UNKNOWN ARTEFACT: Sound Engineers, apply here!
- Thread starter sohara
- Start date
Very true, some do not. I think there is a 'valid range' for notes that goes from the end of one set of clicks to the start of the next. Anything that overlaps the clicks gets lost and anything very close to the boundaries is very quiet - however it does not necessarily follow that these notes were not there is first place and they may still form part of any message. By looking at the spectograms it can quite easily be confirmed if a note exists and whether it is a 1 or a 0 when it overlaps the honks or is very low volume.
In addition, it is sometimes possible to infer whether the bit has been lost form the start or the end by looking at the spacing. it is also sometimes possible to infer what the missing bit is by following the rule that no more than 2 adjacent bits can be identical. For example, if a sequence is 010011 and it can be determined that the final bit is missing, then you know that has to be a 0 even if it is not present in the recording.
I currently believe that it is valid to assume that all sequences have 7 bits as every sequence I have come across so far with fewer bits appears to have one bit outside the valid zone that have been clipped.
This does not stand true for the 'header', which I a treating as an exception and haven't really looked at yet. The header is the few sporadic honks before the first howl and the sequence of honk between the first and seconds howl. This section seems to use a different scheme and can have up to 4 different tones.
I don't. At the moment I am working on the assumption that all sequences are 7 bits (even if all 7 bits can't be discerned), with the exception of the header which needs to be considered separately. For this analysis, sequences with fewer than 7 known bits have been excluded.
I have also only seen 5 bit sequences within the header and I've yet to see a six bit sequence that doesn't fit with the 'valid range' rule.
Please do prove me wrong on either of these though, as they're just theories at the moment.
I did try looking a the stats when filling in missing bits in 6-bit sequences with random 1s or 0s, on the basis that 50% would be correct and the remainder would not be very significant statistically. This did actually change the results somewhat, boosting the r signal above z and k (up to 14), however I removed this again as it didn't feel right!
Last edited:
Just to Post this here too
About the chittering (as in post #1), i got those:
(Yes i'm Aware of EVP)
1. He Bo http://www.dropbox.com/s/akp27bnh5733if7/Code.wav?dl=0
"Gamma-6-6-4-8-5"
2. Ross 154 http://www.dropbox.com/s/znpl5otc1x15l14/twofivefive-ross.wav?dl=0
"2-5-5"
3. Alyuto https://www.dropbox.com/s/z71eh2qz7osidk6/twotwoonefivefive-alyuto.wav?dl=0
"2-2-1-5-5"
Sounds by saman https://forums.frontier.co.uk/showthread.php?t=142912&p=2194672&viewfull=1#post2194672
Guys, just for your information - I revised, compiled and transcribed few new samples (including Saman's) and posted it to another thread today. Full transcription here: https://forums.frontier.co.uk/showthread.php?t=142794&page=24&p=2206486&viewfull=1#post2206486
By the way, won't be better to have these threads merged? It seems we are doing same things on two places...
By the way, won't be better to have these threads merged? It seems we are doing same things on two places...
This shows that the first purrs are lined up, and that the third purrs are offset.
This gives relative location for the whole of the two tracks.
And I just found the problem. It's ED-UA-Section_1 and ED-UA-Section_10. Stupid me. Because when you sort alphabetically, 10 comes after 1 and before 2.
EDIT: thumbnailed the pictures instead of giving raw links.
EDIT 2:
I challenge the idea that there even is a header. You can't immediately look at an object after you jettison it, can you? It certainly took me a few moments to 'acuire' canisters I ejected the other day doing a little testing
I think what people call 'headers' are simply partial sequences, because the first part can't be heard, as the ship turns to face the artefact, or perhaps artefacts start broadcasting at a random point in the message.
It would be very helpful - as I have said a number of times - if more than one player could observe the same artefact at the same time and see if they receive the same singal.
I'd also be very interested to know if two artefacts are jettisoned sequentially, do their signals synchronize, or not? What if the artefacts aren't the source of the signal at all, but they're functionally radios picking up a signal?
Last edited:
Ahh, yes. I see what you mean. I've come across similar, but to be honest I haven't generally been looking with this level of precision as I have not needed to for my purposes. Anyhow, I think the reason for this is that the gap BETWEEN the samples is consistent, but the samples themselves are different lengths. This would explain why it comes back in sync at the end of the sample - both tracks have the same number of 1s and 0s so it all comes back together nicely at the end, but in the middle there is a difference as the number of 1s and 0s that have been played is different. You would see this measuring the from start of one sample to the start of the next, but not if you measured from the end of one to the start of the next (at least thats what I remember seeing in my limited evaluation of this effect).
