- Would such an analysing system once exist ?
- KVRAF
- 6504 posts since 25 May, 2002 from Bobo-dioulasso\BF__Geneva/CH
By advice from someone from the Camel audio team, i' m asking you this :
If you’ve got samples of orchestral hits (like many song’s end of pop musics) or samples of sustained interval or harmonies
Samples with no other musical gesture than a single, decaying hit or a sustained harmony
Would actually exist, or in a near future, a technology able to simply recognize the single interval or harmony contained within these samples ?
If you’ve got samples of orchestral hits (like many song’s end of pop musics) or samples of sustained interval or harmonies
Samples with no other musical gesture than a single, decaying hit or a sustained harmony
Would actually exist, or in a near future, a technology able to simply recognize the single interval or harmony contained within these samples ?
-
- KVRAF
- 6937 posts since 4 Jun, 2004 from Utrecht, Holland
Probably: yes
Software like BandInABox is already able to detect it when I play a chord on a midi keyboard. Use FFT to detect the notes being played in a sample, connect that with a chord detection algorythm and its there!
Now somebody still has to develop it...
Software like BandInABox is already able to detect it when I play a chord on a midi keyboard. Use FFT to detect the notes being played in a sample, connect that with a chord detection algorythm and its there!
Now somebody still has to develop it...
-
- KVRist
- 492 posts since 26 Feb, 2003 from Vancouver BC
Er... when you play a chord on a midi keyboard, you're telling Band-In-A-Box *exactly* what notes you're playing. That's kinda the nature of a midi keyboard. Given that specific note information, determining the name of the chord is a rocket-science free task.
Analysing an audio only sample and parsing the notes (and subsequent chord) being played is doable, but the accuracy will diminish somewhat based on the harmonic content of the instruments participating and the instrumental density of the sample.
K
Analysing an audio only sample and parsing the notes (and subsequent chord) being played is doable, but the accuracy will diminish somewhat based on the harmonic content of the instruments participating and the instrumental density of the sample.
K
-
- KVRist
- 401 posts since 4 May, 2004
I really have a hard time understanding you, but if I'm correct what you're looking for is Akoff Music Composer it will do a good job on most monophonic lines, and even a decent one on polyphonic ones. It won't recognize chords (by name) or anything like that, but it will try to give you MIDI reconstruction of pitches in analyzed wave.
- KVRAF
- Topic Starter
- 6504 posts since 25 May, 2002 from Bobo-dioulasso\BF__Geneva/CH
Thanks for the tips, but what i'm looking actually doesn't exist :peejunk wrote:I really have a hard time understanding you, but if I'm correct what you're looking for is Akoff Music Composer it will do a good job on most monophonic lines, and even a decent one on polyphonic ones. It won't recognize chords (by name) or anything like that, but it will try to give you MIDI reconstruction of pitches in analyzed wave.
I used to create sampler's patches using those kind of samples as described
(song's end for one-shot or pianistic envelopes , sustained harmonies once looped for pad-like envelopes)
In these type of samples, there is very often a note is more intense, that as a higher volume than the other in the harmony...and this is for me the usual (and easier) choice for tuning the sample during edition, though it is the tonic of the harmony or not...
Sounds amazing to create complex harmonies, just by playing a chord when your sampler is based itself by another chord contained in ONE sample
( or interval, the less complex is it content, the most flexible is the use you can have of it )
...but you can imagine that the trick has its limits, especially if you want to stay inside a scale
And what i'm dreaming of at this point of my work is that it would be awesome, after analysing a sample, w haveo a simplified (or even automated ) assignation of them on keyboard relative to their type (major, minor, diminished, etc...) and the scale(s) in wish you're composing a song
They would be, of course many major aspects to consider, and the chord analysing is one of them,
And i dont know if it would be such a big pain in the xxx if it only task basically would be to detect the 2, 3 ( or maybe more ) EXPECTED FONDAMENTALS that a sample is supposed to contain, It seems to me that if you want to recognize a kind of chord inside a sample, the type of samples as described above should be the more appropriate to do such analyse with some success :
( just a single, static, interval or harmony, without any other musical variations or gesture )
And i'm quite persuaded that it if these features would once be developped in samplers, they could be as useful and creative in their own way than all the loop tools and other "rythmical-based" beatslicers are
( well...at least for me, of course, but... )
- KVRAF
- Topic Starter
- 6504 posts since 25 May, 2002 from Bobo-dioulasso\BF__Geneva/CH
In fact, would you please excuse me because i haven't read your post carefullyKrakatau wrote:Thanks for the tips, but what i'm looking actually doesn't exist :peejunk wrote:I really have a hard time understanding you, but if I'm correct what you're looking for is Akoff Music Composer it will do a good job on most monophonic lines, and even a decent one on polyphonic ones. It won't recognize chords (by name) or anything like that, but it will try to give you MIDI reconstruction of pitches in analyzed wave.
