Audio mysticism

[antto]

and if nobody provides counter-proof - it must be true and real
just like.. god or santa claus

Vertion: i'm no longer sure what your pseudo-code is supposed to do, i'm no longer sure what language it's written in, and i have better things to do
wake me up when you have a working demo

[mystran]

and if nobody provides counter-proof - it must be true and real

Pigeon hole principle applies. Q.E.D.

[Gamma-UT]

Vertion, you missed a spot:

Code: Select all

FUNCTION DESCRAMBLE_BLOCK() 
  INSERT_MAGIC_FUNCTION_HERE()
END FUNCTION

[stratum]

INSERT_MAGIC_FUNCTION_HERE()

Not very uncommon :
https://en.wikipedia.org/wiki/Oracle_machine ,works exactly the same way as https://www.pbs.org/empires/thegreeks/b ... /7_p1.html

[quikquak]

Decoding is actually quite fast considering what it can do... it has real applications, more to the archival, long distance transmission and syncronization.. but there are other more subtle and powerful uses. After all, the only words ever worth saying are... I love you.

Code: Select all

FUNCTION MAIN()
	BLOCKNUM = -1;
	BLOCK()
END FUNCTION

FUNCTION BLOCK()
	BLOCKNUM = BLOCKNUM + 1 % 256
	DESCRAMBLE_BLOCK() // INVERSE OF ENCODER SCRAMBLE
	IF BLOCKNUM = 0
		BIT = 0
		INCREMENTOR--
		IF INCREMENTOR = 0
			SAVE_BITSTREAM_AND_EXIT()
		END IF
		STREAM[INCREMENTOR] = BIT
	END IF
	BLOCKTYPE = GETBLOCKTYPE()
SECONDBIT:
	IF BLOCKTYPE = BB_INCREMENT THEN BLOCK()
	IF BLOCKTYPE = BB_UNKNOWN THEN
		IF IS_LAST_UNKNOWN_COMBO() GOTO RETURNCODE
		NEXT_UNKOWN_COMBO()
		BLOCK()
	END IF
	IF BLOCKTYPE = BB_RETURN THEN GOTO RETURNCODE
RETURNCODE:
	IF BLOCKNUM = 0 AND BIT = 0
		BIT = 1
		GOTO SECONDBIT
	END IF
	BLOCKNUM = BLOCKNUM + 255 % 256
END FUNCTION

FUNCTION GETBLOCKTYPE()
	RETTYPE = BB_INCREMENT
	FOR B = 1 TO 8
		HIGHNEIGHBOR = SEEDEDRAND(GET_HIGHPOSITION())
		LOWNEIGHBOR = SEEDEDRAND(GET_LOWPOSITION())
		IF HIGHNEIGHBOR = 1
			IF LOWNEIGHBOR = 0
				RETTYPE = BB_RETURN
			END IF
		ELSE
			IF LOWNEIGHBOR = 1
				IF RETTYPE <> BB_RETURN THEN RETTYPE = BB_UNKNOWN
			END IF
		END IF
	NEXT B
	RETURN RETTYPE
END FUNCTION

I'll check back later... please analyze fully.. I had to hash this out between chores... I used a recursive function to elaborate and example, but the final function is linear for the sake of scale.. you can easily play with the internal stats in a simplified way with that dice rolling website link I provided in an earlier post.. use 8 tetrahedron and you will see why.. chaos has difficulty branching out here, the linear encoding is too strong.. the only way it can go on beyond a few levels.. is if it were DESIGNED. You are welcome to provide counter-proof.. may the real challenges begin.. prove this impossible while keeping the working principle intact.. is this mind bending folly beyond comprehension? or is there something wonderful here to celebrate in great rejoice together?

you decide.

It's all gone a bit 'Silicon Valley' around here! But I have too many 'chores' to do, to revolutionise the world of audio...

[JCJR]

I haven't understanding of what vertion is getting at, but reminds of algorithmic composition on the one hand and DNA on the other.