My ideas about the header are based in the number of tones more than anything else. In DigitalScreams samples, there is a clear section at the beginning, prior to the second howl, where where are up to 4 tones. Following the second howl there are only ever 2. This section is also the only place I've seen 5 bit sequences, and having the howl on either side of these makes me think they are not just partial captures. In my mind there is clearly something different about this section, which is where the idea of a header is coming from.
Agreed, This would be very useful... I've been searching - if I find one I'll give you a shout
I have also been wondering if what we are hearing isn't a message at all. It could be something like the inner mechanics of the artefact that we can hear whirring away as it tries to do something we are completely unaware of. Gowever that is nowhere near as interesting, so I'll park that one for now!
I'm so split on the issue. Part of me really really thinks we're all just wasting our time, it's just noise to imply that the artefact is of biologic nature, and that's all we were supposed to get from "listening to them". But I agree, that's not nearly as interesting! And we're all having fun spinning our gears trying to figure out what (if anything) is encoded in the sounds, right? So what if it's nothing. We're still having a good time.
Better yet, it may be a Fibonacci code: http://www.reddit.com/r/EliteDanger...ssible_uknown_artifact_decoding_breakthrough/
(The reddit link contains my writeup, but I should probably consider moving it here as well. If anyone wants, feel free to do so).
hmm, has anyone looked at lem his masters voice, carl sagan contact book, carl sagan theory, background radiation band, seti research?
if its encoded and alien, they will use math to encode, definitely not something that we use on earth, so morse code is bogus...
if its encoded and alien, they will use math to encode, definitely not something that we use on earth, so morse code is bogus...
Very interesting and pretty logical theory. But, there is just one flaw - there are "000" and "111" sequences present in the signal (000 in Murp's recording, 111 in saman's recording). But, it may be also interesting to briefly test other codes you mentioned. They are human-invented, but as there is no known purpose of unknown artifact, it doesn't eliminate a possibility that thargoids are mimicking it...
Last edited:
Hey guys,
as a musician and audio production enthusiast, I've tried isolating some of the supposed speech parts. I'm not good with math or programming ;-)
My main source is the 67 and respective 16 minute (x4 speed) version.
I've come to the conclusion that what we have sighted so far as voices remains very random guess work.
http://www.dropbox.com/s/akp27bnh5733if7/Code.wav?dl=0 (this one definitely cut a tiny bit off the beginning)
This is the section I've fiddled with the most. I was quickly able to find at least 3 differently sounding instances of this, and each has some clearer and some more processed "words" (think ring modulator/metalizer).
They all vary in the effects, so my next step will be to find more instances to mix it into one "clear" version.
If this is speech, it sure sounds more like german, than english ;-)
as a musician and audio production enthusiast, I've tried isolating some of the supposed speech parts. I'm not good with math or programming ;-)
My main source is the 67 and respective 16 minute (x4 speed) version.
I've come to the conclusion that what we have sighted so far as voices remains very random guess work.
http://www.dropbox.com/s/akp27bnh5733if7/Code.wav?dl=0 (this one definitely cut a tiny bit off the beginning)
This is the section I've fiddled with the most. I was quickly able to find at least 3 differently sounding instances of this, and each has some clearer and some more processed "words" (think ring modulator/metalizer).
They all vary in the effects, so my next step will be to find more instances to mix it into one "clear" version.
If this is speech, it sure sounds more like german, than english ;-)
Another thing that I hit a wall with is finding if there is an important difference in the second section.
I suspect it's a bunch of more hisses.
Someone could check if there's something up with the last column.
Keep in mind this is the long, unprocessed version, and some of those details could disappear with speeding up/slowing down.
I suspect it's a bunch of more hisses.
Someone could check if there's something up with the last column.
Keep in mind this is the long, unprocessed version, and some of those details could disappear with speeding up/slowing down.
Very interesting and pretty logical theory. But, there is just one flaw - there are "000" and "111" sequences present in the signal (000 in Murp's recording, 111 in saman's recording). But, it may be also interesting to briefly test other codes you mentioned. They are human-invented, but as there is no known purpose of unknown artifact, it doesn't eliminate a possibility that thargoids are mimicking it...
