The Filesharing Conspiracy

reply to post by StargateSG7

P.S. Actually theres is a very GOOD reason to be able to do 20,000 tracks
of live mixing or "demuxing" which is used for Audio-based location systems
that can find out (Using an Extreme Example!) such as determining where
EVERY sound within a 3D environment originated from...In my case to
find out the originating location of thousands of gunshots within a gunfight
in a war zone so that an automated shot direction locationing system can
AUTOMATICALLY direct return fire to the originating locations within
MILLISECONDS of those shots being fired and heard by the system

You mean something like the Boomerang system that allready exists.

From Wiki

The Boomerang unit attaches on a mast to the rear of a vehicle and uses an array of seven small microphone sensors. The sensors detect and measure both the muzzle blast and the supersonic shock wave from a supersonic bullet traveling through the air (and so is ineffective against sub-sonic ammunition). Each microphone detects the sound at slightly different times. Boomerang then uses sophisticated algorithms to compute the direction a bullet is coming from, distance above the ground and range to the shooter in less than one second. Users receive simultaneous visual and auditory information on the point of fire from an LED 12-hour clock image display panel and speaker mounted inside the vehicle. For example, if someone is firing from the rear, the system announces "Shot, 6 o'clock", an LED illuminates at the 6 o'clock position, and the computer tells the user the shooter's range, elevation, and azimuth.

It uses seven sensors and far less than 20,000 channels. It also doesn't need a 25 computer network. It isn't automated yet, but it is less than seven years old. So that may be the next step.

The down side to automating return fire at the rate of 6,000 rounds per minute is that you don't have time to confirm if it was (close range) enemy fire or a friendly attempting to defend his position.

Back on topic. You stated that your system could perform the task. Now you say you don't know. Sounds like a severe case of waffling. Plus, if it takes up to eight days it is commercially pointless. I could hire somebody to resing the song, mix it, press it, and have it in the store in eight days.

Lets step up the test though. You said you make it sound like it was done anywhere as well. So let's step it up a notch. Make it sound like it was recorded completely in Westminster Abbey. No artifacts of the original can be left in. You said that you could make it artifact free using logic trees.

reply to post by StargateSG7
[more
Sorry no go. You did not claim you might be able to do it. You claimed it can be done right now. Now you're backing out already saying maybe.

reply to post by daskakik

I got ya brain trust. You have made your point very clear. I have no more questions for you. WOW!

Originally posted by rick1
reply to post by daskakik

 

I got ya brain trust. You have made your point very clear. I have no more questions for you. WOW!

C'mon is it really so bad for me to try before I buy. I pay for whatever I keep and I pay to watch a movie in a theatre if it's worth it. What's so WOW about that?

EDIT to add - See thats the thing. You don't want to strike a deal with me. It's your way (big musics way) or the highway. Maybe I just wanna buy half the album. Sure I can get the single but not the other 2 or three songs I want without paying for the whole thing. What can I do?

[edit on 30-6-2010 by daskakik]

Is it a conspiracy or is it just that is it so easy to do and millions of people are doing it that is would be hard to police it? Then again if they fined a few thousand people say $10 or $20,000 for doing it then it would frighten a lot of people off doing it?

Originally posted by J.Clear

Originally posted by StargateSG7

No Problem! --- This will be fun!

The result could be a total disaster OR it could be totally STUNNING!!!!

[edit on 2010/6/30 by StargateSG7]

We certainly will find out! Just "Hit Me Baby One More Time"

Let's also go with the middle-aged Bass, to make this a significant test.

You say "no problem", but then you say the result will be unpredictable. Eh, those two scenarios don't quite match up, or rather don't match up at all. I'll be expecting the ABSOLUTELY FLAWLESS result you promised initially with great confidence. Spotlight's on ya, buddy!

AND

more
Sorry no go. You did not claim you might be able to do it. You claimed it can be done right now. Now you're backing out already saying maybe.

-----

"I'll be expecting the ABSOLUTELY FLAWLESS result you promised
initially with great confidence. Spotlight's on ya, buddy! "

YES the spotlight will certainly be on my claims....
ME AND MY BIG MOUTH! ....BUT..... I think this would
a be a GREAT TEST so see what software can really do.

Whatever the result, I'll post it so you can either throw
EGG in my face or SHOWER ME with toe-sucking accolades!

