UNKNOWN ARTEFACT: Sound Engineers, apply here!

[CMDR Balboa]

Hey guys, finally got a small amount of time to work on the audio last night. I focused primarily on the long whale noise that ends each passage of small binary notes (the whale noise is referred to elsewhere in this thread as a "howl." After noting, as others have, that there are two types of these whale noises/howls which alternate, I was wondering if the passages they are attached to might be more than simple pairs. What if the different whale noises mean that each such set of paired passages should be treated as a single sequence? When these 14-16 note/purr sequences are translated into decimals they produce 4 digit integers.

Much more interestingly though, what if the different whale noises represent different messages or voices, such that each alternating sequence of 7-8 small notes belongs to a different message, or to two different speakers talking to or over one another? I haven't had the time to split the recordings apart this way, but I can't see how this can be solved without understanding the significance of the alternating howl sounds first.

 

[scorpyon]

At times this whole thing feels like trying to solve a jigsaw puzzle without any edge or corner pieces! Lol!

 

[McBierle]

Hi People!

I found the posting of Bloomoo interesting, as i don't believe there is (only) some binary or whatever code in the Sound.

I've been playing with the various recordings of the 'voice' and I've managed to make the last bit somewhat clearer.

https://www.dropbox.com/s/tmsbzjrojl16ngk/he_bo_3_voice1.wav?dl=0

compare this with the other versions posted earlier,
http://vocaroo.com/i/s0hIXtltcwMq
http://vocaroo.com/i/s1DcReyHZVyf

I don't want to say what I think it's saying for now, to avoid planting ideas in people's heads.
But I'd be interested to hear what other people think.

So i played (as a total Sound noob) a bit with one of the original recordings. In the first post the part i worked on is refered to as "Chittering".
I swear i hear something like "Gamma-6-6-4-8-5" and some more before (i think i deleted some parts there) and after that.

Here is my file: http://www.dropbox.com/s/akp27bnh5733if7/Code.wav?dl=0


Edit: another one: http://www.dropbox.com/s/7yxut4czf9esdg7/Doppelt.wav?dl=0
Two times in a row. Now i hear something like Delta-Gamma-6 and then Gamma-6-6-4-8-5. The last part i still don't understand.

 

[Phyl]

I'm pretty sure now that the main binary sequences are all 7 bits (again not counting the headers in this), however for some sequences either the first or last bit get lost as there is an amplitude envelope that starts immediately after the sequence of clicks and ends at the start of next set of clicks. If any honks are at the extremes of this range they are very quiet as the amplitude ramps up or down. If they are outside this range they do not play at all. Something like this:



Here are some examples of honks at the extreme ends of this range with very low amplitude:





The spacing between the honks is regular regardless of how many can be heard in a sequence. The size of the howl and the spacing of the honks ensures that the final audible honk will always overlap the howl.

Don't know if that is useful, but I found it interesting!

 

[W4rSkuLL]

Man, something is coming... I hope you get that artifact sounds decoded. It looks very organic and Thargoid like. Maybe its probes sent out before an invasion.

 

[EAGLE 5]

All this reminds me of this: http://en.memory-alpha.org/wiki/Breen_language

 

[nikos]

I'm pretty sure now that the main binary sequences are all 7 bits (again not counting the headers in this), however for some sequences either the first or last bit get lost as there is an amplitude envelope that starts immediately after the sequence of clicks and ends at the start of next set of clicks. If any honks are at the extremes of this range they are very quiet as the amplitude ramps up or down. If they are outside this range they do not play at all. Something like this:

View attachment 34561

Here are some examples of honks at the extreme ends of this range with very low amplitude:

View attachment 34562

View attachment 34563

The spacing between the honks is regular regardless of how many can be heard in a sequence. The size of the howl and the spacing of the honks ensures that the final audible honk will always overlap the howl.

Don't know if that is useful, but I found it interesting!

I've thought this is the case for a while (and said so before ), but simply by listening hard, no science!

 

[kobayashi]

Just a thought, knowing FD's and elite's past on the BBC Micro, this could be a homage, the data could be based on the old tape system audio coding using byte standards?

http://en.wikipedia.org/wiki/BBC_Micro
http://en.wikipedia.org/wiki/Kansas_City_standard

There are some tools for converting this old format in to binary, but not sure the recordings will be clean enough to process:
ftp://ftp.forth.org/pub/Forth/Compilers/native/dos/DXForth/kcs08.txt

I admit this seems unlikely as it is way too complicated to be worked out, if FD wanted this to be decoded they would do something much less complex and give hints over several weeks.

 

[McBierle]

Now i have found this: http://www.dropbox.com/s/m8yhiy3beyevxam/twotwoonealpha.wav?dl=0
Something-2-2-1-Alpha

And i found this: http://en.wikipedia.org/wiki/Electronic_voice_phenomenon

 

[Petersen]

The spacing between the honks is regular regardless of how many can be heard in a sequence. The size of the howl and the spacing of the honks ensures that the final audible honk will always overlap the howl. Don't know if that is useful, but I found it interesting!