I'm willing to bet that if you put more ears on it, then this is just a listening fluke. Looking at jspoto's signal source here:
https://github.com/jspoto/UnknownArtifact/blob/master/signals.txt
there are no such 111 or 000 in the file. And even if there is a single place where this is present, the representation can either be error-corrected or taken as is. The Fibonacci coding doesn't outright require the bit pattern.
My understanding (I have not tried this, but from a post I vaguely recall reading) is that people have recorded with one ship jettisoning/retrieving and another ship recording (this is necessaru to get as much recorded as possible during the degradation cycle), and someone mentioned that UA doesn't start its song until it's being listened to (until you turn to face it).
If the jettison ship then listens, I don't know whether it hears a private song or hears the same as the other ship is listening to. But either way, if the song does start when you listen, then that bolsters the headers idea.
How would one determine it doesn't start its song until it's being listened-to? If the only way to hear its transmission is to look at it, how can you tell it's not transmitting if you're not looking at it?
Obviously you can't, however a consistently observed starting segment with different structure is a big hint
We can only work with what we have
Last edited:
Been following this closely and am very impressed with the work you guys are doing. I'm particularly excited about jlouis' recent discovery of a Fibonacci sequence connection. Keep it up!
Thought I'd pitch a couple of ideas that haven't come up yet (in this thread at least):
It may be worth comparing the glyph frequency distribution to typical letter or word frequencies for various languages: http://en.wikipedia.org/wiki/Letter_frequency (see table near the end of the article) and http://en.wikipedia.org/wiki/Zipf's_law. If we can determine that the glyph frequency follows a Zipfian (or other known) distribution then we could use that fact to perform error correction or to help determine what the glyphs represent (letters, words, ?). More data will help. It wouldn't surprise me if they do follow a specific distribution -- someone at Frontier knows their statistics as erf has shown up in a couple places (e.g. rare price vs. distance, shield strength vs. class).
Another angle that may be worth pursuing is to compare the relative phase of howls vs. purrs. Does the first seven purr sequence always start with the same delay after the last header howl? Is that delay significant? Could it be that all recordings with a given delay use one encoding or are part of the same message, sort of like a cross-artefact time-division multiplexing scheme?
- ash
Thought I'd pitch a couple of ideas that haven't come up yet (in this thread at least):
It may be worth comparing the glyph frequency distribution to typical letter or word frequencies for various languages: http://en.wikipedia.org/wiki/Letter_frequency (see table near the end of the article) and http://en.wikipedia.org/wiki/Zipf's_law. If we can determine that the glyph frequency follows a Zipfian (or other known) distribution then we could use that fact to perform error correction or to help determine what the glyphs represent (letters, words, ?). More data will help. It wouldn't surprise me if they do follow a specific distribution -- someone at Frontier knows their statistics as erf has shown up in a couple places (e.g. rare price vs. distance, shield strength vs. class).
Another angle that may be worth pursuing is to compare the relative phase of howls vs. purrs. Does the first seven purr sequence always start with the same delay after the last header howl? Is that delay significant? Could it be that all recordings with a given delay use one encoding or are part of the same message, sort of like a cross-artefact time-division multiplexing scheme?
- ash
I used the UA-Sound from this Video (less Phurrs, more Chittering):
https://www.youtube.com/watch?v=RKT6CVKclZg
https://www.dropbox.com/s/3nxld3v7qrpf0ec/UA_Sound_RedWizzard_Video.wav?dl=0
I´m a Sound Engineer-Noob, but this is the result:
https://www.dropbox.com/s/3bz0cehq3z44ly8/UA_Sound_Chittering.wav?dl=0
...another try:
https://www.dropbox.com/s/iwhagii7j1bok9w/UA_Sound_Chittering_2.wav?dl=0
https://www.youtube.com/watch?v=RKT6CVKclZg
https://www.dropbox.com/s/3nxld3v7qrpf0ec/UA_Sound_RedWizzard_Video.wav?dl=0
I´m a Sound Engineer-Noob, but this is the result:
https://www.dropbox.com/s/3bz0cehq3z44ly8/UA_Sound_Chittering.wav?dl=0
...another try:
https://www.dropbox.com/s/iwhagii7j1bok9w/UA_Sound_Chittering_2.wav?dl=0
Please get enhanced transcript I posted above, there are more samples in (including these I mentioned). Or, listen it by yourself, samples are linked here too...
In any case, I like your theory, but as any other theory - it should lead to some results.
Last edited: