El Thu, 13 Mar 2008 10:02:51 -0700
"Steven Yi"  escribió:


> I saw this link from Alex Ross's site today:
> 
> http://www.sonicstate.com/news/shownews.cfm?newsid=6281

wow, thanks for the link, steven, the demonstration is amazing. i
recommend it to anyone interested in DSP.

i find this absolutely groundbreaking, like the holy grail of DSP. 

I fail to be convinced by this. It's very hard to gauge from a demo 
showing the pieces that work best with the system. It's also hard to 
judge the quality of the sound in that clip. The morph from the 
time-pressure domain to the notation view is seriously cool however!

Rory.




luis jure wrote:
> El Thu, 13 Mar 2008 10:02:51 -0700
> "Steven Yi"  escribió:
> 
> 
>> I saw this link from Alex Ross's site today:
>>
>> http://www.sonicstate.com/news/shownews.cfm?newsid=6281
> 
> wow, thanks for the link, steven, the demonstration is amazing. i
> recommend it to anyone interested in DSP.
> 
> i find this absolutely groundbreaking, like the holy grail of DSP. 
> 
> 
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Rory Walsh a écrit :
> I fail to be convinced by this. 

same here. mostly viral marketing imho.

Stef

Stéphane Rollandin wrote:
> Rory Walsh a écrit :
>> I fail to be convinced by this. 
>
> same here. mostly viral marketing imho. 
My first reaction is to agree, but on reflection I think it's just 
another cool piece of music technology.

I'm reasonably certain it will not improve the quality of popular music. 
Nothing else has worked so far. :)

Best,

dp

indeed. It's impressive but no more than a magic show! I would be more  
interested in analyzing and tweaking the internal spectral attributes  
of a sound of any complexity than breaking a guitar chord into its  
component notes. Can't remember the last time I thought about "notes"!!

P


On 14 Mar 2008, at 11:49, Dave Phillips wrote:

> Stéphane Rollandin wrote:
>> Rory Walsh a écrit :
>>> I fail to be convinced by this.
>>
>> same here. mostly viral marketing imho.
> My first reaction is to agree, but on reflection I think it's just  
> another cool piece of music technology.
>
> I'm reasonably certain it will not improve the quality of popular  
> music. Nothing else has worked so far. :)
>
> Best,
>
> dp
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"

For me the impressive thing is the user interface. It's interesting how
people are mainly focusing on the underlying polyphonic segmentation
technology. But what impresses me is the slick way in which the
functionality is made available to the user via what appears to be a very
friendly and well designed UI.

BTW, I agree about the note-centricity of it, I would like to see a UI
like this where you can visually manipulate/transform each note's spectrum
and time stretch by dragging-out the partials, then perform the polyphonic
segmentation on the result ad infinitum...

Jamie

> indeed. It's impressive but no more than a magic show! I would be more
> interested in analyzing and tweaking the internal spectral attributes
> of a sound of any complexity than breaking a guitar chord into its
> component notes. Can't remember the last time I thought about "notes"!!
>
> P
>
>
> On 14 Mar 2008, at 11:49, Dave Phillips wrote:
>
>> Stéphane Rollandin wrote:
>>> Rory Walsh a écrit :
>>>> I fail to be convinced by this.
>>>
>>> same here. mostly viral marketing imho.
>> My first reaction is to agree, but on reflection I think it's just
>> another cool piece of music technology.
>>
>> I'm reasonably certain it will not improve the quality of popular
>> music. Nothing else has worked so far. :)
>>
>> Best,
>>
>> dp
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>> "unsubscribe csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"

I agree,

After toying with Melodyne recently on the back of my Britney mania, the
thing i came away with that impressed me most was the interface - 

the whole "controllers in lanes" thing p*sses me off. Melodyne plops the
"controller" data right there up against the note/ space it's associated
with. Pure genius? try bleeding obvious! ; ) 

present the data to the user as he is concieving it, not as your machine is
expecting to read it (as a dedicated, separate controller lane / global
variable...). This reinforced this idea for me high on my own GUI priorities
list...

That & it's capacity to separate drift from vibrato, & of course all the
format tweeky stuff. (with every parameter displayed right there against the
note, & migratable with the note in ne simple action...)

