The Wagtunes Corner (Featuring My Best Yet)

[wagtunes]

You listen to this last track that I posted and you tell me that this sounds quantized.

https://soundcloud.com/steven-wagenheim ... lead-vocal

[donkey tugger]

2. Quantization. Doing it your way is too much work. As I said, I didn't quantize because I was lazy. So the last thing I'm going to do is quantize and then undo my work. That requires more work than I'm willing to put into a track.

Many sequencers (and some plug-ins, such as Kontakt) will have 'humanize' functions where you can add a small amount of timing randomization. If you're going for anything which is meant to be realistic, if you're not going to manually edit, I'd at least be adding a little bit of timing inconsistency - 2 minutes of a job!

I have not seen anything in my DAW called timing randomization. I wouldn't even know where to look for it. Besides, my ears are now so tuned to perfect quantization that IF I did find it and IF I did use it, I'd probably hate it anyway because I'd hear where it was off.

The point is to use it sparingly - a few percent at most. No drummer/guitarist/keyboard player is ever going to play exactly on the beat every note - there'll always be small (and it doesn't need to be much) deviations, and this is what makes things sound more natural.

But you want to know what the irony is? In the 100 plus tracks that I've posted here, if you read through all the complaints, you will read...

Hate Vocaloid
Orchestral arrangements don't sound real
Mix is uneven

Not ONE person ever said to me, out of 100 plus tracks...

"Damn man! How come your tracks sound so robotic? What do you do? Quantize everything to death?"

And the reason nobody has said that is because they DON'T sound quantized to death. The only reason this has even become an issue is because I opened my big mouth. Had I not done that, this issue NEVER comes up.

That's what I find laughable about all of this. This is typical KVR.

"Ooo, ooo, ooo. He just said something that we can jump all over him for. He quantizes everything. Let's hammer him for it."

Even though NOBODY could even tell.

LMAO. Gotta love this place.

Easy tiger, was just a suggestion. Won't bother again.

[sleepcircle]

i don't think anyone was hammering you. it didn't sound like it to me, anyway.


i mentioned something i do, and you said you wouldn't/couldn't do it. the other guy offered suggestions. you presented a doubt. the guy replied with a clarification as to what he meant.

it was just the flow of conversation. he was trying to offer you options which he didn't think you knew about. you never actually said "please stop talking about this because i'm definitely not going to do anything else."

if you had, i'm sure neither of us would've continued talking about the subject.


i personally didn't say you HAD to humanize your stuff. Tonnes of people arrange stuff in perfect time—and i didn't mean to imply you had to do any differently; I just mentioned why I did what i did. I forgot that it could seem like a veiled criticism. I'm sorry for not watching my tongue better.



anyway, this has angered you, clearly, so I don't think anyone will mention it again.

[wagtunes]

i don't think anyone was hammering you. it didn't sound like it to me, anyway.


i mentioned something i do, and you said you wouldn't/couldn't do it. the other guy offered suggestions. you presented a doubt. the guy replied with a clarification as to what he meant.

it was just the flow of conversation. he was trying to offer you options which he didn't think you knew about. you never actually said "please stop talking about this because i'm definitely not going to do anything else."

if you had, i'm sure neither of us would've continued talking about the subject.


i personally didn't say you HAD to humanize your stuff. Tonnes of people arrange stuff in perfect time—and i didn't mean to imply you had to do any differently; I just mentioned why I did what i did. I forgot that it could seem like a veiled criticism. I'm sorry for not watching my tongue better.



anyway, this has angered you, clearly, so I don't think anyone will mention it again.

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

In the case of Cubase, you end up with movements that end up being detrimental to each other because it's not "smart" randomization. So after you've done this, you still have to go back and listen to your track and move the notes that are obviously moved too much in either direction. It's a sloppy process that nobody has perfected yet. It's not like absolute quantization where the software will absolutely set each note to an exact value. That's easy. Random? Not so easy and requires a lot of "after work" to get it to sound right or you end up with some serious timing issues.

The Cubase forums are littered with this problem.

So again, it all comes back to how much work I want to put into making my stuff sound "human" given I'm using a robot for vocals.

I can't be the only one who sees the ridiculousness of doing this.

[sleepcircle]

i wasn't thinking that you were a stubborn mule. and yes, automatic randomization does present those problems.

EDIT: as an addendum, i realize that specialty forums are usually full of know-it-all snobs who have no compunction about being cruel and arrogant while shielding themselves under the umbrella that they are "simply being honest."

i cannot guess how much you have received in 21 pages worth of music and 100 songs, but it was not my intention to add to it. I apologize.

[wagtunes]

i wasn't thinking that you were a stubborn mule. and yes, automatic randomization does present those problems.

EDIT: as an addendum, i realize that specialty forums are usually full of know-it-all snobs who have no compunction about being cruel and arrogant while shielding themselves under the umbrella that they are "simply being honest."

i cannot guess how much you have received in 21 pages worth of music and 100 songs, but it was not my intention to add to it. I apologize.

