Insights for mixing on an MPC?

djst wrote: I was interested in eliminating any unintentional clipping/distortion and simply wanted to validate what sounded like made up claims from yet another MPC YouTuber who's more interested in profiting from other people than to share the love and knowledge. I guess that's a part of the MPC community DNA that I'll simply have to accept.

Oh come on, give it a rest, this site has over 1.5 million posts, most of which are MPC owners freely helping each other for the past 20 years. You ask one question here and get a couple of replies pointing you to some paid resources and you won't stop yapping about it. Take a look around, this forum has always primarily been about free exchange of knowledge. Feel free to share some of yours and stop ya whining!

As for that youtuber, I don't know him, but these guys spend the time putting up free tutorials and they owe us nothing in return. I don't have an issue with the guy using his free tutorial as a way of picking up potential clients.

I'm reading that the MPC converts all audio to 32 bit float internal format when loading, so I would assume that inside the MPC's signal path things are kept in 32 bit float, and so there is no clipping +0dB. Except in distortion/clipping effects, compressors etc. With floats there is no clipping when adding numbers. 32 bit floats "clip" somewhere around +1500dB. But you get problems when your signal leaves the floating point world and enters integer world, then you'll have to scale it down to fit the target number range, so that's why it's a good idea to keep things at reasonable levels. Things are designed with this in mind - if you use compressors etc., the range of controls is so that maximum somewhere near 0. I don't think you'll find a compressor plugin that lets you set the threshold to +1500dB even though it would be technically possible. The same goes with distortion/overdrive plugins, tape saturation, etc., they are designed for a certain range of signal magnitudes. But most of the plugins won't clip "digitally", like some very old DAWs used to. Is any modern DAW using integers internally?

Does the MPC have some kind of in-between hard clipping?

On main window, you have a master meter.
Never get to the red, always stay a bit under it and all will be fine, keygroups, autosampling, mixing and so on.

NearTao wrote:@djst - I suspect you're getting stuck in the land of objective vs subjective. Also as much what is observable/testable and what people are speculating. So you are right in questioning everything (and even what I am about to post) because a lot of information is just "unknown" how it is really implemented, and it isn't like manufacturers are always going to come clean or explain what they implemented for a lot of reasons like secret sauce or due to licensing constraints or whatever.

Specifically for the clipping issue though, as I am sure you are aware in most DAWs, that anything over a 0db signal you will get digital clipping, or in other words your signal gets cut off as it goes over 0db. More bits does not magically fix this issue, it is a hard limit, but what manufacturers can do is change where 0db hits. They can do this is 16bit, 24bit, 32bit, 64bit, etc... all they do is change their meters to have what is really -4db for example to read as 0db, and this would give them 4db of headroom at the expense of lower resolution. at 16 or 24 bit you might notice this, but at 64bit or 128bit you're very unlikely to hear this, so you can provide even more headroom.

On the MPC I would say it is not good practice to play your signals hot, in other words at or above 0db, except for the mixer levels that will show above 0db (I suspect the meter is using a trick like -4db reading as 0db for visual purposes only), as far as my experiments have shown, it clips hard at 0db anywhere in the chain. This is for any part of the chain, so if you have samples normalized to 0db and play them polyphonically you are going to get digital clipping if you don't reduce the audio level at the sample, and you want to also consider the Program master level, as well as the track level. Similarly for plugins, you want to make sure that you are exiting the plugin below 0db (unless you want that sound). Personally I try to start at -9 to -15db, and then adjust to taste in the mixer section from there.

This is mostly from what I have observed. It is very possible I am wrong, but from the video I do believe that the author is wrong or at least mischaracterizing some of the problems in the signal chain.

Hope that helps...

This was the most constructive and helpful answer in this thread - thank you!

Your observation that the MPC really does clip at its stated 0dB anywhere in the signal chain except at the channel mixer is very valuable info. This might explain why I've heard some audible distortion/clipping at times when I really wasn't expecting it.

This also suggests that the MPC works very differently from the typical DAW. This excellent video made by the producers of Reason goes into detail about its signal chain works, and the basic message is that there's practically no such thing as a signal running too hot in Reason (except of course at the very final master output): https://www.youtube.com/watch?v=T-TpPLzRpsU

I was hoping the MPC would behave similarly, but your reply suggests it doesn't, so I really should pay attention to the signal earlier in the chain. Like the guy on that earlier YouTube video is suggesting too, so I guess that means he's mostly right in the end!