Expression of dna results in wide variety with more complexity than the parent dna chain. So far as I recall, last time I looked, dna was being considered a "construction algorithm" rather than "blueprint". A complete blueprint for an organism would require more data than merely the construction algorithm. Even if we data-compress the blueprint it requires more data than the construction algorithm.

Starting from the same DNA pattern we tend to get "nearly the same" organism. In real world biology there are numerous confounding factors, mother's hormones, nutrient levels yadda yadda. But a "digital dna" designed to build songs might be made more deterministic than biology.

Just sayin, maybe rather than lossless-data-compressing as tiny possible the audio of "Louie Louie" by the Kingsmen-- Maybe a deterministic algorithm could be devised which, when executed, naturally results in the "Louis Louis" audio?

Algorithmic composition is similar-- You can take a small number of rules and generate arbitrarily long songs. So the trick would be to devise the smallest set of rules which always results in "Louie Louis" and nothing else!

Maybe impossible. Maybe theoretically possible but not feasible. Haven't the foggiest idea. I wouldn't know where to start.

One could specify "Louis Louis" in general fashion with a remarkably small number of bytes in a band in a box file-- The band in a box file much smaller than the resulting midi file, and the resulting midi much smaller than the final audio.

But the trick would be a similar but smarter process which results in the EXACT AUDIO of that old cheezy song. Or any other song.

[Chaotikmind]

I haven't understanding of what vertion is getting at, but reminds of algorithmic composition on the one hand and DNA on the other.

Expression of dna results in wide variety with more complexity than the parent dna chain. So far as I recall, last time I looked, dna was being considered a "construction algorithm" rather than "blueprint". A complete blueprint for an organism would require more data than merely the construction algorithm. Even if we data-compress the blueprint it requires more data than the construction algorithm.

Starting from the same DNA pattern we tend to get "nearly the same" organism. In real world biology there are numerous confounding factors, mother's hormones, nutrient levels yadda yadda. But a "digital dna" designed to build songs might be made more deterministic than biology.

Just sayin, maybe rather than lossless-data-compressing as tiny possible the audio of "Louie Louie" by the Kingsmen-- Maybe a deterministic algorithm could be devised which, when executed, naturally results in the "Louis Louis" audio?

Algorithmic composition is similar-- You can take a small number of rules and generate arbitrarily long songs. So the trick would be to devise the smallest set of rules which always results in "Louie Louis" and nothing else!

Maybe impossible. Maybe theoretically possible but not feasible. Haven't the foggiest idea. I wouldn't know where to start.

One could specify "Louis Louis" in general fashion with a remarkably small number of bytes in a band in a box file-- The band in a box file much smaller than the resulting midi file, and the resulting midi much smaller than the final audio.

But the trick would be a similar but smarter process which results in the EXACT AUDIO of that old cheezy song. Or any other song.

You're describing the Kolmogorov complexity of a data sequence here. It's theoretically sound, but it's impractical to use as you can imagine.
Finding the program that will generate your data will take eons.

Edit:
But it's maybe tractable with enough a priori info on a specific kind of data.

[BertKoor]

The ultimate data compression algorithm is a jukebox.
Press button B12 (8 bytes) and it plays "Thriller - Michael Jackson" (40 megabytes)
Alas only known songs can be compressed.

[fese]

Who needs compression anyway? Just put it in the cloud, it’s the solution for everything! If that doesn’t help, throw some ML at it. IT is so easy!

[kamalmanzukie]

sorry to bump this ancient thread, but this was the only real lead i got from advanced search when trying to find the situation that inspired this tweet i made the other week, which i think sums up the thread as well as a lot about the developers forum in general. it is this:

KVR smart is like abstract deity, where insight pours unbound, eternal. then there's a 'magic plugin' kind of guy this attracts. and this poor guy, this aspect of pure mind made known to self, has to listen to dip-sh*t, about how flipping bits on a known random seed will allow infinite data compression, until there's a polite exit

if anyone has a lead on when or where a discussion like this took place, i'd be very grateful