Low- and mid-level productions with crazy levels

[Machina]

Hi guys,

I'm sure many of you have noticed this: When surfing on the internet - maybe on myspace or some other music sites - you often hear low- and mid-class recordings, either from smaller bands or studios, that mostly lack many of the things like clarity, definition, good leveling, great sounds and so on, but are loud as hell. They're distorting and clipping and everything you can imagine. Soundwise not even close to a "pro" mix/master, but sometimes it's loudness that many artists confound with punch. I mean it's crazy! Those guys try to be louder than all the top guys in the biz?! Who needs to be louder than e.g. the new Coldplay?! It's insane...
I really don't want to start another loudness war discussion here and personally I really dig the sound of some cds that some may call squashed (the new Jimmy Eat World or Green Day or Paramore or whatever) BUT I'd like to know how you guys deal with this situation and what you think about it? I mean sometimes it's not enough to say that we don't care, because we know we can do a better sounding mix or master. In the end the clients pay the money and our bills...

Cheers,

Chris

[Edward_Vinatea]

Originally Posted by Machina

BUT I'd like to know how you guys deal with this situation and what you think about it? I mean sometimes it's not enough to say that we don't care, because we know we can do a better sounding mix or master. In the end the clients pay the money and our bills...

Machina - If you were the artist I know I can talk you out of squashing the life out of your mixes. If you were the artist's engineer I would work it out with you so that the client gets what he is looking for. Same goes if you were the manager or the producer ---

Regards,

[Machina]

Hi Edward,

thanks for your reply! I know there are a thousand goods arguments why one shouldn't squash it to death, but even if they're convinced in the first place I've seen bands coming back to the ME asking for more loudness anyways.
But as I said it's not about artists wanting to be as loud as their favourite cd from XY (which is still kind of understandable somehow ) but more about those low- and mid-level productions that sound crappy but have insanly hot levels. I've heard so many blowing off intro guitars and when the band comes in you instantly go

Cheers,

Chris

[MASSIVE Master]

I protest, then do what they want while controlling the damage as well as I can.

You can't talk most of them out of it - Especially the producer/label reps. The artists are easy - They tend to not want to sound like crap. The labels are busy with a pi$$ing contest of some sort.

[Ben F]

There was an old saying where I used to work in regards to crappy mixes: "If you can't make it sound good make it loud!"

[ninohernes]

Originally Posted by MASSIVE Master

I protest, then do what they want while controlling the damage as well as I can.

Couldn't have said it better myself.

[Machina]

@ Ben F:
That's excactly what I think when I'm listening to these kind of recordings. If it's not good then at least make it loud...

[Edward_Vinatea]

Originally Posted by MASSIVE Master

You can't talk most of them out of it - Especially the producer/label reps. The artists are easy - They tend to not want to sound like crap. The labels are busy with a pi$$ing contest of some sort.

Hi Massive Master - FWIW the concept of making things louder is over a decade old. At the risk of sounding controversial or whatever let me say that I once talked to not one but two multi-platinum producers {Jerry Kasenetz and Jeff Katz} out of squashing the life of a mix and also talk them out of producing a song that was poorly written. Then talked them in to delivering the vocal tracks of RAM JAM's biggest hit - remixed it and then later I was fired . Not because the record was too soft or too loud but because it was "too crazy" for these 2 bubblegum music producers to handle. My point is you can talk even to giants if you are humble enough.

Regards,

[24-96 Mastering]

In my experience you can often reason with those directly involved in the production process, because it is their work, their labour of love that will be damaged.
With A&Rs though, in my experience, if they want the level, they want the level. And it's hard to successfully advise against it, because they usually are aware of the drawbacks and knowingly accept them.
The best way of making your opinion count in such cases where the label is the driving force behind high loudness, is - if possible, i.e. if there's direct contact - to advise the producer or artist, those who have crafted the recording, of your recommendations and, if they agree, suggest to make their influence count.

[Edward_Vinatea]

Originally Posted by 24-96 Mastering

With A&Rs it can be hard though.

They were the 'A&R' department as well. But yes - times have changed. ---

Regards,

[MASSIVE Master]

My point is you can talk even to giants if you are humble enough.

I'd never suggest otherwise. But you can only lead a horse to water.