HouseWithoutMouse wrote:I'm reading that the MPC converts all audio to 32 bit float internal format when loading, so I would assume that inside the MPC's signal path things are kept in 32 bit float, and so there is no clipping +0dB. Except in distortion/clipping effects, compressors etc. With floats there is no clipping when adding numbers. 32 bit floats "clip" somewhere around +1500dB. But you get problems when your signal leaves the floating point world and enters integer world, then you'll have to scale it down to fit the target number range, so that's why it's a good idea to keep things at reasonable levels. Things are designed with this in mind - if you use compressors etc., the range of controls is so that maximum somewhere near 0. I don't think you'll find a compressor plugin that lets you set the threshold to +1500dB even though it would be technically possible. The same goes with distortion/overdrive plugins, tape saturation, etc., they are designed for a certain range of signal magnitudes. But most of the plugins won't clip "digitally", like some very old DAWs used to. Is any modern DAW using integers internally?

Does the MPC have some kind of in-between hard clipping?

Right, this is what I'd assume personally before reading some answers in this thread, in other words that it behaves like any modern DAW and simply won't clip anywhere but at the end of the signal line when converting down to the integer domain. I mean why would the MPC be designed so that the audio signal runs the risk of clipping just because you play a chord on a software synth that is playing one note at near 0 dB? It just seems stupid to me. Hence me questioning the video advice.

There's no clipping inside the signal path. (Except in plugins which deliberately clip the signal to some level.) You can test it yourself with the following effect chain in a program's insert effects, for example the Electric plugin:
- Insert 1 : AIR Channel Strip, EQ:OFF, GATE:OFF, COMP:OFF, OUTPUT:+24 dB
- Insert 2 : AIR Channel Strip, EQ:OFF, GATE:OFF, COMP:OFF, OUTPUT:+24 dB
- Insert 3 : AIR Channel Strip, EQ:OFF, GATE:OFF, COMP:OFF, OUTPUT:+24 dB
- Insert 4 : AIR Channel Strip, EQ:OFF, GATE:OFF, COMP:OFF, OUTPUT:+24 dB

And this one in one of the Master Output inserts:
- Insert : AIR Channel Strip, EQ:OFF, GATE:OFF, COMP:OFF, OUTPUT:-96 dB

The end result is: no change, no clipping, no nothing. 4 x +24 dB + 1 x -96 dB = +- 0 dB.

This is because the internal signal processing is done with floating point numbers. There is no clipping, unless a plugin deliberately clips/limits/compresses because it's a clipper/limiter/compressor plugin.

If the insert effect in the master output is bypassed, there is HORRIBLE CLIPPING because of +96 dB gain hitting the master outputs.

If you want, you can add more gain to the plugin. Route it via a submix and add another 4 x +24dB of gain there. Then counter it with another -96 dB in the main output. No clipping, even though between the submix and the master, there is +192 dB of gain. But no clipping. Or at least there wasn't on my MPC One when I tested this.

Look at what the main output's first AIR Channel Strip's input level meter is showing for a signal with no clipping: (playing Finlandia by Sibelius on the Electric piano plugin, beautiful song)


The level meter bars naturally max out, but the signal isn't clipped.

Housewithoutmouse wind the loudness wars.

What you did was very helpful. A pseudo-scientific test to see where clipping/ distortion occurs and if it occurs. Keep up the good work!

HouseWithoutMouse wrote:There's no clipping inside the signal path. (Except in plugins which deliberately clip the signal to some level.) You can test it yourself with the following effect chain in a program's insert effects, for example the Electric plugin:
...

The level meter bars naturally max out, but the signal isn't clipped.

Hats off to you, this was the answer I was looking for! I think we can finally settle this topic (yes, the YouTube guy is no doubt wrong and is just hoping to prey on ignorance to sell his services) and I'm glad that logic wins in the end.

Thank you HouseWithMouse!

That doesn't necessarily have to mean much in practice. You'll still have to keep your levels at reasonable ranges, where the plugins have been designed to operate. Avoiding going over 0 is a good principle. If you do it because you think you'll get digital clipping inside the channel... on some other digital systems you might get that, just not here. But you want to maintain control over your mix, and you do that by being methodical, setting targets and "avoid" limits.

But it's good to know your tools. Usually there's internal headroom above 0 ... how much, and what happens when you go over it, depends. Some mixers clip softly at some point, maybe even producing nice results, others might clip in a harsh way. In floating point based DAW mixers there's virtually unlimited headroom. Some DAWs emulate analog soft-clipping/saturating behavior, like Harrison Mixbus (if I remember correctly).

