Slashdot Mirror
The Weak Signal Challenge - Decode and Win $100
superid writes "Several years ago while reading comp.dsp I found a link to The Weak Signal Challenge. On that page is a .wav file of a morse code signal bounced off the moon. The page author Mike Cook is offering $100 to the next person to successfully decode the morse code.
Since I was the one who originally solved this, I promised Mike that I wouldn't divulge the answer or provide any clues. I can say thought that I didn't use anything special other than traditional signal processing techniques, octave, matlab, and patience. I think that overall I spent about 24 hours total sitting at my '486.
I think it would be great to generate some interest in this. Maybe someone could come up with a novel solution and win $100!"
It starts out, "ALL YOUR BA--"
I'll finish up translating the rest tomorrow.
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
This is nothing more than a recording of a McDonald's drive-through speaker.
Man $100 for 24 hrs of work? Is that even minimum wage?
- Moomin
http://www.rathergood.com/moon_song/
"I think that overall I spent about 24 hours total sitting at my '486." With my Intel Pentium II processor, with the power to make the internet come alive, i should be able to complete the problem in one tenth the time of your 486! and since this is multimedia, my MMX processor should really speed things up!
"This should send the guys at SETI on a wild goose chase."
I would be interested in the motivation behind the challange, its clearly not a 'how good are your ear' challange as the first winner used a computer. Does he hope to boost some obscure area of signal processing or just hand out $100 just for the fun of it?
On a gamer forum, mentioning a '486 would get you laughed out in about ten seconds.
;)
On a non-technical forum, mentioning a '486 would get a bunch of (the digital equivalents of) blank stares.
On SlashDot, only a couple of people think it's anything out of the ordinary...
It would seem that the Internet has a wide variety of computing cultures
Honey, I shrunk the Cygwin
So your time is worth approximately $4.17 an hour I take it ;-)
;-)
Very cool link though - sounds like a fun challenge.
Damn though, somehow I think if I end up doing this my time will come out being worth about $0.05 an hour... all for the fun of it
-Colin
50 SWM, seeking 30-40 SWF for laughs and good times. Please contact me at XXX-XXX-XXXX.
I can clearly hear the morse code beeps. With some head phones (if I knew morse code) - I could decode this using only the world's most powerful real-time signal processing and pattern recognition device - the human brain.
-josh
"Eat at Joes" or something equilly monumental
Rus
Cheap UK and US VPS
Maybe someone could come up with a novel solution...
Turn the volume up - a lot.
Everything in the world is controlled by a small, evil group to which, unfortunately, no one you know belongs.
It's TCP/IP over Morse code.
Talk about lag...
I promised Mike that I wouldn't divulge the answer or provide any clues. I can say thought that I didn't use anything special other than traditional signal processing techniques, octave, matlab, and patience.
Why, you've given it all away! I can tell that since you mentioned traditional signal processing techniques that the answer is clearly not in the neural network in front of you but being that you know that I know you're a Sicilian you obviously wouldn't put it in a simple band pass filter in front of me! Aha ha...Aha ha ha *clunk*
"Stop! Your radio waves are killing our moon babies! First it was Country Music, and now this."
Table-ized A.I.
Hey Suid - you made Slashdot!
Have you compiled your kernel today??
Umm.. thats not quite the point. The page says this:
Do you think you are good at copying weak signals? Well here's a test for you. I am posting a zipped 1 minute wav file of a VERY weak EME station calling me. I am offering $100 and a free copy of the FFTDSP42 to the first person who can tell me the call sign of the calling station. The signal is strong enough to just copy my call (AF9Y) near the middle of the 1 min period. The mystery station is sending a simple repeat of his call and my call. The characters "DE" may or may not be between the two calls.
So you CAN hear one of the call-signs... its harder than just listening carefully. Otherwise it would make no sense.
You spent 24 hours trying to win $100?
Dude, you could make more than that working at Wendy's.
- A.P.
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
No, I could hear the signal from the very beginning of the wav file. Very faint, but it was there.
-josh
This is nothing more than a recording of a McDonald's drive-through speaker.
It is the repetion in the message that allows humans to know what they are really saying. Thus, we hear, "Vu Vu Von Vie Vih Vah", and know from prior experience that they are saying, "Do you want fries with that?" Otherwise, I often would not know what the hell they are saying.