[Aisle 6]

Hi Machina.
It would be interesting to know if these loud lower budget productions are even mastered. Stay with me!
I have become more aware of low budget mixes that are hellishly loud simply because they have been mixed on a native system, ie 32bit floating point. This arithmetic has way way more headroom than say a HD system running 48 bit fixed point arithmetic. I am running HD myself and have had clients bring in their projects for me to mix, which they have recorded at home, only to find them spanking the clip leds and distorting. I have had to turn all tracks in a multi track session down by up to 18db or more across all tracks to prevent overs. The client always says, it was not distorting on my system, and they are right. The inaccuracy of 0db in floating point math is actually beneficial to low end users who do not even know how to manage gain structure. Ironically it gives them the ability to create louder mixes than is possible on a fixed point system. Then add to that the mastering process and viola! ignorance creates something that is ridiculously loud. Crazy stuff.
Cheers,
Scott.

[Edward_Vinatea]

Originally Posted by MASSIVE Master

I'd never suggest otherwise. But you can only lead a horse to water.

...but you can't make them drink! -

BTW, care to participate in our little shoot-out? your personal views IMO would be one of the most respected and anticipated on this forum.

Best regards Massive Mastering,

[Machina]

Hi guys,

thx for all your answeres!

@Scott:

Oh, that sounds interesting! I guess most of the low-/mid-level masters/mixes I hear we're done on a native Logic or Cubase sytem. I always asked myself how the mix could be so bad (technically), but how it was possible for those guys to create such a crazy loud master without TOTALLY distorting it like hell. I mean it DOES kinda distort, but at a loudness level that's even hard to accomplish for professional and really good MEs I know and work with. Don't get me wrong: I know they can do MUCH better sounding masters. I was just wondering about loudness.
I'm not really that into arithmetic stuff, but 32 bit floating vs 48bit fixed point could be the explanation for that phenomenon. I couldn't imagine those guys sitting there and wondering how much clipping is still not audible and what instances of compression, limiting and distorting would produce less artifacts and so on
Thx for that suggestion!

Anybody else noticed this?

Cheers,

Chris

[macc]

Originally Posted by Aisle 6

Hi Machina.
It would be interesting to know if these loud lower budget productions are even mastered. Stay with me!
I have become more aware of low budget mixes that are hellishly loud simply because they have been mixed on a native system, ie 32bit floating point. This arithmetic has way way more headroom than say a HD system running 48 bit fixed point arithmetic. I am running HD myself and have had clients bring in their projects for me to mix, which they have recorded at home, only to find them spanking the clip leds and distorting. I have had to turn all tracks in a multi track session down by up to 18db or more across all tracks to prevent overs. The client always says, it was not distorting on my system, and they are right. The inaccuracy of 0db in floating point math is actually beneficial to low end users who do not even know how to manage gain structure. Ironically it gives them the ability to create louder mixes than is possible on a fixed point system. Then add to that the mastering process and viola! ignorance creates something that is ridiculously loud. Crazy stuff.
Cheers,
Scott.

I absolutely 100% concur and identify with this! I could not agree more, right down to the 18dB thing. When I get mixes to work on (from 2 bands in particular) the very first thing I do is Ctrl-A and bring the gain down by 18dB. All because of gain structure.

[Adam Dempsey]

Originally Posted by MASSIVE Master

I'd never suggest otherwise. But you can only lead a horse to water.

But if you lead a horse to water when it's thirsty...



At least I'd like to think so. (Having just spent a day helping run workshops with 350 school students).

[Edward_Vinatea]

Originally Posted by Aisle 6

I have become more aware of low budget mixes that are hellishly loud simply because they have been mixed on a native system, ie 32bit floating point.

Incorrect - please read on.

This arithmetic has way way more headroom than say a HD system running 48 bit fixed point arithmetic.

The benefits from processing with 48 fixed as opposed to 32 floating are IMO humanly impossible to be perceived. It all goes down to 24 bit anyway.

I am running HD myself and have had clients bring in their projects for me to mix, which they have recorded at home, only to find them spanking the clip leds and distorting. I have had to turn all tracks in a multi track session down by up to 18db or more across all tracks to prevent overs.

This is just giving you more headroom but it doesn't resolve sound quality issues created at capturing samples or tracking . If you a got a bass guitar track distorting at different sections of the mix - lowering it on your DAW will not restore the sound quality. That distortion will be re-scaled 18 dB lower than the original source. It may even sound worse {theoretically speaking}.

The client always says, it was not distorting on my system, and they are right.

Whose system your client(s) or yours? If it's the client's end then is most likely due to poor detail studio monitors.