No problem. While I have you, this is a track I posted earlier. Does it suffer from the same vocal problem?

https://soundcloud.com/steven-wagenheim ... ittle-at-a

[Frantz]

https://soundcloud.com/steven-wagenheim ... lead-vocal

Good one! Strong melody and chord changes throughout.

[wagtunes]

https://soundcloud.com/steven-wagenheim ... lead-vocal

Good one! Strong melody and chord changes throughout.

Thanks Frantz.

[ghettosynth]

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

Not to be a pedant, but, this is not really "the reason." It's random enough to be called random for all practical purposes. The "reason" is that random isn't really a good solution for humanization. That is, humans have variation but are far less "random" than any bog standard computer random number generator.

This is why "groove" quantization was invented. Whether it's a good solution or not isn't my claim here, only that it tries to capture how humans vary their timing.

I'm not saying that there isn't a certain amount of randomness in human playing, there is, but it's not a complete solution to humanization to just apply randomness to timing, even in small amounts.

[wagtunes]

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

Not to be a pedant, but, this is not really "the reason." It's random enough to be called random for all practical purposes. The "reason" is that random isn't really a good solution for humanization. That is, humans have variation but are far less "random" than any bog standard computer random number generator.

This is why "groove" quantization was invented. Whether it's a good solution or not isn't my claim here, only that it tries to capture how humans vary their timing.

I'm not saying that there isn't a certain amount of randomness in human playing, there is, but it's not a complete solution to humanization to just apply randomness to timing, even in small amounts.

I actually agree with you, which is why I quantize perfectly and leave it at that. The alternative is my stuff sounds even worse than in does now.

[wagtunes]

Title track from my CD "No Signal"

I came up with the idea for this CD from my PC. When I shut it down and my monitor loses power, it pops up a message in blue "No Signal"

You can't make this stuff up.

No Signal

Night screams
Bad dreams
Cold sweats
Regrets

I'm scared
You dared
The cost
All lost

No signal coming from you
No signal coming unglued
No signal tell me what to do

No signal life is unfair
No signal no one seems to care
No signal reach me if you dare

New moon
Same tune
I come
You run

This game
Must change
Right now
Some how

No signal coming from you
No signal coming unglued
No signal tell me what to do

No signal life is unfair
No signal no one seems to care
No signal reach me if you dare

https://soundcloud.com/steven-wagenheim ... /no-signal

[ghettosynth]

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

Not to be a pedant, but, this is not really "the reason." It's random enough to be called random for all practical purposes. The "reason" is that random isn't really a good solution for humanization. That is, humans have variation but are far less "random" than any bog standard computer random number generator.

This is why "groove" quantization was invented. Whether it's a good solution or not isn't my claim here, only that it tries to capture how humans vary their timing.

I'm not saying that there isn't a certain amount of randomness in human playing, there is, but it's not a complete solution to humanization to just apply randomness to timing, even in small amounts.

I actually agree with you, which is why I quantize perfectly and leave it at that. The alternative is my stuff sounds even worse than in does now.

Well, ok, but, perfect quantization doesn't make it human either. "You" make it human, by definition. But, let's talk about me so that it doesn't sound like I'm criticizing your music, which is not my intent.

My playing varies from ok, to wtf was I thinking. So, I have to quantize so that it doesn't sound like "bad human" playing. But, quantization doesn't make it sound like "good human" playing, it will often sound robotic. When my playing is ok, I can just tighten it up by using less than 100% quantization and I think that this is an ok solution to the problem. You are still retaining some of the human feel while moving all notes closer to the grid. This is much less effective when my playing isn't good at all, however.

One thing that groove quantization does is vary the amplitude as well as the timing. I've found that this idea, along with randomness applied to certain events happening, e.g., probabilistic drum hits, helps to make it more human. Again though, you can't just apply uniform randomness to the hits to expect it to sound real, the probabilities have to have some similarity to humanistic playing.

Anyway, I do think that less than 100% quantization can help. Randomness can also help to some extent but only in very small amounts and it's usually better when applied in limited contexts, e.g., one voice of two playing the same part. I think that the key with all of these tools is to not apply them blindly but to think about how what they do plays a similar role to what a musician would do.

[wagtunes]

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

Not to be a pedant, but, this is not really "the reason." It's random enough to be called random for all practical purposes. The "reason" is that random isn't really a good solution for humanization. That is, humans have variation but are far less "random" than any bog standard computer random number generator.

This is why "groove" quantization was invented. Whether it's a good solution or not isn't my claim here, only that it tries to capture how humans vary their timing.

I'm not saying that there isn't a certain amount of randomness in human playing, there is, but it's not a complete solution to humanization to just apply randomness to timing, even in small amounts.

I actually agree with you, which is why I quantize perfectly and leave it at that. The alternative is my stuff sounds even worse than in does now.

Well, ok, but, perfect quantization doesn't make it human either. "You" make it human, by definition. But, let's talk about me so that it doesn't sound like I'm criticizing your music, which is not my intent.

