View Full Version : Next Project: Sampling == Noise
September 4th 06, 07:06 PM
My next project, for audio is going to be sampling. And I think I've
found a way to isolate lyrics and tracks of music from one another.
All you have to do is create a fuzzy sample of the music yourself. And
then look at that sample as noise that you want to take out of whatever
music you want to work with. So you can just erase the guitar from a
song, and keep the drums or vocals. Cool idea right?
Anyone else know of ways to do this?
September 4th 06, 08:43 PM
If you look at a kareoke song as noise, then you can just sample out
the lyrics from the real song with a noise filter. And you can just
sample out the guitar too, or the drums. And with my Wav2bmp
convertor, you can sample out pretty much everything even when they are
encrypting pop music with vocoders. Because you can just create a
fuzzy immitation, that represents the sound and get close enough.
So download it and check it out:
<a
href="http://grimoire.genesismuds.com">http://grimoire.genesismuds.com</a>
My best sound effect yet: <a
href="http://grimoire.genesismuds.com/perfect.wav">http://grimoire.genesismuds.com/perfect.wav</a>
wrote:
> My next project, for audio is going to be sampling. And I think I've
> found a way to isolate lyrics and tracks of music from one another.
> All you have to do is create a fuzzy sample of the music yourself. And
> then look at that sample as noise that you want to take out of whatever
> music you want to work with. So you can just erase the guitar from a
> song, and keep the drums or vocals. Cool idea right?
>
> Anyone else know of ways to do this?
Joseph Littleshoes
September 4th 06, 08:44 PM
wrote:
> My next project, for audio is going to be sampling. And I think I've
> found a way to isolate lyrics and tracks of music from one another.
> All you have to do is create a fuzzy sample of the music yourself. And
> then look at that sample as noise that you want to take out of whatever
> music you want to work with. So you can just erase the guitar from a
> song, and keep the drums or vocals. Cool idea right?
>
> Anyone else know of ways to do this?
>
Don't music studios have a machine to do this? an equalizer?
http://www.softlab-nsk.com/ddclipro/audiofx/fx_firequ.html
---
JL
September 4th 06, 08:54 PM
Looks like a good effect. I have an equalizer in audacity that I am
using and its free. But even if I have an equalizer to balance the
music with the lyrics, it isn't going to help crack a real song and
take the lyrics out of the music.
Joseph Littleshoes wrote:
> wrote:
> > My next project, for audio is going to be sampling. And I think I've
> > found a way to isolate lyrics and tracks of music from one another.
> > All you have to do is create a fuzzy sample of the music yourself. And
> > then look at that sample as noise that you want to take out of whatever
> > music you want to work with. So you can just erase the guitar from a
> > song, and keep the drums or vocals. Cool idea right?
> >
> > Anyone else know of ways to do this?
> >
>
>
> Don't music studios have a machine to do this? an equalizer?
>
> http://www.softlab-nsk.com/ddclipro/audiofx/fx_firequ.html
> ---
> JL
September 5th 06, 01:25 AM
wrote:
> If you look at a kareoke song as noise, then you can just sample out
> the lyrics from the real song with a noise filter. And you can just
> sample out the guitar too, or the drums. And with my Wav2bmp
> convertor, you can sample out pretty much everything even when they are
> encrypting pop music with vocoders. Because you can just create a
> fuzzy immitation, that represents the sound and get close enough.
>
> So download it and check it out:
> http://grimoire.genesismuds.com
Totally unimpressive, as these effects can be easily
produced more easily by any number of existing means.
In this particular sample, you simply have introduced a
lot of comb filtering and such.
So why is your way, which seems far more involved,
more intricate, far less intuitive in the audio domain,
Does not seem to be able to ever work in real time,
in any way better?
Roy W. Rising
September 5th 06, 01:26 AM
wrote:
> My next project, for audio is going to be sampling. And I think I've
> found a way to isolate lyrics and tracks of music from one another.