MPC-Tutor wrote:

djst wrote: I was interested in eliminating any unintentional clipping/distortion and simply wanted to validate what sounded like made up claims from yet another MPC YouTuber who's more interested in profiting from other people than to share the love and knowledge. I guess that's a part of the MPC community DNA that I'll simply have to accept.

Oh come on, give it a rest, this site has over 1.5 million posts, most of which are MPC owners freely helping each other for the past 20 years. You ask one question here and get a couple of replies pointing you to some paid resources and you won't stop yapping about it. Take a look around, this forum has always primarily been about free exchange of knowledge. Feel free to share some of yours and stop ya whining!

As for that youtuber, I don't know him, but these guys spend the time putting up free tutorials and they owe us nothing in return. I don't have an issue with the guy using his free tutorial as a way of picking up potential clients.

djst wrote: ...more interested in profiting from other people than to share the love and knowledge. I guess that's a part of the MPC community DNA that I'll simply have to accept.

You know, this isn't the fastest moving forum. Sometimes it can take people a while to get an answer to something, and I've always imagined thats because people are somewhere else, making music or some shit, but not sitting at their computers arguing why x is better than y. But, someone will eventually get around to helping, if you have some patience. Personally I think you've been kind of rude to people who've taken the time to try to help. Maybe you're just a bit highly strung and took it the wrong way....

Whatever, glad you finally got the answers you wanted.

Crumb$ wrote:

djst wrote: ...more interested in profiting from other people than to share the love and knowledge. I guess that's a part of the MPC community DNA that I'll simply have to accept.

You know, this isn't the fastest moving forum. Sometimes it can take people a while to get an answer to something, and I've always imagined thats because people are somewhere else, making music or some ****, but not sitting at their computers arguing why x is better than y. But, someone will eventually get around to helping, if you have some patience. Personally I think you've been kind of rude to people who've taken the time to try to help. Maybe you're just a bit highly strung and took it the wrong way....

Whatever, glad you finally got the answers you wanted.

If I've upset someone, I apologize. If anything, I may have been rude to that YouTuber who spread misinformation hoping that people would buy his online courses. And yes, when the first two replies immediately suggested I buy other books/services to find the answer to my question, that made me generalize about the MPC community. There are some really smart and helpful people in this community, as we've seen above, and I appreciate the help.

Not sure how that makes me highly strung in your opinion, I'd say some people reacting to this thread may be. Regardless, I considered this thread settled a couple of weeks ago and I'm not really sure why it was suddenly resurrected.

Hope you'll have a great Sunday.

We all read text different ways… at different times… and the context today may not be the same in the past or the future.

Anyways… for whatever it is worth it did not bother me

Just for chuckles I tried how loud I could make the signal at the master outs. Using all 4 program inserts, all 4 submix inserts, and all 4 master inserts, totalling 12 x AIR Channel Strip, each with gain +24 and all 4 bands of EQ with maximum gain. We're at +500 dB, and there's still 1000 dB headroom before the theoretical limit of 32 bit floating point numbers.



BUT it's possible to get clipping somewhere inside plugins and instruments. For example, the (very nice) Odyssey plugin seems to distort/clip easily, particularly if the VCA GAIN feature is used. Crank up VCA GAIN all the way and start adding filter resonance:



Those sound like digital overs to me, happening somewhere inside the Odyssey plugin. I don't know what that really is, are they using integers inside the plugin or is there some forced clipping, or is that clipping based on modeling Odyssey hardware, and so corresponds to what would actually happen. To me it sounds like digital clipping though.

MPC-Tutor wrote:Well it's very easy to clip at the individual pad/keygroup level - hot samples, multiple layers, filters, FX etc, so gain staging the individual pad/keygroups within the program is definitely part of the normal workflow. Gain stage individual pads/keygroups, then adjust the overall program level, and so on, etc etc.

In that video he's actually adjusting the wrong 'level' control - he's only adjusting the 'layer 1' level which is bad practice when the aim is to reduce the total pad level as that method fails once you have multiple layers in use. He really should have used the pad mixer to explain this better, but that's just nitpicking, I know what he's trying to get at.

I was messing around trying to figure out gain staging and it seems the pad mixer adjusts level post insert, so if you don’t want your signal going into an effect as hot you have to reduce the layer level, which is pre insert. It makes a lot of the fx sound better and less distorted.

https://www.youtube.com/watch?v=zI7e7A065mI