Next time I'll leave the grandiose claims at home,
but since I DID SAY it could be done....I (WE!) will see
just how good we are --- or NOT ! he he he he!

I'll post the result no matter what!

---

And the BIGEGST IRONY of this...is that I will have to fork over
the $15 CANADIAN to ACTUALLY BUY the Britney Spears album
so I can get the original 16-bit recording quality file to be able
to run this test fairly!

OH THE HORROR, THE HORROR, THE HORROR
of making Britney an X percentage of $15.00 richer!

[edit on 2010/6/30 by StargateSG7]

Originally posted by StargateSG7

Originally posted by J.Clear

Originally posted by StargateSG7

No Problem! --- This will be fun!

The result could be a total disaster OR it could be totally STUNNING!!!!

[edit on 2010/6/30 by StargateSG7]

We certainly will find out! Just "Hit Me Baby One More Time"

Let's also go with the middle-aged Bass, to make this a significant test.

You say "no problem", but then you say the result will be unpredictable. Eh, those two scenarios don't quite match up, or rather don't match up at all. I'll be expecting the ABSOLUTELY FLAWLESS result you promised initially with great confidence. Spotlight's on ya, buddy!

-----

"I'll be expecting the ABSOLUTELY FLAWLESS result you promised
initially with great confidence. Spotlight's on ya, buddy! "

YES the spotlight will certainly be on my claims....
ME AND MY BIG MOUTH! ....BUT..... I think this would
a be a GREAT TEST so see what software can really do.

Whatever the result, I'll post it so you can either throw
EGG in my face or SHOWER ME with toe-sucking accolades!

Next time I'll leave the grandiose claims at home,
but since I DID SAY it could be done....I (WE!) will see
just how good we are --- or NOT ! he he he he!

I'll post the result no matter what!

---

Hahaha!! No worries, I won't do that, but you have set the bar very high for yourself in the first few posts you made in this topic..

Haha, you won't torrent her, no?

Originally posted by rick1
reply to post by StargateSG7
[more
Sorry no go. You did not claim you might be able to do it. You claimed it can be done right now. Now you're backing out already saying maybe.

I agree with Rick. Looking forward to hearing this tech that will put me out to pasture!

Originally posted by MikeNice81

Back on topic. You stated that your system could perform the task. Now you say you don't know. Sounds like a severe case of waffling. Plus, if it takes up to eight days it is commercially pointless. I could hire somebody to resing the song, mix it, press it, and have it in the store in eight days.

Lets step up the test though. You said you make it sound like it was done anywhere as well. So let's step it up a notch. Make it sound like it was recorded completely in Westminster Abbey. No artifacts of the original can be left in. You said that you could make it artifact free using logic trees.

I called troll on this quite a while back. I'm willing to give the guy a chance to prove this out, but I'm also betting that it won't be up to scratch at all or that he'll never come back. I strongly believe, actually, that this is the last we'll hear of it. I state now that I think he won't be able to do it.

So - the test we've set here by consensus:

Taking Britney Spears' voice, transforming it perfectly into the voice of a middle-aged male bass with perfect tone, intonation, resonance, timbre etc. The voice must also be transferred to the location of Westminster Abbey with no artifacts. My monitors await the results!!

EDIT: I should say also, I work a lot with choirs, so - impress me!!

[edit on 30-6-2010 by J.Clear]

I'll waffle a bit by saying that I can't PERSONALLY predict
the final result (NO HUMAN CAN!) because of the SHEER NUMBER
OF VARIABLES that are calculated...but much like
the Lotto you can't win until you play!

Even if the result sounds absolutely horrific....it might
be FUNNY and thus go viral - Think Luciano Pavarotti
trying sing "Baby, Baby! Hit Me One More Time" and
since he's now dead, this could really freak people out!

I'll keep you updated....U2U if you want specific information
about what I'm doing...Regarding Westminster Abbey,
I don't suppose anyone has an acoustic model of that
place that they could send me.....it's kinda of hard for
me to physically acoustic-model a place I've never been to.
24 UNCOMPRESSED channels of Pink Noise at 192 kHz, 24 bits
with MS-EXCEL format 64-bit or 80-bit floating-point-based
timing info in the microsecond (NOT millisecond) range and
a pinpoint 3D map of the room dimensions and where the
mics were placed in Autocad 14 format would be IDEAL!

Again...this might be EGG ON MY FACE or the OSCAR!

We'll see..........