I used to create sampler's patches using those kind of samples as described
(song's end for one-shot or pianistic envelopes , sustained harmonies once looped for pad-like envelopes)
In these type of samples, there is very often a note is more intense, that as a higher volume than the other in the harmony...and this is for me the usual (and easier) choice for tuning the sample during edition, though it is the tonic of the harmony or not...
Sounds amazing to create complex harmonies, just by playing a chord when your sampler is based itself by another chord contained in ONE sample
( or interval, the less complex is it content, the most flexible is the use you can have of it )
...but you can imagine that the trick has its limits, especially if you want to stay inside a scale
And what i'm dreaming of at this point of my work is that it would be awesome, after analysing a sample, w haveo a simplified (or even automated ) assignation of them on keyboard relative to their type (major, minor, diminished, etc...) and the scale(s) in wish you're composing a song
They would be, of course many major aspects to consider, and the chord analysing is one of them,
And i dont know if it would be such a big pain in the xxx if it only task basically would be to detect the 2, 3 ( or maybe more ) EXPECTED FONDAMENTALS that a sample is supposed to contain, It seems to me that if you want to recognize a kind of chord inside a sample, the type of samples as described above should be the more appropriate to do such analyse with some success :
( just a single, static, interval or harmony, without any other musical variations or gesture )
And i'm quite persuaded that it if these features would once be developped in samplers, they could be as useful and creative in their own way than all the loop tools and other "rythmical-based" beatslicers are
( well...at least for me, of course, but... )
indeed yes,
if that program (PC only, unfortunately for me...) can analyse an entire interpretation of a single polyphonic instrument like a piano or guitar. evenafter some necessary adjustment, it should be able to do the analyse of a simple and static content of the samples i'm refering of...
-
- KVRAF
- 13446 posts since 14 Nov, 2000 from Hannover / Germany
My wild guess would be that one day we'll see such a program (or let's better say, an algorhythm, which then could be implemented in all sorts of ways).
Thing is, the human ear is allready able to analyze complexed chords, regardless of intense harmonic content. So, a program after all should be able to do it as well.
But then: The trained human ear is pretty good at distracting actual musical content (played notes) from "unused" stuff (harmonics or frequencies that we might not even be able to hear).
A program would just analyze *everything* in a given sound, and I could very well imagine that it'd be quite hard to decide whether something is actually a played note or only a result of some notes being played together. The ear (well, actually the brain) is doing such things automatically, and this sort of "filter" would IMO be the toughest thing to code.
Still, I do believe that we will see such an algorhythm one day.
In fact, for monophonic material it's almost working perfectly allready (just have a look at Melodyne, it even works great on signals containing rich overtones), so it'd only be a matter of someone teachong an algorhythm to divide polyphonic material into useable pitches (hm, I should have a look at that Arkoff thingy)
I wouldn't happen to know whether there's strong enough interest in such a thing though... maybe, if everybody cried out loud, coders would hurry up a bit more. But to me it seems not many people would actually need it.
Interesting thing though.
Thing is, the human ear is allready able to analyze complexed chords, regardless of intense harmonic content. So, a program after all should be able to do it as well.
But then: The trained human ear is pretty good at distracting actual musical content (played notes) from "unused" stuff (harmonics or frequencies that we might not even be able to hear).
A program would just analyze *everything* in a given sound, and I could very well imagine that it'd be quite hard to decide whether something is actually a played note or only a result of some notes being played together. The ear (well, actually the brain) is doing such things automatically, and this sort of "filter" would IMO be the toughest thing to code.
Still, I do believe that we will see such an algorhythm one day.
In fact, for monophonic material it's almost working perfectly allready (just have a look at Melodyne, it even works great on signals containing rich overtones), so it'd only be a matter of someone teachong an algorhythm to divide polyphonic material into useable pitches (hm, I should have a look at that Arkoff thingy)
I wouldn't happen to know whether there's strong enough interest in such a thing though... maybe, if everybody cried out loud, coders would hurry up a bit more. But to me it seems not many people would actually need it.
Interesting thing though.
There are 3 kinds of people:
Those who can do maths and those who can't.
Those who can do maths and those who can't.
-
- Banned
- 12367 posts since 30 Apr, 2002 from i might peeramid
i expect it might be entirely possible offline, eg. voice analysis s/w like praat can give you several bands of formant data, but next someone will ask for it in realtime, VST format please :p pending your transdimensional pheripherals.
you come and go, you come and go. amitabha neither a follower nor a leader be tagore "where roads are made i lose my way" where there is certainty, consideration is absent.
- KVRAF
- Topic Starter
- 6504 posts since 25 May, 2002 from Bobo-dioulasso\BF__Geneva/CH
Well...xoxos wrote:i expect it might be entirely possible offline, eg. voice analysis s/w like praat can give you several bands of formant data, but next someone will ask for it in realtime, VST format please :p pending your transdimensional pheripherals.
