rpm backwards, ranges inbetween and over depending on record velocity and direction. Speed does not need to be 'rock solid', a bit of flutter is fine as long as it's within ranges that don't negatively affect sound playback for scratching samples.
  
• You will retain the intellectual property rights to your component, but you MUST upload it to the user library and allow other registered Reaktor users free access to it and to include it in derivative works (such as ammobox)
  
• New rule: It has to be at least as accurate as the existing zero-crossing technique in HangTheDJ -- no arguments no BS. If you piece together something that isn't better, you don't win. I doubt this will be a problem but I just want to make it clear we're here to make things better.
  

Whoever meets these criteria first will be awarded the $150USD via paypal or via money order if you choose. I live in the US but will be happy to send it to you overseas if necessary.

Now, I realize this is a long shot, I want to see ammobox improved and if I continue working on it solo it will take a long time for me to get the FFT stuff sorted out, $150USD isn't much but it's a commitment from me to the community, I believe in Reaktor and I believe in rewarding people who help out.

I want to make ammobox a real competitor for Ms. Pinky and I want it to remain free.

If nobody steps up to the plate, I'll just figure this out on my own, no biggie.

-Nathan Ramella

 

I would like to note that all donations being collected on behalf of ammobox are going to be put towards this contest. Currently I have received $40 from nice people who appreciate the work, so if I don't get any more the other $110 will come out of my pocket.

If I recieve more than $150 in donations, all of that money will go to the contest and I will update the thread accordingly.

Consider the gauntlet thrown down.

 

I don’t know if you will want to keep this post here in the thread so feel free to have it deleted after you read it.
This could be an interesting idea, perhaps this could serve as a test for the contest concept you mention in the ‘Problem child’ thread.

But I would make to make oa few suggestions for the future if this works out:

1. One account geared for collecting monies for funding contest, any monies received will be used in the next contest etc.

2. Rather then be and independent project, meaning that you alone are the only person having responsibility for it, as this would be a good deal of work to run, maintain and plan there could be some sort of volunteer console for judging entries, drafting the details of a contest, and oversight of the monies in the account (to protect the person collecting monies from any accusation of theft as this could happen, this may also help make people feel more confidant that the monies they send are really going to this purpose…Its the internet suspicion when it comes to money is an important issue that would need to be dealt with).

3. Some means by which people can access the entries of the contest, perhaps NI would be will to do this in the UL if the idea took off, for now a thread in the forum would have to do.

Just some ideas, you may already have consider all of these things, and are just getting the ball rolling, but I thought that I would express that this is an interesting idea.

 

Honestly the amount of money that will be moved around for this project is likely to be so small that the idea of a great internet heist isn't terribly likely.

As far as my trustworthiness, I care more about my reputation than I do a couple hundred or a couple thousand dollars.

I'd probably sell one of you guys to be a sex slave if we were talking multiple millions of dollars, that would likely suppress my moral issues with it. But anything less than 2-3 million USD and I'll play it straight.

If this goes on likely what I would do is add anyone who wants to be added to the 'You've received a payment!' paypal email so that every time a donation comes in everyone on the list would get notified.

But, I'm not going to hold my breath just yet.. This is after all the first experiment!

The way I'm dealing with this currently is posting the name and the amount each person who has donated to the wiki page for the project (http://www.remix.net/wiki/ammobox), simply put it's all out in the open, there is no subterfuge.

If this experiment is successful, I'd definitely be interested seeing more organization to it, judging, all that good stuff. But it depends on participation and how this idea is received.

I think the rule is if you have an entry for this contest, upload it to the UL. Everyone can take a look at it. As far as judging, I can probably do an ok on my own this round but if anyone wants to volunteer to assist in technical analysis of the entries (if there are any!) your help would be greatly appreciated. I understand things from a functionality standpoint, I will likely have less of a nuanced perspective on the aesthetics of DSP solutions.

I'm glad you like the idea, I hope other people do as well, I hope this generates some motivation around the idea of making a community marketplace by fostering creativity and getting solutions that benefit everyone.

 

I am not questioning your trustworthiness. I am not sure if I sounded that way or not.

I am only saying that if this was to take off having someone else involved on that end may short fuse some potentail problems. I have seen it happen in another places where this sort of thing developed and it exploded in the face of a really good guy, and to note those people all know one another... the internet could be a bit of trick etc. I am just looking to safeguard you or anyone else if this became more then the current contest.

 

Nope, no worries. We'll see how this goes. If it works and people want to continue it, we'll figure out a way of standardizing how it operates.

Truth be told, I don't really even want to run them, I'm lazy. But since nobody else is I'm left with a non-lazy solution. If I could snap my fingers and have all the Reaktor stuff I wanted without any effort, I'd probably just do that instead. heh.

 

Nobody going to give this a shot? Bueller? Bueller?

 

CONTEST IS OVER! I WIN!

ammobox continues to be a one man show.

Took me 16.5 hours.

 

even though the prize is won, could i shoot some cash over to you (since you are the winner) for one on one understanding on how all this works? i'm really interested in this project, but i haven't had any time to get into it. but i really want to get a working understanding of FFT and how you got this guy up and running.

 

Basically I set up an 8 windowed 512 bin FFT , I'm then doing peak detection amongst the bins every time a window emits a frame of processed data, by doing that I get a pretty accurate detection of the speed.

I'm still stuck a little bit on the phase detection portion although I know what needs to happen, I'm just not sure how to do it yet -- essentially after spending about $400 on books and nary a response from any interested parties I cancelled the contest once I was able to get the FFT working enough that it was simply a matter of time before I could get it working as a drop-in replacement.

Once I get the phase stuff sorted out I'll be uploading the new ammobox with new UI and some bug fixes.