I disagree with this (not that my word is absolute, just offering my observations). The howls always occur every 30 seconds. I have taken multiple parts of multiple recordings, aligned the howls, and the purrs (honks) are sometimes aligned, and sometimes not, by varying degrees. Thus the purrs (honks) must be irregularly spaced. I even tried lining up the first purr (honk) between some sequences, and later honks were still offset (not to mention the howls).

 

[Phyl]

I've been looking at the frequency distribution of the symbols. Can anyone better than me with stats help me out interpreting this? I can dust off the old stats-books, but if anyone can save me having to do that it would be most useful

With 42 possible symbols and a sample set of 191, I've mapped the following frequency distribution, however I'm not sure if this is statistically significant? There is a small bais towards higher frequencies of some symbols, and having 11 and 12 occurrences of some symbols seems unlikely by chance, but I know random numbers can do funny things sometimes!



There does seem to be a small correlation with adjacent symbols as well - this highest frequency symbol (k) is preceded by the same symbol (6) on three of its 12 occurrences. I haven't looked much more deeply than this yet though.

The full frequency table for the results I have so far is below. The symbols I've used are just to represent the binary string in order to make it easier to work with them, they do not have any meaning of themselves.

Symbol	Instances
k	12
z	11
r	9
6	9
f	8
u	8
x	8
y	8
v	7
5	7
t	6
n	5
q	5
8	5
a	4
c	4
g	4
h	4
i	4
j	4
p	4
s	4
w	4
1	4
7	4
9	4
&	4
b	3
e	3
o	3
2	3
3	3
!	3
%	3
l	2
@	2
^	2
m	1
4	1
£	1
$	1
d	0

 

[Phyl]

I disagree with this (not that my word is absolute, just offering my observations). The howls always occur every 30 seconds. I have taken multiple parts of multiple recordings, aligned the howls, and the purrs (honks) are sometimes aligned, and sometimes not, by varying degrees. Thus the purrs (honks) must be irregularly spaced. I even tried lining up the first purr (honk) between some sequences, and later honks were still offset (not to mention the howls).

I meant the spacing within a 7-bit sequence is regular. If you align the first honk (purr) of a sequence with another first honk in a different sequence, then the honks will all line up within those sequences, however as you noted the howls will not. I wanted to know if the spacing within a single sequence was different for the 6 bit sequences, which it was not.

Are you referring to misalingment across subsequent sequences? I have not seen it within a single sequence and would be interested if you have found this? The spacing between the 7-bit sequences does vary and so far I have assumed it is random. I believe this is the cause of the 6-bit sequences as one of the bits falls outside of the 'valid' range.

Interestingly the first howl does not conform with the spacing of the others. If you line up the second howl across multiple tracks, all subsequent howls will be aligned but there is variation in the positioning of the first.

 

[Archvillain]

Phyl - the forums won't let me give your posts any rep until I give more to others, but I love what you're doing on several fronts.
In addition to investigating various avenues, you've added evidence behind some hunches.

Framing the missing honks and the quieter honks as both factors of the same simple envelop that all blocks are conforming to - that hits my "that's gotta be it" buttons and gives me more confidence in assuming that 6-digit blocks are incomplete, rather than merely suspecting they are and feeling like I still have to hedge and try working it both ways, or look for possible significance. I love it when an unknown becomes firmer and less speculative. It still raises the issue of why the honks are not chronologically constant with the howl, there may be significance there, but I'm now going to be running with the assumption that the blocks are 7 honks, 6 means something got clipped. Thanks.

 

[Petersen]

I meant the spacing within a 7-bit sequence is regular. If you align the first honk (purr) of a sequence with another first honk in a different sequence, then the honks will all line up within those sequences, however as you noted the howls will not. I wanted to know if this spacing was different for the 6 bit sequences, which it was not.

Are you referring to misalingment across sequences or within the same sequence? I have not seen it within a single sequence and would be interested if you have found this? The spacing between 7-bit sequences does vary and so far I have assumed it is random. I believe this is the cause of the 6-bit sequences as one of the bits falls outside of the 'valid' range.

Interestingly the first howl does not conform with the spacing of the others. If you line up the second howl across multiple tracks, all subsequent howls will be aligned but there is variation in the positioning of the first.

Look at EA_UA Sections 1 and 2 from Glynie (post #15). Align the first purrs after the third howl. Both are 6-purr sequences, but the timing is different. The first two purrs are spot on. The next two purrs, EA-UA Section 2 lags slightly, then the two howls are almost perfectly aligned, and the 6th tone (overlapped by the howl) also sounds aligned.

I do see the kind of behavior you're talking about in EA_UA Sections 4 and 5, though. If you align the first purrs of the two sequences you get exactly the same purr timing up until the first 6/7 discrepancy. Howls are misaligned, but purrs are spot on 'til one hits a 6 and one hits a 7 purr sequence.