Im only on 56k, so haven't been able to watch the new fangled glory... must
remember to take a look next time i'm in the vicinity of some broadband...

So irrespective of what it can do, i think there is a potentially a genesis
of a whole resynthesis composition interface in the melodyne interface fer
sure. Not to mention whatever DSP chicanery is going on within.

I think the only thing standing in opposition to my statement at the moment
from a resynthesis point of view is the capacity to realise a smooth
interpolation of separate analysis files across the pitch / freq domain to
render audio / multiple .pvx / analysis files as a kind of "matrix" for
seemless glissando type effects.... (something that's living in the back of
my mind for future examination...)

Formant / ratio based data analysis & reduction probably lends itself easier
to this kind of application - i've been looking at using SDIF type data
defined in ratios to the fundamental rather that absolute freq values as a
way forward with this...




Jamie Bullock wrote:
> 
> 
> For me the impressive thing is the user interface. It's interesting how
> people are mainly focusing on the underlying polyphonic segmentation
> technology. But what impresses me is the slick way in which the
> functionality is made available to the user via what appears to be a very
> friendly and well designed UI.
> 
> BTW, I agree about the note-centricity of it, I would like to see a UI
> like this where you can visually manipulate/transform each note's spectrum
> and time stretch by dragging-out the partials, then perform the polyphonic
> segmentation on the result ad infinitum...
> 
> Jamie
> 
>> indeed. It's impressive but no more than a magic show! I would be more
>> interested in analyzing and tweaking the internal spectral attributes
>> of a sound of any complexity than breaking a guitar chord into its
>> component notes. Can't remember the last time I thought about "notes"!!
>>
>> P
>>
>>
>> On 14 Mar 2008, at 11:49, Dave Phillips wrote:
>>
>>> Stéphane Rollandin wrote:
>>>> Rory Walsh a écrit :
>>>>> I fail to be convinced by this.
>>>>
>>>> same here. mostly viral marketing imho.
>>> My first reaction is to agree, but on reflection I think it's just
>>> another cool piece of music technology.
>>>
>>> I'm reasonably certain it will not improve the quality of popular
>>> music. Nothing else has worked so far. :)
>>>
>>> Best,
>>>
>>> dp
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
> 
> 
> 
> 
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
> 
> 


-----
*******************
www.phasetransitions.net
hermetic music * python * csound * possibly mindless ranting
various werk in perpetual delusions of progress....

-- 
View this message in context: http://www.nabble.com/-ot--Melodyne-DNA-tp16033390p16062864.html
Sent from the Csound - General mailing list archive at Nabble.com.

I am guessing but the spectrum manipulation from a partial should be 
trivial in and of itself. Generating the partial requires some analysis 
and predictable continuity / stability in sounds. An algorithm to 
recognize and respond to 'note phenomena' that creates an event for a 
synth / filter bank would be the most CSoundish thing DNA could 
accomplish. I am willing to bet that statistical inferences are used to 
decide what a note sounds like. (pun intended) Another factor that most 
instruments allow you to throw dice on is the assumption that the attack 
will be full of noise components, which will die quickly, and harmonic 
components which will not. Finally there is resonance of the instrument 
body which involves varying levels of formant-ish sound. These factors 
taken together would be enough, but you can add the possibility of 
stereo delay analysis to dig notes out of bigger recordings.

All of these algorithmic notions can be implemented as a collective set 
of analysis tools. A statistical engine can look at the output of each 
contributing algorithm to help make a fuzzy logic determination of what 
constitutes a note. The analysis obviously needs Wavelet analysis as a 
foundation. Time and frequency are essential co-variables for this kind 
of analysis.

The bottom line is that if this is indeed the architecture in use, then 
the output of the analysis tools could be made available through the 
plugin (without giving away deep dark secrets) to trigger external 
events. That would give you the manipulation/ transformation of 
spectrum. I suspect they are holding back on that feature, hoping to 
give some reasonable control inside the plugin. If no 'in the plugin' 
processing is envisioned, then I would hope they eventually release a 
version that splits out the formants from the frequency harmonics so 
that each component can be manipulated independently.