Table-ized A.I.
... to drink your Ovaltine.
Portable versions of Firefox, GIMP, LibreOffice, etc
Try writing a book, if you don't get a best seller you would probably earn even less
This is one small signal for man, one giant signal for mankind.
Special Relativity: The person in the other queue thinks yours is moving faster.
Reminds me of story somebody told me once, but I have yet to verify it.
During preparations for the Apollo moon missions, some members of NASA were sent out to record greetings messages from various communities to be included on a recording that was to be left on the moon by the astronauts.
After visiting a Native American tribe, one Native American man refused to tell NASA what he had just said into the recorder in his native tongue. So, they eventually found a translator, and the message said something like, "Watch out for these guys. They will take your land too. Explorers, my ass".
Table-ized A.I.
I'd assume the beeps are a single tone. Run a fourier xform on the signal and look for any dominant tones (peaks) in the spectrum. Once the frequency is known, run the signal through a narrow band filter at that frequency to eliminate the noise. After that the tones should be recognizable enough to decode by ear.
Vote for Pedro
Here: Morse Code info.
Now all you need is a pair of headphones. I'm sure you can find a decent pair for less than $100 and make some quick money.
Try it. I'm betting it's harder than it sounds, but hey: people's hearing does vary. You might just be good at this.
'Sensible' is a curse word.
Resistance is futile
This is a test. This is a test of the emergency sig system. This has been only a test.
For my senior design project, we had to make some dedicated hardware to exactly this. It was solved using an FFT inside a Vertex-II Pro FPGA using a sliding window.
Now, as far as this contest goes, this can be done quite easilly in a computer. In fact, we had made a bitwise simulation of the FPGA using matlab before doing any Verilog to make sure that our design worked. My favorite was the little setup that we did in LabView. Took about 30 minutes to make, and would solve this quite easilly. It was very nice to see the result in color intensity with frequency along the x axis, and time across the y. Seeing the data in there was quite easy even though the SNR was so low that you couldn't even hear most of the tones.
In short, if you want to do this, read the data into matlab (or octive), do an FFT on a small window of the data, slide the window over, repeat. Take all the FFT's and plot in an intenisty graph (not sure what the matlab term is, but I am sure it has such a thing).
You see? It's like I've always said. You can get more with a kind word and a 2x4 than you can with just a kind word.
Can someone please open-source the solution (probably after winning the reward). I'd also like to see the original author's solution -- matlab code, and whatever.
Thx!
Maybe someone could come up with a novel solution and win $100!
Sounds like the best solution could just be a Gaussian filter could be passed over all the values to lessen the noise, then have the remaining values leveled on a linear scale - then finally have the translated solution passed through a common acronym/spell checker.
-Gwala
#!/bin/csh cat $0
Yeah, but Wal-Mart doesn't let you stock shelves in your underwear while eating Doritos and chugging Mountain Dew.
No, I don't want to talk about it.
KFG
I took a professional signal analysis software package and analyzed the file in detail. There is indeed a signal, quite discernible, at about 570 Hz. However the package that I have is visual only, it is not suitable for audio playback, and I am too lazy to add audio output to it... not worth of $100 for sure. Also, the signal drifts considerably.
From af9y.com:
"The mystery station is one of the calls in this list of all known 2 mtr EME stations: all2eme.txt (18 K Bytes)"
peter@df5jt $ wget http://www.webcom.com/~af9y/all2eme.txt
peter@df5jt $ cat all2eme.txt|grep -i df5jt
DF5JT
peter@df5jt $
Where's my 100 bucks?
google is your friend. First came across this technology in the 1970's in a paper from the Naval Post Grad School. The problem them was how to communicate with submerged subs using only very low frequency keyed CW. google will lead you to more modern applications, e.g., "A technique for removing noise and emphasizing coherent events from multiple channels of seismic data."
"Please send help! Drowning! Will not last much longer!"
Outdoor digital photography, mostly in New Engl
Why bother translating it? Getting all technical is the best way to ruin perfectly good music.
I bet the answer is 47 - it's always bloody 47..
--A Million-in-one chance will work out 99% of the time...