My playing varies from ok, to wtf was I thinking. So, I have to quantize so that it doesn't sound like "bad human" playing. But, quantization doesn't make it sound like "good human" playing, it will often sound robotic. When my playing is ok, I can just tighten it up by using less than 100% quantization and I think that this is an ok solution to the problem. You are still retaining some of the human feel while moving all notes closer to the grid. This is much less effective when my playing isn't good at all, however.

One thing that groove quantization does is vary the amplitude as well as the timing. I've found that this idea, along with randomness applied to certain events happening, e.g., probabilistic drum hits, helps to make it more human. Again though, you can't just apply uniform randomness to the hits to expect it to sound real, the probabilities have to have some similarity to humanistic playing.

Anyway, I do think that less than 100% quantization can help. Randomness can also help to some extent but only in very small amounts and it's usually better when applied in limited contexts, e.g., one voice of two playing the same part. I think that the key with all of these tools is to not apply them blindly but to think about how what they do plays a similar role to what a musician would do.

I understand what you're saying. But doesn't something like a kick and a crash hitting on the same beat have to be right on the beat or it sounds like crap because of the fast attack and release? You can absolutely hear if the crash is slightly ahead or behind the beat and it sounds like garbage.

Agree, disagree?

[ghettosynth]

Just to show I'm not a total stubborn mule, I looked up random quantization in Cubase. It's simple to execute. The problem is, according to what I've read, it has very unpredictable results. In short, it doesn't work all that well, which doesn't surprise me. The reason is because random algos aren't really random. It's like shuffle mode on your iPod. The algo built into it, by definition, makes it not really random.

Not to be a pedant, but, this is not really "the reason." It's random enough to be called random for all practical purposes. The "reason" is that random isn't really a good solution for humanization. That is, humans have variation but are far less "random" than any bog standard computer random number generator.

This is why "groove" quantization was invented. Whether it's a good solution or not isn't my claim here, only that it tries to capture how humans vary their timing.

I'm not saying that there isn't a certain amount of randomness in human playing, there is, but it's not a complete solution to humanization to just apply randomness to timing, even in small amounts.

I actually agree with you, which is why I quantize perfectly and leave it at that. The alternative is my stuff sounds even worse than in does now.

Well, ok, but, perfect quantization doesn't make it human either. "You" make it human, by definition. But, let's talk about me so that it doesn't sound like I'm criticizing your music, which is not my intent.

My playing varies from ok, to wtf was I thinking. So, I have to quantize so that it doesn't sound like "bad human" playing. But, quantization doesn't make it sound like "good human" playing, it will often sound robotic. When my playing is ok, I can just tighten it up by using less than 100% quantization and I think that this is an ok solution to the problem. You are still retaining some of the human feel while moving all notes closer to the grid. This is much less effective when my playing isn't good at all, however.

One thing that groove quantization does is vary the amplitude as well as the timing. I've found that this idea, along with randomness applied to certain events happening, e.g., probabilistic drum hits, helps to make it more human. Again though, you can't just apply uniform randomness to the hits to expect it to sound real, the probabilities have to have some similarity to humanistic playing.

Anyway, I do think that less than 100% quantization can help. Randomness can also help to some extent but only in very small amounts and it's usually better when applied in limited contexts, e.g., one voice of two playing the same part. I think that the key with all of these tools is to not apply them blindly but to think about how what they do plays a similar role to what a musician would do.

I understand what you're saying. But doesn't something like a kick and a crash hitting on the same beat have to be right on the beat or it sounds like crap because of the fast attack and release? You can absolutely hear if the crash is slightly ahead or behind the beat and it sounds like garbage.

Agree, disagree?

Well, yes and no. That is standard advice. Lock the on beat notes and play with the timing of the off beat notes, e.g., clap/snare. But it doesn't always work well that way and it depends a lot on style. You can move them forward or backward (in tiny amounts) to promote a feel, but again, like a drummer. If you move them all randomly, it just sounds bad, if you move all of them forward or backward by the same amount then you will just sound like a perfectly timed but off-beat drummer. So you have to do more than just select every single note and say "apply random quantization."

Usually the first beat in some groove determined time frame, e.g., one, two, four, or eight measures, should be on. But, the rest can vary depending on the groove.

But, with respect to my advice for varying one of two voices randomly, I wasn't really talking about kick and crash. That said, I can guarantee you that NO human in the world plays them exactly on time (assuming that the crash is played with an arm and the kick with a leg) and that EVERY human in the world has some randomness AND groove in terms of the relationship in time between the two hits. Point being, if it sounds like crap, then you are using too much randomness/groove. Moreover, if you are locking them perfectly, then you sound like no drummer in existence because drummers are human.

[wagtunes]

@ghetto

Translation of all the above. It is a crap ton of work to really make this stuff sound human and good at the same time.

And that's why I quantize and call it a day.

Unless you have an easy way to do this and somehow I doubt that.