Time stretch, swing, quantization, tempo change, key change and style 
change are all in the plugin to be released in the fall according to the 
web site. I don't have the plugin, but I have heard from reliable 
sources that the existing plugin works pretty well. It just does not do 
DNA. However DNA does not require any new analysis tools, essentially it 
is just a midi rompler on top of the existing engine. I think it may be 
a bit more sophisticated than that with the way it treats 'formants'. 
Any way the results are much better than auto-tune.

What I would really like to hear is what would happen if Angelo Farina 
collaborated using his new analysis/synthesis engine. Put those two 
technologies together and the universe as we know it would change 
completely.

Michael




> BTW, I agree about the note-centricity of it, I would like to see a UI
> like this where you can visually manipulate/transform each note's spectrum
> and time stretch by dragging-out the partials, then perform the polyphonic
> segmentation on the result ad infinitum...
>
> Jamie
>
>   
>> indeed. It's impressive but no more than a magic show! I would be more
>> interested in analyzing and tweaking the internal spectral attributes
>> of a sound of any complexity than breaking a guitar chord into its
>> component notes. Can't remember the last time I thought about "notes"!!
>>
>> P
>>
>>
>> On 14 Mar 2008, at 11:49, Dave Phillips wrote:
>>
>>     
>>> Stéphane Rollandin wrote:
>>>       
>>>> Rory Walsh a écrit :
>>>>         
>>>>> I fail to be convinced by this.
>>>>>           
>>>> same here. mostly viral marketing imho.
>>>>         
>>> My first reaction is to agree, but on reflection I think it's just
>>> another cool piece of music technology.
>>>
>>> I'm reasonably certain it will not improve the quality of popular
>>> music. Nothing else has worked so far. :)
>>>
>>> Best,
>>>
>>> dp
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>> "unsubscribe csound"
>>>       
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>     
>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>
>
>
>
> E-mail message checked by Spyware Doctor (5.5.0.178)
> Database version: 5.09420
> http://www.pctools.com/spyware-doctor/
>
>   





E-mail message checked by Spyware Doctor (5.5.0.178)
Database version: 5.09420
http://www.pctools.com/spyware-doctor/

> I would like to see a UI
> like this where you can visually manipulate/transform each note's  
> spectrum
> and time stretch by dragging-out the partials, then perform the  
> polyphonic
> segmentation on the result ad infinitum...

This would be my dream come true!! It's what Diphone from IRCAM tries  
to do without success (http://forumnet.ircam.fr/703.html?&L=1). It's  
the whole issue of symbolic representation of sonic attributes: are  
sounds just too complex? Maybe spear and audiosculpt get close, but  
still there is so much work to be done particularly on the UI front as  
you mention.

Best
Peiman




On 14 Mar 2008, at 13:48, Jamie Bullock wrote:

>
> For me the impressive thing is the user interface. It's interesting  
> how
> people are mainly focusing on the underlying polyphonic segmentation
> technology. But what impresses me is the slick way in which the
> functionality is made available to the user via what appears to be a  
> very
> friendly and well designed UI.
>
> BTW, I agree about the note-centricity of it, I would like to see a UI
> like this where you can visually manipulate/transform each note's  
> spectrum
> and time stretch by dragging-out the partials, then perform the  
> polyphonic
> segmentation on the result ad infinitum...
>
> Jamie
>
>> indeed. It's impressive but no more than a magic show! I would be  
>> more
>> interested in analyzing and tweaking the internal spectral attributes
>> of a sound of any complexity than breaking a guitar chord into its
>> component notes. Can't remember the last time I thought about  
>> "notes"!!
>>
>> P
>>
>>
>> On 14 Mar 2008, at 11:49, Dave Phillips wrote:
>>
>>> Stéphane Rollandin wrote:
>>>> Rory Walsh a écrit :
>>>>> I fail to be convinced by this.
>>>>
>>>> same here. mostly viral marketing imho.
>>> My first reaction is to agree, but on reflection I think it's just
>>> another cool piece of music technology.
>>>
>>> I'm reasonably certain it will not improve the quality of popular
>>> music. Nothing else has worked so far. :)
>>>
>>> Best,
>>>
>>> dp
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe
>> csound"
>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"