EDIT:

but I'm now going to be running with the assumption that the blocks are 7 honks, 6 means something got clipped. Thanks.

I'm curious what your thoughts are on Saman's kaunan_a_1. After the 4th full howl (because the recording starts in the middle of a howl). And after the 7th and 8th. All those sequences of purrs are extremely unusual.

Actaully, that whole recording is rife with anomalies (and I'm not talking about ship sounds, though there's a healthy helping of those). The howls, are really mistimed, the intervals between howls get longer, and then get shorter, but always seem a little longer than the norm.

 

[jspoto]

I'm pretty sure now that the main binary sequences are all 7 bits (again not counting the headers in this), however for some sequences either the first or last bit get lost as there is an amplitude envelope that starts immediately after the sequence of clicks and ends at the start of next set of clicks. If any honks are at the extremes of this range they are very quiet as the amplitude ramps up or down. If they are outside this range they do not play at all. Something like this:

I'd like to believe this, but some of the samples do not have a resolvable 7th sound. And I'm afraid that observer bias is "Addding" 7th notes in some cases

Either way, this is why having an organized set of data to work with is helpful. We can get the community to re-comb thorough data and improve the reliability and accuracy of the binary transcriptions

Edit: I'll attempt to come up with specific examples for shorter "words" for discussion

 

[McBierle]

http://www.dropbox.com/s/z71eh2qz7osidk6/twotwoonefivefive-alyuto.wav?dl=0

two-two-one-five-five / alyuto

 

[jspoto]

There does seem to be a small correlation with adjacent symbols as well - this highest frequency symbol (k) is preceded by the same symbol (6) on three of its 12 occurrences. I haven't looked much more deeply than this yet though.

The full frequency table for the results I have so far is below. The symbols I've used are just to represent the binary string in order to make it easier to work with them, they do not have any meaning of themselves.

.....

Interesting...nice work there Phyl.

On the topic of how to represent the valid binary 'words' : I initially used an alphanumeric scheme like you do above however words of different lengths present an issue -- how do you differentiate between a 5-bit glyph and a 6 or 7 bit one? Also, if we determine that we're dealing with bit lengths of more than 7 or 8 bits as new information comes to light, we'll run out of single alphanumeric symbols.

For now, I've adopted a glyph scheme of (length)-(position), where position is derived from a complete pre-computed table of all possible glyphs, for a given word size. Thus, '7-02' or '6-12' are used in my Python code, with table entries dumped at the end of the processing:

I encourage everyone to play with that stuff here:https://github.com/jspoto/UnknownArtifact I've included an (incomplete) list of sample binary data, as well as a number of audio source files for comparision, which I'll continue to expand

- - - Updated - - -


Also, I plan to add .csv output support to my UAProcess.py script, which should help people with the kinds of processing that Phyl has been pioneering...

 

[jspoto]

Cat Herding

Gah! So difficult to keep track of dicussions on multiple forum threads -- the irony

I'll attempt to process more data and get the latest audio samples some time over the next day or so

 

[jspoto]

Also, I haven't seen this info posted elsewhere, but for anyone who hasn't seen it, some interested ideas an directions on the bit data:
https://github.com/jspoto/UnknownArtifact/issues/8

Takeaway: jlouis observes that a NAND parity bit could exist, and still adhere to the "no runs > 2" rule

 

[Phyl]

Look at EA_UA Sections 1 and 2 from Glynie (post #15). Align the first purrs after the third howl. Both are 6-purr sequences, but the timing is different. The first two purrs are spot on. The next two purrs, EA-UA Section 2 lags slightly, then the two howls are almost perfectly aligned, and the 6th tone (overlapped by the howl) also sounds aligned.

I do see the kind of behavior you're talking about in EA_UA Sections 4 and 5, though. If you align the first purrs of the two sequences you get exactly the same purr timing up until the first 6/7 discrepancy. Howls are misaligned, but purrs are spot on 'til one hits a 6 and one hits a 7 purr sequence.

Not sure I follow exactly where you mean? I've tried here - comparing the first section to the second section:



This is the first time I've seen these samples though, so thanks for the pointer. So far I've been working almost entirely with DigitalScrems UA_1 - UA_5 samples.

Anyhow in this section, they do seem to align to me (at least to within a few m/s, which could be due to interference masking the details). The bottom sample, which is from the second section, is a 7 bit sequence and in the below screen capture it can be seen that the last bit of the top sequence falls outside the valid zone as it overlaps the clicks, so has been lost. I still consider these to be 7 bit sequences, it's just that we don't know what all the bits are in the top one!



If this is not where you mean, are you able to take a screenshot so make it easier to see?

- - - Updated - - -

Phyl - the forums won't let me give your posts any rep until I give more to others

Lol, no need for reps. Just trying to make sense of things one step at a time...