Login / Register  0 items | $0.00 New

Audio mysticism

stratum
KVRian
 
1296 posts since 29 May, 2012

Postby stratum; Sat Aug 05, 2017 5:14 am Re: Audio mysticism

This part of the article is particularly interesting

Actually, as noted by Dennis Gabor (best known for
his invention of holography, but who also worked in
audio) back in 1946, the ears actually analyse the
frequency content of sounds in time faster than
suggested by the uncertainty principle by a factor of
about 7. The seeming logical contradiction with the
fundamental theoretical limit of time/frequency
resolution is avoided by the ear’s use of a priori or
previously assumed knowledge of the nature of typical
sounds but at the expense of getting the analysis
‘wrong’ when sounds not of the assumed form occur.
No one has yet succeeded in devising a method of
simultaneous time/frequency analysis that beats the
uncertainty principle limits on resolution using a priori
information similar to the ear. Existing methods of
analysis do not resolve enough detail in the two
domains simultaneously to predict reliably how a filter
will sound.


It's quite clear that while wavelets wouldn't do that either, perhaps it might be possible to use better time resolution obtained from the higher frequencies to infer localization info about the lower frequencies since most of the time frequencies that are integer multiples of each other come from a single source (like a vibrating string) and therefore related?
~stratum~
User avatar
nonnaci
KVRist
 
215 posts since 7 Feb, 2017

Postby nonnaci; Sat Aug 05, 2017 5:58 am Re: Audio mysticism

stratum wrote:It's quite clear that while wavelets wouldn't do that either, perhaps it might be possible to use better time resolution obtained from the higher frequencies to infer localization info about the lower frequencies since most of the time frequencies that are integer multiples of each other come from a single source (like a vibrating string) and therefore related?


Localization is mostly a binaural phenomenon relying on different cues depending on part of the source spectra. e.g difference in onset time-delay or group-delay is responsible for low frequency localization on azimuth plane; wavelets don't improve temporal resolution in low-frequency so there's little benefits here. High-frequency content conveys elevation information based on notch patterns in the spectra derived from subjective anthropometry. That said, time-delay is still relevant here and wavelets could be used to resolve cock-tail type localization problems of say mixed narrow and wide-band sound-sources.
stratum
KVRian
 
1296 posts since 29 May, 2012

Postby stratum; Sat Aug 05, 2017 6:14 am Re: Audio mysticism

By localization I had actually mean time resolution (i.e. localization in the t-axis), sorry about the misleading term.

Yes wavelets don't improve it, but since the article says that the human ear does better than what is theoretically possible, I had thought, well, this may be one way of doing it, i.e. imagine harmonics from a single vibrating string, 1...N, obviously the N'th harmonic will have the best temporal resolution (we know its 'location' in the t-axis better), and we may infer something about the first harmonic using that information. Isn't that what the article also implies by saying "The seeming logical contradiction with the fundamental theoretical limit of time/frequency resolution is avoided by the ear’s use of a priori or previously assumed knowledge of the nature of typical sounds but at the expense of getting the analysis ‘wrong’ when sounds not of the assumed form occur." I couldn't see what may be additionally assumed so that we may get better temporal resolution than what is theoretically possible. That sounds like a sentence with an open meaning, they don't specify any details about what the human ear assumes, i.e. we don't have any additional clue about the temporal localization (i.e where it starts or ends) of the N'th harmonic, so it's unlikely that the human ear achieves better than theoretical temporal resolution for it.

Using this rationale, perhaps some EQ's mess with such clues and therefore sound 'colored'?
~stratum~
User avatar
nonnaci
KVRist
 
215 posts since 7 Feb, 2017

Postby nonnaci; Sat Aug 05, 2017 6:32 am Re: Audio mysticism

stratum wrote:1...N, obviously the N'th harmonic will have the best temporal resolution (we know its 'location' in the t-axis better), and we may infer something about the first harmonic using that information. Isn't that what the article also implies by saying "The seeming logical contradiction with the fundamental theoretical limit of time/frequency resolution is avoided by the ear’s use of a priori or previously assumed knowledge of the nature of typical sounds but at the expense of getting the analysis ‘wrong’ when sounds not of the assumed form occur." I couldn't see what may be additionally assumed so that we may get better temporal resolution than what is theoretically possible. That sounds like a sentence with an open meaning, they don't specify any details about what the human ear assumes.