The inaccuracy of 0db in floating point math is actually beneficial to low end users who do not even know how to manage gain structure. Ironically it gives them the ability to create louder mixes than is possible on a fixed point system.

Internal processing (better math, more steps, etc} does not yield more loudness - i.e. a 1kHz tone at -0dBFS on a 24 bit file is still the same value if converted to 16 bit at -0dBFS. The same happens for 100Hz tone at -0dBFS on 24 bit is still -0dBFS on 16. However, you will perceive the 1kHz tone as louder than the 100 Hz tone.

I hope I understood your comments and got this right otherwise please someone correct me if I am wrong.

Best regards,

[UnderTow]

Originally Posted by Aisle 6

Hi Machina.
It would be interesting to know if these loud lower budget productions are even mastered. Stay with me!
I have become more aware of low budget mixes that are hellishly loud simply because they have been mixed on a native system, ie 32bit floating point. This arithmetic has way way more headroom than say a HD system running 48 bit fixed point arithmetic. I am running HD myself and have had clients bring in their projects for me to mix, which they have recorded at home, only to find them spanking the clip leds and distorting. I have had to turn all tracks in a multi track session down by up to 18db or more across all tracks to prevent overs. The client always says, it was not distorting on my system, and they are right. The inaccuracy of 0db in floating point math is actually beneficial to low end users who do not even know how to manage gain structure. Ironically it gives them the ability to create louder mixes than is possible on a fixed point system. Then add to that the mastering process and viola! ignorance creates something that is ridiculously loud. Crazy stuff.
Cheers,
Scott.

This doesn't make any sense to me. At the end of the line, even floating point DAWs need to converter the signal to fixed point to go out the D/A converters. If the master bus is not clipping in the floating point DAW, a properly exported mix won't clip in a fixed point DAW either. (And vice versa). The only difference is that individual tracks can go over 0 dB FS without clipping. The master output is still limited to 0 dB FS or it will clip.

Maybe you are referring to a very specific situation: ProTools LE sessions loaded into ProTools HD. Yes in this case the tracks that were above 0 dB FS in LE will now clip in HD. But you must realise that if the master output of the LE session was not clipping, there was no no "loudness" advantage whatsoever. The only advantage was that the individual tracks did not clip despite the bad gain staging. (Assuming there were not plugins in the session that clip at 0 dB FS).

Alistair

[Aisle 6]

Originally Posted by Edward_Vinatea

Incorrect - please read on.

The benefits from processing with 48 fixed as opposed to 32 floating are IMO humanly impossible to be perceived. It all goes down to 24 bit anyway.

Althoug the HD system is constantly truncating from 48 bit to 24 bit through every insert. Arghhh!

This is just giving you more headroom but it doesn't resolve sound quality issues created at capturing samples or tracking . If you a got a bass guitar track distorting at different sections of the mix - lowering it on your DAW will not restore the sound quality. That distortion will be re-scaled 18 dB lower than the original source. It may even sound worse {theoretically speaking}.
Correct. I am not talking about the recorded sound, only headroom.

Whose system your client(s) or yours? If it's the client's end then is most likely due to poor detail studio monitors.
My monitors are ATC SCM20's and I have also experimented on Questeds and Duntechs. All revealed the same result and all are very highly regarded monitors.

Internal processing (better math, more steps, etc} does not yield more loudness - i.e. a 1kHz tone at -0dBFS on a 24 bit file is still the same value if converted to 16 bit at -0dBFS. The same happens for 100Hz tone at -0dBFS on 24 bit is still -0dBFS on 16. However, you will perceive the 1kHz tone as louder than the 100 Hz tone.
Correct again, but not grasping the issue. Would you say that 48bit is better math than 32 bit floating??? It certainly retains more post decimal point remainders before rounding or truncation. But it simply exhibits less headroom. Push that headroom out ridiculously and check the result on each system, it is possible to get a much louder mix on LE than HD. I own HD and it sucks. Although there is the argument that correct gain structure is correct gain structure and that is the truth. But an LE can almost ignore that fact and not suffer the consequences.

I hope I understood your comments and got this right otherwise please someone correct me if I am wrong.

Best regards,

You are misunderstanding my statement and I can assure you that I have extensively tested this theory and it is supported by the arithmetic.
I am not talking about the D/A stage. I am talking about the internal summing. You can drive the crap out of the internal floating point buss and just turn the master fader down. No distortion. Now take that same mix session into a HD rig and the overs through the internal busses are absolutely audible. This is as you suggest purely to do with the increased headroom offered by floating point arithmetic.
Cheers,
Scott.