> All you have to do is create a fuzzy sample of the music yourself. And
> then look at that sample as noise that you want to take out of whatever
> music you want to work with. So you can just erase the guitar from a
> song, and keep the drums or vocals. Cool idea right?
>
> Anyone else know of ways to do this?
Methinks you write before you think. Delve deeply into the concept of what
you seek to do and discover the fuzziness of your conjecture. OR ... make
it work, go public and let us all buy stock in your company!
--
~ Roy
"If you notice the sound, it's wrong!"
Erwin Hessle
September 5th 06, 02:14 AM
wrote:
> wrote:
> > If you look at a kareoke song as noise, then you can just sample out
> > the lyrics from the real song with a noise filter. And you can just
> > sample out the guitar too, or the drums. And with my Wav2bmp
> > convertor, you can sample out pretty much everything even when they are
> > encrypting pop music with vocoders. Because you can just create a
> > fuzzy immitation, that represents the sound and get close enough.
> >
> > So download it and check it out:
> > http://grimoire.genesismuds.com
>
> Totally unimpressive, as these effects can be easily
> produced more easily by any number of existing means.
> In this particular sample, you simply have introduced a
> lot of comb filtering and such.
>
> So why is your way, which seems far more involved,
> more intricate, far less intuitive in the audio domain,
> Does not seem to be able to ever work in real time,
> in any way better?
The mistake you are making is looking for the answer to your question
from the perspective of someone who's not a total whacked out nutjob.
Erwin Hessle, 8=3
September 5th 06, 02:43 AM
Well, I at least have a first try at taking some samples out of music.
Just listen to this loop I created with Wav2bmp:
http://grimoire.genesismuds.com/stolen.wav
wrote:
> wrote:
> > If you look at a kareoke song as noise, then you can just sample out
> > the lyrics from the real song with a noise filter. And you can just
> > sample out the guitar too, or the drums. And with my Wav2bmp
> > convertor, you can sample out pretty much everything even when they are
> > encrypting pop music with vocoders. Because you can just create a
> > fuzzy immitation, that represents the sound and get close enough.
> >
> > So download it and check it out:
> > http://grimoire.genesismuds.com
>
> Totally unimpressive, as these effects can be easily
> produced more easily by any number of existing means.
> In this particular sample, you simply have introduced a
> lot of comb filtering and such.
>
> So why is your way, which seems far more involved,
> more intricate, far less intuitive in the audio domain,
> Does not seem to be able to ever work in real time,
> in any way better?
Roy W. Rising
September 5th 06, 05:05 AM
wrote:
> Well, I at least have a first try at taking some samples out of music.
> Just listen to this loop I created with Wav2bmp:
> http://grimoire.genesismuds.com/stolen.wav
>
I listened. And ... ? (shrug)
--
~ Roy
"If you notice the sound, it's wrong!"
September 5th 06, 12:20 PM
wrote:
> > Totally unimpressive, as these effects can be easily
> > produced more easily by any number of existing means.
> > In this particular sample, you simply have introduced a
> > lot of comb filtering and such.
> >
> > So why is your way, which seems far more involved,
> > more intricate, far less intuitive in the audio domain,
> > Does not seem to be able to ever work in real time,
> > in any way better?
> Well, I at least have a first try at taking some samples out of music.
> Just listen to this loop I created with Wav2bmp:
> http://grimoire.genesismuds.com/stolen.wav
Not surprisingly, you utterly failed to address any of
the questions.
Your new samples are nothing unique, nothing new,
nothing that couldnlt be done faster, easier and probably
better on any number of existing techiques.
For all the bluster you'ce created, your samples are
notable for being ordinary, uninteresting, and boring.
Richard Smol
September 5th 06, 02:25 PM
wrote:
> My next project, for audio is going to be sampling. And I think I've
> found a way to isolate lyrics and tracks of music from one another.
> All you have to do is create a fuzzy sample of the music yourself. And
> then look at that sample as noise that you want to take out of whatever
> music you want to work with. So you can just erase the guitar from a
> song, and keep the drums or vocals. Cool idea right?
>
> Anyone else know of ways to do this?