Ah I see your point although this a-priori schemta could be said for most transient sounds being broadband and of having originated from a common source-location. This is why white-noise bursts are commonly used in localization tests and not pure-tones ;)

On another bit about a-priori structures, monoaural localization of moving broadband noise-type sources are statistically significant. To me, it seems like neurons of single-ears are able to map coloured white-noise to directions.
User avatar
Vertion
KVRist
 
307 posts since 29 Oct, 2016

Postby Vertion; Sat Aug 05, 2017 11:36 am Re: Audio mysticism

antto wrote:oh now it all makes perfect sense


This is real. I will provide the source code in pseudo code form to anyone who asks me on a peaceful day. I am freely presenting verifiable truth and challenging you directly to see the truth. I prove that infinite data compression is real, will your faith increase? :)
ENVIRON
KVRer
 
11 posts since 4 Feb, 2017

Postby ENVIRON; Sat Aug 05, 2017 1:39 pm Re: Audio mysticism

Consider your challenge accepted. I'd like to see it.
User avatar
Vertion
KVRist
 
307 posts since 29 Oct, 2016

Postby Vertion; Sat Aug 05, 2017 2:16 pm Re: Audio mysticism

ENVIRON wrote:Consider your challenge accepted. I'd like to see it.


Very well, I spoke about this with my wife. I think the easiest way for me to transmit the algorithm, is to make a video of me teaching the algorithm to her. The plan is, she will interview me tonight with questions to represent everyone who wishes to learn this. Since she does not know the algorithm, teaching her will be teaching all you wish to know. In my excitement, I would like to upload tonight, however, tomorrow around noon PST I will upload to youtube and then post the link here as soon youtube provides. :)
User avatar
antto
KVRAF
 
2474 posts since 4 Sep, 2006, from 127.0.0.1

Postby antto; Sat Aug 05, 2017 3:23 pm Re: Audio mysticism

infinite detail, now infinite data compression
my skepticism towards this went up exponentially



soon: infinite skepticism
It doesn't matter how it sounds..
..as long as it has BASS and it's LOUD!

irc.freenode.net >>> #kvr
User avatar
Zaphod (giancarlo)
KVRAF
 
2323 posts since 23 Jun, 2006

Postby Zaphod (giancarlo); Sat Aug 05, 2017 3:32 pm Re: Audio mysticism

nonnaci wrote:
Zaphod (giancarlo) wrote:(Yes it is old... But each age is raising new questions, we don't have an answer to everything)


It was a good read nonetheless. From experience, psychoacoustic effects should not be ignored as elements of short-filter designs such as eq. are magnified when applied to long-filters. Case in point, pre-ring and pole resonances produce obvious colourations in reverb design.



Eheh Yes
User avatar
Vertion
KVRist
 
307 posts since 29 Oct, 2016

Postby Vertion; Sun Aug 06, 2017 10:44 am Re: Audio mysticism

https://www.youtube.com/watch?v=YSEAg9z ... 6oWTpNU5x6

Here is the interview she did asking me to explain the data compression algorithm. This video is in 3 parts on this playlist.

I'm sure there will be more requests for clarification. It was pretty late at night. If you have any questions for me to answer, I will add more detail. Be sure you ask until you are absolutely clear on the details. I will clarify all details and answer all relevant question for a clear understanding of how the algorithm works.
stratum
KVRian
 
1296 posts since 29 May, 2012

Postby stratum; Sun Aug 06, 2017 11:23 am Re: Audio mysticism

Infinite data compression is indeed possible, just have an infinite dictionary of symbols that you can refer to :wink:

I couldn't understand what you are trying to accomplish by explaining that though, and didn't watch the video because it was too long. There are many things that are possible in principle but aren't possible at all in practice.
~stratum~
User avatar
antto
KVRAF
 