It seems the particular case that i was thinking about in relation with sample's assignation to a scale doesn't require a real-time process,...that gives me some hope !!
- KVRAF
- 2549 posts since 7 Jul, 2003 from Huntington, WV
I think this is a task that is easier for a human to execute. I really don't think there is currently any software which does exactly what you want. Also, what if your original source material is in tune with itself, but not recorded at "concert pitch"? How will the computer automatically know what to do with such recordings?
On a slightly different topic, does anyone make a spectrum analyzer which is calibrated in musical notes, instead of Hertz? It might make things easier for Krakatau if he could see a spectral plot of a waveform containing a chord, and instantly see how the peaks of the plot lined up with various notes of the musical scale. In fact, I would like to have that myself.
There has been lots of times when I looked at a spectral plot and noted that a certain peak or dip in the spectrum was so centered on a value of x Hertz, and then had to translate that into the nearest musical note. If the legend of the spectral plot were labeled in units of musical notes, that would eliminate the translation task from my workflow.
Does that simple capability exist in any current software?
thanks,
McLilith
On a slightly different topic, does anyone make a spectrum analyzer which is calibrated in musical notes, instead of Hertz? It might make things easier for Krakatau if he could see a spectral plot of a waveform containing a chord, and instantly see how the peaks of the plot lined up with various notes of the musical scale. In fact, I would like to have that myself.
There has been lots of times when I looked at a spectral plot and noted that a certain peak or dip in the spectrum was so centered on a value of x Hertz, and then had to translate that into the nearest musical note. If the legend of the spectral plot were labeled in units of musical notes, that would eliminate the translation task from my workflow.
Does that simple capability exist in any current software?
thanks,
McLilith
-
- KVRist
- 245 posts since 1 Mar, 2004 from London
Transcribe http://www.seventhstring.com/xscribe/overview.html
used to be pretty handy at that. I briefly tried version 5 and it is now on 7 so try it out.
It had a keyboard mapped out and the wave you were working on and by highlighting part of the wave it would show you over the keyboard all the "notes" being played (though obviously some are harmonics). It would also play back at half speed to play along - its a musicians tool. I think it would do this job really well. It helped me figure out some Thelonious Monk chords, and that is no easy job
Steve
used to be pretty handy at that. I briefly tried version 5 and it is now on 7 so try it out.
It had a keyboard mapped out and the wave you were working on and by highlighting part of the wave it would show you over the keyboard all the "notes" being played (though obviously some are harmonics). It would also play back at half speed to play along - its a musicians tool. I think it would do this job really well. It helped me figure out some Thelonious Monk chords, and that is no easy job
Steve
-
- KVRian
- 1325 posts since 1 Sep, 2004
YES !
This analysis technology already exists !
It's called EARS (in connection with a human brain).
Musical experience may complete the system...
Sorry for my sarcasm, but's like to try to overdrive a mosquito with a monster track ( merlely to extract the harmonies of his dead-scream while doing that ).

This analysis technology already exists !
It's called EARS (in connection with a human brain).
Musical experience may complete the system...
Sorry for my sarcasm, but's like to try to overdrive a mosquito with a monster track ( merlely to extract the harmonies of his dead-scream while doing that ).
- KVRAF
- Topic Starter
- 6504 posts since 25 May, 2002 from Bobo-dioulasso\BF__Geneva/CH
Sorry because, i'm the lame wanker in chief, that has left me completely depth and i need an alternative solution you see...jackle&hyde wrote:YES !
This analysis technology already exists !
It's called EARS (in connection with a human brain).
Musical experience may complete the system...
Sorry for my sarcasm, but's like to try to overdrive a mosquito with a monster track ( merlely to extract the harmonies of his dead-scream while doing that ).
- KVRAF
- 2549 posts since 7 Jul, 2003 from Huntington, WV
I see arrogance and sarcasm in almost all of your posts, but at least this time you seem to be attempting a level of playful sarcasm.jackle&hyde wrote:YES !
This analysis technology already exists !
It's called EARS (in connection with a human brain).
Musical experience may complete the system...
Sorry for my sarcasm, but's like to try to overdrive a mosquito with a monster track ( merlely to extract the harmonies of his dead-scream while doing that ).
I have to admit, J&H, this is a big improvement.
McLilith
-
- KVRian
- 1325 posts since 1 Sep, 2004
Sorry Guy.
But I give a shit of what your opinion is.
(This statement is very personally to the addressed person above. Because he tries to judge me whereever he can but never being asked to to that.)
But I give a shit of what your opinion is.
(This statement is very personally to the addressed person above. Because he tries to judge me whereever he can but never being asked to to that.)