----

P.S.

QUOTE:

Plus, if it takes up to eight days it is commercially pointless. I could hire somebody to resing the song, mix it, press it, and have it in the store in eight days.

END_QUOTE:

On a commercial note, it wouldn't be pointless if you have the
required amount of computer horsepower...I'm doing this on a
SMALL NETWORK which means more time....however if I run the
system on a 1024 processer DSP/GPU/CPU system, which is it's
INTENDED platform, then results would be near instantaneous.

And an array processer is in the $150,000 range which is NOT that
much compared to a 2 million dollar SSL Scenaria audio workstation
system which SOME high end studios have...would you rather spend
$150 to $200k or 2 million dollars...you can also compare against
the cheaper Master Studio System at ($500k)

www.solid-state-logic.com...

or our cheaper $200k array processor system
hmmm saves $300k which sounds like a good deal to me.

---

In response to earlier comments regarding the Boomerang
system, I was well aware of that system, but ours has
INTEGRATED Audio locationing and Vision Recognition with
autonomous Target Hunting, Tracking & Object Identification
mated to an autonomous robotic armature system that
has 6-axis of movement for VERY FAST TRACKING
(millisecond response times) and RESPONSE.


[edit on 2010/6/30 by StargateSG7]

[edit on 2010/6/30 by StargateSG7]

reply to post by StargateSG7


Don't sweat it. Most of the worlds population are not sound engineers. It only has to be flawless for a select few. For the rest it only has to be good enough. If you don't believe me just check out the pop charts.

Originally posted by StargateSG7
I'll waffle a bit by saying that I can't PERSONALLY predict
the final result (NO HUMAN CAN!) because of the SHEER NUMBER
OF VARIABLES that are calculated...but much like
the Lotto you can't win until you play!

Even if the result sounds absolutely horrific....it might
be FUNNY and thus go viral - Think Luciano Pavarotti
trying sing "Baby, Baby! Hit Me One More Time" and
since he's now dead, this could really freak people out!

WAFFLE.



That's not what he meant. He meant the original claim is now waffle. So no more dancing, please. Just get working!

I'll keep you updated....U2U if you want specific information
about what I'm doing...Regarding Westminster Abbey,
I don't suppose anyone has an acoustic model of that
place that they could send me.....it's kinda of hard for
me to physically acoustic-model a place I've never been to.
24 UNCOMPRESSED channels of Pink Noise at 192 kHz, 24 bits
with MS-EXCEL format 64-bit or 80-bit floating-point-based
timing info in the microsecond (NOT millisecond) range and
a pinpoint 3D map of the room dimensions and where the
mics were placed in Autocad 14 format would be IDEAL!

Again...this might be EGG ON MY FACE or the OSCAR!

We'll see..........

I might have an impulse of it somewhere, I'll have a look. But, no more dancing. You're adding more variables on to it now that you didn't say was necessary before. What you've asked for there noone here is likely to be able to supply. You're spoofing, to use an Irish term. I'm not going to accept what you've asked for above as a reason for not being able to do it. You figure it out, because you said it's possible, right here, right now. No more talk, let's just get it done please, if you can!

Originally posted by daskakik
reply to post by StargateSG7

 

Don't sweat it. Most of the worlds population are not sound engineers. It only has to be flawless for a select few. For the rest it only has to be good enough. If you don't believe me just check out the pop charts.

To be honest, the way it's likely to sound, anyone will hear it. You'll also hear the silence if he disappears

Originally posted by J.Clear

Originally posted by daskakik
reply to post by StargateSG7

 

Don't sweat it. Most of the worlds population are not sound engineers. It only has to be flawless for a select few. For the rest it only has to be good enough. If you don't believe me just check out the pop charts.

To be honest, the way it's likely to sound, anyone will hear it. You'll also hear the silence if he disappears

Didn't mean that it couldn't be heard but that it would matter. I'd rather hear something off than silence just to see what he came up with and I think you guys may scare him off.

He don't sound scared to me I'd be much more concerned about it being a spoof and him no-showing, to be honest. Look at the last bit he just threw in. Do you have any 24 track, 24 bit recordings with pinpoint accuracy of location of Westminster Abbey lying around? I think the making excuses may already have begun.