Sounds like something already implemented in Cool Edit Pro (now Adobe
Audition). There you can use noise cancellation by actually subtracting
a background-noise from another track. This can for instance be abused
to extract the vocals from a track, a long as you have a part with
vocals only. It works fairly fine too, but sometimes leads to
artifacts... which may or may not be cool.
RS
September 5th 06, 05:07 PM
Thanks for the tip Richard Smol. I don't think I'll buy adobe
audition, but it should be hard to write my own noise filter, that
cancels out background sounds or other instruments. It is just pattern
recognition. I just need to find some people who have experience
working with noise cancelation or pattern recognition, and get a team
going. Can you think of any other software that does this kind of
work?
Richard Smol wrote:
> wrote:
> > My next project, for audio is going to be sampling. And I think I've
> > found a way to isolate lyrics and tracks of music from one another.
> > All you have to do is create a fuzzy sample of the music yourself. And
> > then look at that sample as noise that you want to take out of whatever
> > music you want to work with. So you can just erase the guitar from a
> > song, and keep the drums or vocals. Cool idea right?
> >
> > Anyone else know of ways to do this?
>
> Sounds like something already implemented in Cool Edit Pro (now Adobe
> Audition). There you can use noise cancellation by actually subtracting
> a background-noise from another track. This can for instance be abused
> to extract the vocals from a track, a long as you have a part with
> vocals only. It works fairly fine too, but sometimes leads to
> artifacts... which may or may not be cool.
>
> RS
philicorda
September 5th 06, 06:25 PM
On Tue, 05 Sep 2006 09:07:23 -0700, CoreyWhite wrote:
> Thanks for the tip Richard Smol. I don't think I'll buy adobe
> audition, but it should be hard to write my own noise filter, that
> cancels out background sounds or other instruments. It is just pattern
> recognition. I just need to find some people who have experience
> working with noise cancelation or pattern recognition, and get a team
> going. Can you think of any other software that does this kind of
> work?
Download a program called 'pure data'. (It's free.)
http://puredata.info/
Look at the example Help->Browser->Audio Examples->I09 Sheep from Goats.
It's an example of seperating sinusoids from noise, and is about as
simple an example of doing this as it gets. Which is still quite
complicated for me at least.
September 5th 06, 06:44 PM
I found another program too:
http://www.1esa.com/STC_SoundCleaner.htm
It is also free.
philicorda wrote:
> On Tue, 05 Sep 2006 09:07:23 -0700, CoreyWhite wrote:
>
> > Thanks for the tip Richard Smol. I don't think I'll buy adobe
> > audition, but it should be hard to write my own noise filter, that
> > cancels out background sounds or other instruments. It is just pattern
> > recognition. I just need to find some people who have experience
> > working with noise cancelation or pattern recognition, and get a team
> > going. Can you think of any other software that does this kind of
> > work?
>
> Download a program called 'pure data'. (It's free.)
> http://puredata.info/
> Look at the example Help->Browser->Audio Examples->I09 Sheep from Goats.
>
> It's an example of seperating sinusoids from noise, and is about as
> simple an example of doing this as it gets. Which is still quite
> complicated for me at least.
philicorda
September 5th 06, 07:56 PM
On Tue, 05 Sep 2006 10:44:03 -0700, CoreyWhite wrote:
> I found another program too:
> http://www.1esa.com/STC_SoundCleaner.htm
>
> It is also free.
Aye, it's free and it does ten times as many things as that simple 'sheep
from goats' example.... but it won't teach you anything about what's going
on 'under the hood', which you need to understand to actualise your
ambitious ideas.
Get into using pure-data. If you are into the more experimental side of
messing around with audio you'll have lots of fun with it.
September 5th 06, 08:17 PM
Yeah maybe I should. This sound cleaner software it awful. It is made
to run on windows 95.
philicorda wrote:
> On Tue, 05 Sep 2006 10:44:03 -0700, CoreyWhite wrote:
>
> > I found another program too:
> > http://www.1esa.com/STC_SoundCleaner.htm
> >
> > It is also free.