[Aisle 6]

Originally Posted by UnderTow

This doesn't make any sense to me. At the end of the line, even floating point DAWs need to converter the signal to fixed point to go out the D/A converters. If the master bus is not clipping in the floating point DAW, a properly exported mix won't clip in a fixed point DAW either. (And vice versa). The only difference is that individual tracks can go over 0 dB FS without clipping. The master output is still limited to 0 dB FS or it will clip.

Correct. I was not talking about the D/A stage.

Maybe you are referring to a very specific situation: ProTools LE sessions loaded into ProTools HD. Yes in this case the tracks that were above 0 dB FS in LE will now clip in HD. But you must realise that if the master output of the LE session was not clipping, there was no no "loudness" advantage whatsoever. The only advantage was that the individual tracks did not clip despite the bad gain staging. (Assuming there were not plugins in the session that clip at 0 dB FS).

Alistair

You are correct. It is all about clipping the internal buss. But this allows for crazy loud mixes through an LE system. Myself, another Producer that I work with regularly and my mastering engineer have all tested this and proven it. It is purely that floating point arithmetic has more headroom internally. On an LE system, you can spank the crap out of the channels (and the inserts, but don't even get me stared on the undithered 24bit inserts in Pro Tools HD) right through to the mater fader. Go ahead drive them up as far as you like! Then just turn the master fader down to prevent overs through the D/A. Viola! Hammered, loud mix. Put that same mix on a HD system and cover your ears.
Cheers,
Scott.

[Edward_Vinatea]

Ok - sorry- let's start from the top again. I might learn something in here too.

Originally Posted by Aisle 6

Hi Machina.
This arithmetic has way way more headroom than say a HD system running 48 bit fixed point arithmetic.

So you are saying that the internal operational computations increased the dynamics of the signal and its plugin effects? So, you achieved more dynamic range?

I am running HD myself and have had clients bring in their projects for me to mix, which they have recorded at home, only to find them spanking the clip leds and distorting.

You mean that you got session files from another software or maybe it was the same as yours? Isn't it easier to mix with BWF files? Can you clarify a bit more?

I have had to turn all tracks in a multi track session down by up to 18db or more across all tracks to prevent overs. The client always says, it was not distorting on my system, and they are right.

Are we talking *real* distortion (like really harsh crap} or just clipping?

Originally Posted by Aisle 6

I am talking about the internal summing. You can drive the crap out of the internal floating point buss and just turn the master fader down. No distortion. Now take that same mix session into a HD rig and the overs through the internal busses are absolutely audible.

Can I ask what HD rig you have? Is it a DAW?

This is as you suggest purely to do with the increased headroom offered by floating point arithmetic.

I did? Headroom relates to dynamics no? How does it change the word of your audio file?

Please let me know.

Thanks --

[Aisle 6]

Originally Posted by Edward_Vinatea

Ok - sorry- let's start from the top again. I might learn something in here too.


So you are saying that the internal operational computations increased the dynamics of the signal and its plugin effects? So, you achieved more dynamic range?

I would not say that it is more dynamic range as such, but it is to do with the way that floating point algorithms see 0db.

You mean that you got session files from another software or maybe it was the same as yours? Isn't it easier to mix with BWF files? Can you clarify a bit more?

All my experiments were done in Pro Tools. Sessions brought in from Pro tools LE and launched on my Pro Tools HD rig. This means that all the mixing including levels and panning are the same. The only difference is the arithmetic. LE uses 32 bit floating point and HD uses 48 bit fixed point.

Are we talking *real* distortion (like really harsh crap} or just clipping?

We are talking audible digital distortion, not just xmas tree lights.


Can I ask what HD rig you have? Is it a DAW?
I did? Headroom relates to dynamics no? How does it change the word of your audio file?

As I said earlier it is Pro Tools HD (DAW). On the surface there should be no changes in the data when the session is imported between LE & HD. The only difference is the mathematics.

Please let me know.

Thanks --

Does this make more sense now??? It is quite frustrating for me considering that I pay considerably more for my HD system than an LE user.
Cheers,
Scott.

PS. I am sorry that I was unable to break your quotations up between my responses.
I do not know how that works. : P

[Edward_Vinatea]

Originally Posted by Aisle 6

Does this make more sense now??? It is quite frustrating for me considering that I pay considerably more for my HD system than an LE user.