[edit on 30-6-2010 by J.Clear]

Originally posted by daskakik

Originally posted by J.Clear

Originally posted by daskakik
reply to post by StargateSG7

 

Don't sweat it. Most of the worlds population are not sound engineers. It only has to be flawless for a select few. For the rest it only has to be good enough. If you don't believe me just check out the pop charts.

To be honest, the way it's likely to sound, anyone will hear it. You'll also hear the silence if he disappears

Didn't mean that it couldn't be heard but that it would matter. I'd rather hear something off than silence just to see what he came up with and I think you guys may scare him off.

---

No I'm not scared...I KNOW what the system CAN DO,
it's just a matter of me getting down to rolling up my
t-shirt sleeves all the way up to the shoulders and simply
getting ON WITH IT !!!! Like I said even if its sounds
absolutely screechingly horrific, I'll at least get the
satisfaction of knowing that the system works
on real-world music (However Badly or Well!).

And since it's based on a reconfigurable logic tree,
my first try won't be the last one since the system can
"Learn" over time as more environments and vocalization
models are input to it...Think of it as learning from it's past
mistakes and since it can't ever become embarrassed, I can
make it try, try and try again until it gets it RIGHT!
And over time, the system gets faster & faster due to
the "Training Effect" as some logic trees are discarded
or added as the system resynthesises a particular
series of acoustic models.

P.S. I was actually being QUITE SERIOUS about the
Westminster Abbey thing...I KNOW that there are some
acoustic models of it out there and MAYBE someone has it here,
otherwise I'll add a generic cathederal acoustic model to the
final output and the Westminster Abbey version will have to
wait until I get over to London in the next few months!




[edit on 2010/6/30 by StargateSG7]

Originally posted by StargateSG7


No I'm not scared...I KNOW what the system CAN DO,
it's just a matter of me getting down to rolling up my
t-shirt sleeves all the way up to the shoulders and simply
getting ON WITH IT !!!! Like I said even if its sounds
absolutely screechingly horrific, I'll at least get the
satisfaction of knowing that the system works
on real-world music (However Badly or Well!).

And since it's based on a reconfigurable logic tree,
my first try won't be the last one since the system can
"Learn" over time as more environments and vocalization
models are input to it...Think of it as learning from it's past
mistakes and since it can't ever become embarrassed, I can
make it try, try and try again until it gets it RIGHT!
And over time, the system gets faster & faster due to
the "Training Effect" as some logic trees are discarded
or added as the system resynthesises a particular
series of acoustic models.

P.S. I was actually being QUITE SERIOUS about the
Westminster Abbey thing...I KNOW that there are some
acoustic models of it out there and MAYBE someone has it here,
otherwise I'll add a generic cathederal acoustic model to the
final output!

Go for it. And I'd love to read more about your logic trees, if you could direct me towards any interesting reading on the topic.

EDIT: You see, the big huge stumbling block I have right here is that how does it know it's not correct, or that it sounds stupid. Who sets the quality variables? That is, potentially, where your whole entire argument right from your very first post falls down. If YOU have to make it try again, you are influencing the choice and it is no longer automated. YOU will have made the quality decision on whether it was good or not. The human factor will NOT have been eliminated AT ALL. Effectively, getting down to brass tacks, nothing will have changed.

From my experience, this is exactly where any automated system I've seen to date falls down, and was Mike's point quite a fair way back. Human input is still needed for the quality judgement. Input is still required because the computer doesn't know what's good or not. You must set the parameters and ultimately at the moment for any system I've come across, it's just not possible to set enough of them or accurately enough to replace a human. It's totally impractical, unfeasible and I'd hazard to say totally impossible in the present day.

[edit on 30-6-2010 by J.Clear]

[edit on 30-6-2010 by J.Clear]

Regarding the Human input thing, we must take a cue from
Fractal Graphics where we supply an initial SINGULAR series
of quality decision inputs, but like a human, the logic trees
eventually filter out the good from the bad based upon
previous experience. It's the INITIAL WEIGHTING of the
variables in the logic tree which is where we start but
afterwards the process is fully automated and the more
dataset inputs that are given to the system, the more
it can discern bad from good quality sound.

However as a caveat, if the initial GROWTH conditions/inputs
are BAD then those datasets are the initial comparators
and will be REINFORCED as more data-sets are input.
This gives rise to the "Juvenile Delinquent" fuzzy logic tree
which gets worse and worse at deciding what's good or bad
because it's INITIAL BIRTH (i.e. initial data-sets) was "bad"
to begin with.