>
> Aye, it's free and it does ten times as many things as that simple 'sheep
> from goats' example.... but it won't teach you anything about what's going
> on 'under the hood', which you need to understand to actualise your
> ambitious ideas.
>
> Get into using pure-data. If you are into the more experimental side of
> messing around with audio you'll have lots of fun with it.
philicorda
September 5th 06, 08:36 PM
On Tue, 05 Sep 2006 12:17:50 -0700, CoreyWhite wrote:
> Yeah maybe I should. This sound cleaner software it awful. It is made
> to run on windows 95.
The one I suggested will be even worse :). It's really only the skeleton
of an example. You can play around with how it works though, and get some new sounds.
I still don't really understand how Fourier analyses and re-synthesis
works, but messing with the pd examples cleared it up a bit.
September 5th 06, 08:55 PM
philicorda wrote:
> On Tue, 05 Sep 2006 12:17:50 -0700, CoreyWhite wrote:
>
> > Yeah maybe I should. This sound cleaner software it awful. It is made
> > to run on windows 95.
>
> The one I suggested will be even worse :). It's really only the skeleton
> of an example. You can play around with how it works though, and get some new sounds.
>
> I still don't really understand how Fourier analyses and re-synthesis
> works, but messing with the pd examples cleared it up a bit.
No, I think pure data is exactly what I need! Thank you for telling me
about it. The Wav2bmp program I coded, was intended to be a way i
could do exactly what pure data is doing. Program and manipulate the
raw bytes of sound files. It should be hard to code an adaptive noise
filter with it. I've been reading papers written on doing it.
Just goog "adaptive background noise" and you should get a few hits of
other people coding programs that do this. You just isolate the
background noise and take it out. Easy enough.
What else can you do with pure data? I'm really excited already, and
am download the full program.
Roy W. Rising
September 6th 06, 12:46 AM
wrote:
"You just isolate the background noise and take it out. Easy enough."
Right. Have you thought about what you said? Even better, have you ever
heard the results of trying to do what you said? Easy enough? Not!
--
~ Roy
"If you notice the sound, it's wrong!"
Erwin Hessle
September 6th 06, 01:05 AM
Roy W. Rising wrote:
> wrote:
>
> "You just isolate the background noise and take it out. Easy enough."
>
> Right. Have you thought about what you said? Even better, have you ever
> heard the results of trying to do what you said? Easy enough? Not!
Don't forget Corey's a genius.
No, wait a minute. Nutjob, not genius.
Erwin Hessle, 8=3
Scott Dorsey
September 6th 06, 02:00 AM
Roy W. Rising > wrote:
wrote:
>"You just isolate the background noise and take it out. Easy enough."
>
>Right. Have you thought about what you said? Even better, have you ever
>heard the results of trying to do what you said? Easy enough? Not!
All you need is a filter that only passes signal and rejects noise. No
problem.
It's just like the filters that pass gasoline and filter out water. You
pump ocean water through them and eliminate oil spills, while running
your car on the output.
Cats use something like this principle... when they are lying on a dark
object they shed light hair while when they are lying on a light object
they shed dark hair.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Tomislav Crncic
September 6th 06, 06:55 AM
["Followup-To:" header set to alt.music.makers.dj.]
On 2006-09-05, > wrote:
> Thanks for the tip Richard Smol. I don't think I'll buy adobe
> audition, but it should be hard to write my own noise filter, that
> cancels out background sounds or other instruments. It is just pattern
> recognition. I just need to find some people who have experience
> working with noise cancelation or pattern recognition, and get a team
> going. Can you think of any other software that does this kind of
> work?
how do you plan to use pattern recognition to cancel out
the noise or background instruments? noise is random signal
which covers all frequencies, and for background instruments
every note has a different spectrum, and also levels can be
different for each note. good luck, but i do not think this
is ever going to work. at least not in our lifetime.
--
now all give tribute to g.c.coleman!
vBulletin® v3.6.4, Copyright ©2000-2025, Jelsoft Enterprises Ltd.