This sig is in Spanish when you're not looking....
Well, I guess we're fucked then, eh? Better watch out for giant domino-wannabes plummeting from the sky...
i downloaded the "zip" file and ran mplayer on it... it's BITTERSWEET by FUEL... WTF
What is slashdot?
It's STENDEC (or ETA LATE).
"And a voice was screaming: 'Holy Jesus! What are these goddamn animals?'" - HST
there must be some certain-frequency signal in the "noise" that occurs in the noise for two durations: the duration of a morse dit and dah. Wouldn't an FFT and analysis of the relative amplitudes and durations suggest what the message is (to a person knowing morse code)?
So, why do people use LabView for this?
detecting and decoding are two very different things.
**I call bullcrap. You listening to 3 seconds of it instaltly make you an expert ** Well, he's right. I also can clearly hear the dots and beeps in that file, with nothing more than turning up the treble amp on a pair of laptop speakers. Three other "witnesses" in the room with me at the time can also hear it just fine. And yes, we listened to the whole thing. Doesnt sound like a very long message. Since I've got better things to do with my time than jot it all down and translate it, I'll leave it to someone who actually has to work to earn the $100, not someone who can hear it. PS. You know those "unhearable, ultrasonic" pest repeller things? I know 5 people, inluding myself, who can hear those damn things ticking :p
*There's Klingons on the starboard bow, scrape em off Jim!*
Put the file into XMMS or Winamp, depending if your in linux or windows. Turn on the AMP/Equalizer, and drag everything except 12-14k down to nothing. Play, turn up speakers, unplug ears and listen. You'll hearn a pattern starting out with what sounds to ..-...-.-
It goes by way to fast for me to pick up much more than that. But I suppose I could get less lazy and slow it down.
*There's Klingons on the starboard bow, scrape em off Jim!*
Took the wave file, loaded it onto a windows pc. Loaded Nero wave editor. Select All. Analize Noise, Remove Noise, adjust equalizer... Clear, perfect morse code file... All in 15 seconds work... Sad, isnt it?
*There's Klingons on the starboard bow, scrape em off Jim!*
I was indeed able to pull out a legible signal using Cool Edit 2000, I don't speak morse. I sent the file to the folks at the center.
I will post any results to my journal.
If I win anything it is going to the EFF. (Protecting grannies from the RIAA. 24/7)
Arnold is looking pretty good right now.
Anyone want to recommend what programs to use to change my levels so i can decipher static from morse?
Anyone want to recommend a program so i can adjust my audio levels to hear past the static so i can decipher the damn morse?
Q. How do you fix a broken chimp?
A. With a monkeywrench.
Got it!!!!
WKRP
"The avalanche has already started. It's too late for the pebbles to vote." - Kosh
You mean that signal spike in SETI@HOME was just an ovalten commerical...I thought it was ET phoning home....
"The problem with socialism is eventually you run out of other people's money" - Thatcher.
"Hey /. I found this Differential Equations take-home final that has been BOUNCED OFF THE MOON and I'll give $100 to the first person that completes it! Should only take about 24hrs, and I think it would be great to generate some interest in this."
my karma will be here long after I'm gone
You're using wrong character encoding so the lameness filter deduces that it will show up as a bunch of garbage to most people. You could try recoding it to ASCII first...
My dad ran crypto for the Navy back in the 50's, was even given a citation from the Pentagon for his ability to pick out weak Morse Code on the fly..and decode the stuff, realtime, in his head! Spent like 6 years up in Adak, Alaska listening to Russian freighter traffic. Three cheers for Cold War paranoia!
:)
I'll pass this along to him.
Bowie J. Poag
use letters instead of . and -, maybe i and t (see the dot and dash along the middle of the row?)
Man,
The solution is called matched filter...
If you need to ask is because youre not going to understand anyway...
how long until
It's easy: subtract white noise and there you go :-)
I'd assume the beeps are a single tone. Run a fourier xform on the signal and look for any dominant tones (peaks) in the spectrum. Once the frequency is known, run the signal through a narrow band filter at that frequency to eliminate the noise.
The frequency of the signal may change during transmission, due to Doppler effect. In this sample it constantly increases. So running a Fourier transform on the signal won't show you any distinct peaks.