----

Here's a background on some of the articles I am familiar with
and have read: Some of the stuff I have is "Classified Materiel"
and is thus not discussable in this forum, whereas other material
is freely available:

Hope it helps:


Fuzzy Logic:
en.wikipedia.org...


Neural Nets:
en.wikipedia.org...


Grid Processing:
en.wikipedia.org...


Bayesian Logic:
en.wikipedia.org...


Boolean Logic:
en.wikipedia.org...


Speech Synthesis:
en.wikipedia.org...


New Yamaha VOCALOID Singing Synthesis Software Generates Superb Vocals On A PC:
www.vocalist.org.uk...


A sound and complete fuzzy logic system using Zadeh's implication operator:
www.springerlink.com...


A tree-based statistical language model for natural language recognition
by LR BAHL - 1989 - Cited by 223 - Related articles
ieeexplore.ieee.org...


Audio Cues needed for listeners to feel as if their environment is natural:
Free-Field Spatialized Aural Cues for Synthetic Environments:
oai.dtic.mil...


Flow-Structure-Acoustic Interaction in a Prestressed Human Larynx Model :
www.springerlink.com...


Underwater acoustic modeling and simulation By Paul C. Etter:
books.google.ca... 3f1zKcP_lKLpvaSTvnI&hl=en&ei=EuUrTKOyGpjonQfbvfXTCQ&sa=X&oi=book_result&ct=result&resnum=5&ved=0CCsQ6AEwBA#v=onepage&q&f=false

Modelling sound production from an aerodynamical model of the human newborn larynx:
www.sciencedirect.com... canchor=&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=c5a35e0e34d0a22b54cf0b0ee2c3ded2


Aerodynamic and Acoustic Effects of Abrupt Frequency Changes in Excised Larynges:
www.britannica.com...

Open SOurce Acoustic Models for Synthetic Speech:
www.speech.cs.cmu.edu...

Emotional Speech Synthesis:
emotion-research.net...


ARTICULATORY SPEECH SYNTHESIZER by Chang-Shiann Wu
www.aclweb.org...


Medically Oriented Logic trees are adaptable to ANY type of dataset:
www.lulu.com...


and this is a small sampling of the THOUSANDS of articles
that are stuffed in my file cabinets.

reply to post by StargateSG7

And an array processer is in the $150,000 range which is NOT that
much compared to a 2 million dollar SSL Scenaria audio workstation
system which SOME high end studios have...would you rather spend
$150 to $200k or 2 million dollars...you can also compare against
the cheaper Master Studio System at ($500k)

You could just go with an API 1608 and expander. With full expansion package it is $89,000. Plus API is a completely discrete circuit topography and completely modular with any of the "lunch box" preamps or eqs on the market. For less than $100,000 it is fully integrated with Pro Tools. They use this particular in one of Black Bird's studios. In case people aren't familiar Black Bird is considerred the best country studio in the US.

Telefunken USA use the API 1608 for testing their new microphone designs. Plus it does 5.1 monitoring.

Really though you could get a Control Eight or a C24 and build your own custom API, SSL, Avalon, Apogee, system for less than $25,000. Throw in Waves Mercury and the set up is complete for less than $40,00. That doesn't include Mics. A mac Pro with a 2.66ghz Quad Core Xenon processor adds another $1,700. We could call it a complete studio for with Mics for less than $100,000.

(I intentionally left out the building or space because it hasn't been mentioned before now.)

Still much cheaper than Buying a $200,000 dollar array plus all of the necessary cables, microphones and everything else.

The amount of computing power you are talking about drags us completly away from the "mid price" home computer that started the computer tech discussion.

Like I said cost and availability still make it pointless at this time. In a decade it might be a valid option. I doubt it though since you still compare getting good results to playing the lotto.

My point still stands that buying a small home computer will not give you the same results as a good professional studio.

Thanks for the links Stargate. I will be reading them in the days to come.

About Vocaloid, I've used it. In less time than it took to program four part harmonies for a lead my assistant had allready went in and recorded two parts with the lead. I then got the singer to sing his own fourth part.

Plus the quality of output was glitchy on uncommon words and letter combinations.

It couldn't give me the right drawl for a New Orleans sounding vocal. It did okay overall but the emotion was completely missing. Even people that didn't know when I used it commented that the backups sounded stiff or stale.

Admitedly we are talking 2006 so things may have changed.