Hi I am sorry but I am not sure what you are asking me now. I did leave a few questions for me to get a mental picture of what happened.

EDIT: I didn't see your answers before - sorry! and I can only speak from experience so I don't have an answer for that distortion. Maybe UnderTow is right. The thing is in any request from a client to have me mix something, If they are not PT TDM sessions, I tell them to get BWF files.

Regards,

[tonmeister]

Originally Posted by Aisle 6

32bit floating point has way way more headroom than say a HD system running 48 bit fixed point arithmetic.

erm, I don't think this statement is correct, HD systems have 288dB of dynamic range, stacks of headroom and full resolution "footroom" also. The 48 bit mixer uses a 56bit accumulator.
this from the 201 coursebook: "This process provides enough headroom to sum 128 full-code tracks with all the faders at +12 without causing any internal clipping on the "input" side of the bus. (you would have to trim the master fader however to avoid clipping the DAC) It also allows channel faders to be pulled down to nearly -90dB before the channel stops contributing a full 24 bits to the mix."

I haven't got the stats on 32bit float, but I can't see how there would be any clipping problem loading your LE mixes into HD

[UnderTow]

Originally Posted by Aisle 6

You are correct. It is all about clipping the internal buss. But this allows for crazy loud mixes through an LE system. Myself, another Producer that I work with regularly and my mastering engineer have all tested this and proven it. It is purely that floating point arithmetic has more headroom internally. On an LE system, you can spank the crap out of the channels (and the inserts, but don't even get me stared on the undithered 24bit inserts in Pro Tools HD) right through to the mater fader. Go ahead drive them up as far as you like! Then just turn the master fader down to prevent overs through the D/A. Viola! Hammered, loud mix. Put that same mix on a HD system and cover your ears.
Cheers,
Scott.

But this won't make the mix any louder. If you pull down the master fader so that it isn't clipping, there is no clipping going on and the end result is the same as if nothing had passed 0 dB FS. I still don't see what you are driving at. You can do clipping in any DAW.

Alistair

[UnderTow]

Originally Posted by tonmeister

erm, I don't think this statement is correct, HD systems have 288dB of dynamic range, stacks of headroom and full resolution "footroom" also.

32 bit float gives you 1536 dB of dynamic range.

I haven't got the stats on 32bit float, but I can't see how there would be any clipping problem loading your LE mixes into HD

If the individual channels and buses pass 0 db FS in the LE session, they will clip when the session is loaded in PT HD.

Alistair

[Aisle 6]

Originally Posted by tonmeister

erm, I don't think this statement is correct, HD systems have 288dB of dynamic range, stacks of headroom and full resolution "footroom" also. The 48 bit mixer uses a 56bit accumulator.
this from the 201 coursebook: "This process provides enough headroom to sum 128 full-code tracks with all the faders at +12 without causing any internal clipping on the "input" side of the bus. (you would have to trim the master fader however to avoid clipping the DAC) It also allows channel faders to be pulled down to nearly -90dB before the channel stops contributing a full 24 bits to the mix."

I haven't got the stats on 32bit float, but I can't see how there would be any clipping problem loading your LE mixes into HD

Try it sometime...a Pro Tools LE mix with channels contributing to overs, (over 0dbfs) but with no audible distortion will, when loaded into a Pro Tools HD rig sound like nails scratching down a blackboard. Digital distortion!

What your course book probably does not tell you is that the HD system truncates the insert points to 24 bit (half resolution) with no dithering unless the plug in chosen has it built into it's own code. That little chestnut throws a few things out in the world of mathematics.

[Aisle 6]

Originally Posted by UnderTow

(Assuming there were not plugins in the session that clip at 0 dB FS).

Alistair

When was the last LE session you saw that did not have plug ins used in the final mixes???

[tonmeister]

Originally Posted by Aisle 6

Try it sometime...HD system truncates the insert points to 24 bit (half resolution) with no dithering

interesting, funny the wee details they neglect to mention... will have to test this.
One of the original selling points of HD is that it is virtually impossible to max out the bus, not so it seems.stike

[UnderTow]

Originally Posted by Aisle 6

When was the last LE session you saw that did not have plug ins used in the final mixes???

I didn't mean just any plugins. As I wrote, I specifically meant plugins that clip at 0 dB FS.

Alistair