2474 posts since 4 Sep, 2006, from 127.0.0.1

Postby antto; Sun Aug 06, 2017 12:02 pm Re: Audio mysticism

i can't currently watch it because of youtube html5 video problems

it's funny how so many hundreds of megabytes have to be wasted for communicating information which can fit into a few kilobytes worth of text
or is there more to it?
It doesn't matter how it sounds..
..as long as it has BASS and it's LOUD!

irc.freenode.net >>> #kvr
quikquak
KVRist
 
298 posts since 6 Aug, 2005, from England

Postby quikquak; Sun Aug 06, 2017 12:45 pm Re: Audio mysticism

Infinite? :hihi: I suppose if the results were only finite, you could pass them back in and compress it again?
:pray:
User avatar
Vertion
KVRist
 
307 posts since 29 Oct, 2016

Postby Vertion; Sun Aug 06, 2017 5:34 pm Re: Audio mysticism

quikquak wrote:Infinite? :hihi: I suppose if the results were only finite, you could pass them back in and compress it again?
:pray:


The stream is constant and infinite because it loops the transforms on the same data.. the 256 bytes remain the same size (along with the 4 byte incrementer). It is not a form of compression that relies on one-to-one mapping of a finite stream. It is constantly transforming the data and utilizing the input stream to transform the 256 bytes in a way that leaves directions on how to return to the original signal via reciprocation... it goes back the way it came. By utilizing noise and statistics, this is possible... this isn't sort of, or kind of true. This is real. If you need more detail to clarify your understanding of the exact working of the algorithm, I will be happy to provide. If someone else can explain the algorithm with more clarity, I welcome that explanation. I welcome all peaceful inquiries on the algorithm.

It is also not a slow compression, considering it is infinite, it is useful both as an unlimited archiver, deep distance transmissions, and full network synchronicity. Decoding is slower, but since it is linear, it actually does a fairly fast job... on an infinite data stream.

I challenge everyone once, again.. I will give every detail I have. As dominos fall one after the other in a perfect progression, this algorithm works. The decoder must deal with a few levels of randomness, but it is within practical limits and is always self correcting (linear recursion).

If anyone wants to get it right for the sake of everyone else's understanding, they are welcome to get every detail from me.. I simply have a lot of detail on this, and know it from my own perspective, so I would appreciate the right questions from the truth seeker. This is a linear algorithm, that also must deal with noise on it's decision tree during the decode.
User avatar
Vertion
KVRist
 
307 posts since 29 Oct, 2016

Postby Vertion; Sun Aug 06, 2017 7:51 pm Re: Audio mysticism

To go into a higher level of detail, here is some quick pseudo code for the encoder portion of the algorithm. :)

Code: Select all
// Pseudo Code For Endless Encoder

// Variables
file_size_in_bits int   // How big is the file in bits
file_bit[file_size_in_bits] bool   // File data in here, Boolean array, bit by bit
circles[256] byte // output data 1
inc unsigned 32 int   // wrapping int, resets after max_value, output data 2

// Main
loop i from 0 to file_size_in_bits
   inc++
   encode_bit = file_bit[i]
   loop block from 0 to 31
      reciprocal_shuffle(block, block+31 % 32, inc) // use current block to shuffle the previous block
      noise_encode(block, encode_bit, inc)
   end loop block
end loop i

function bool seeded_random(bigint seed) // random number generator with proper distribution

function void reciprocal_shuffle(current_block, previous_block, inc_seed)
   // uses current block data to scramble the previous block using the increment step,
   // uses xor, add/sub, ror, etc to create a recipricol scramble (reversible)

function void noise_encode(block, encode_bit, inc_seed)
   loop t from block * 8 to block * 8 + 7
      if seeded_random(inc_seed << 32 + circles[t] << 24 + t << 1 + encode_bit) == 1
         circles[t]++
      else  // it returned a zero
         circles[t]--
   end loop t

PreviousNext

Moderator: Moderators (Main)

Return to DSP and Plug-in Development