Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[nickharambee]

i understand that reply gain does not affect the original music file, but i am unsure as to whether it affects the quality of sound when burning to a cd.  does it merely change the volume level, or does it affect the quality of the sound as well?

thanks, nick.

[jleerigby]

I believe that the answer is no unless you turn the DSP setting on when burning.  

This was explained by xen-uno in the other thread you started.  If you weren't sure about his answer it would have been better to bump the original thread rather than start a new one - then we would all have the background of the previous discussion and may be better able to help you.

http://www.musicex.com/cgi-bin/yabb/YaBB.cgi?board=MediaCenter;action=display;num=1059670382

[nickharambee]

thanks for your message jlee, and my apologies for starting a new thread.  my point is that i am looking for a process where i can burn tracks to cd without loss of quality, but having equalised the volume levels using a program like replay gain.  as i understand it, in order to have replay gain applied to tracks when burning i need to select the DSP/crossfade box prior to burning.  but you say that if i do this, i wil lose sound quality.  is this right?

thanks, nick.

[jleerigby]

I see your problem.  You can turn off all the DSP settings except RG before the burn.  That will ensure that equalisation etc does not affect quality - but how do you stop it applying crossfade.  If it were me I would try changing the crossfade setting to Standard before the burn.

Have you done some tests and comparisons?

[nickharambee]

i have done this exact process and it works fine as far as crossfading goes.  i have tried to do some sound quality comparisons and it is difficult to tell the difference using my ears.  mine are not the best of ears though, and as i use a lot of my my music for djing on large sound systems, sound quality is important.  i did feel that the level of volume reduction was too great though.  is there a better way of comparing sound quality, e.g. with a program like cool edit pro?

thanks, nick.

[jleerigby]

Can't help you there I'm afraid Nick.  My djing is limited to my front room!  Time for the experts to chip in...

[JohnT]

There's no quality loss doing replay gain other than a reduction of volume on the louder tracks in the playlist. The softest tracks are reproduced exactly and then the louder tracks have their volume reduced proportionally to create an even volume across the playlist. No distortion is introduced but of course anytime you reduce the volume you lose a few bits of accuracy of reproduction, the same as if you manually turned down the volume on the player.

[nickharambee]

thanks johnt,

so does this mean that if, having turned up the volume on my stereo, to compensate for the reduction in volume created using replay gain when burning, that the quality of sound would be the same as listening to the same track burnt without replay gain (at the same volume as the replay gain'd track).  i hope this makes sense.  what i am in effect asking is, is the reduction in quality only due to the reduction in volume, or it is an irreversible reduction in sound quality that cannot be rectifoed by turning up the volume on my stereo!

thanks, nick.

[nickharambee]

hiss on cds?  not yet heard this, so probably not turning my stereo up loud enough.  ok, so the difference in quality is not in the sound itself, but rather extra background noise due to lower input signal?  is this right?

nick.

[Bartabedian]

In your circumstance for the uses you describe, don't use RG...it' just not going to do what you need. If you are truly concerned with quality because of the large system throw but still want to "level" your files, you'd be better off with proper HQ digital editing tools, like Sound Forge (which has the Wave Hammer plug) or the Waves L1 or L2 plugs, digital brick wall limiters that will slam the music onto CD right at 0 level without any loss of quality or distortion, but be warned these tools aint cheap, the best stuff never is...

[Matt]

hiss on cds?  not yet heard this, so probably not turning my stereo up loud enough.  ok, so the difference in quality is not in the sound itself, but rather extra background noise due to lower input signal?  is this right?

Yes.  Digital volume reduction throws away information.  Normally that information is just random noise bits, so you'd never know the difference.  However, it's still by definition a lossy transform -- not lossless.

In your circumstance for the uses you describe, don't use RG...it' just not going to do what you need. If you are truly concerned with quality because of the large system throw but still want to "level" your files, you'd be better off with proper HQ digital editing tools, like Sound Forge (which has the Wave Hammer plug) or the Waves L1 or L2 plugs, digital brick wall limiters that will slam the music onto CD right at 0 level without any loss of quality or distortion, but be warned these tools aint cheap, the best stuff never is...

There is no higher quality way to digitally adjust volume than what MC does. (MC uses 32-bit precision)  All digital volume adjustment is a lossy DSP, so "sound quality" is sort of the wrong lingo here.  (i.e. does bass boost help or hurt the 'sound quality'? -- you can't really answer that)

[gkerber]

Yes.  Digital volume reduction throws away information.  Normally that information is just random noise bits, so you'd never know the difference.  However, it's still by definition a lossy transform -- not lossless.


There is no higher quality way to digitally adjust volume than what MC does. (MC uses 32-bit precision)  All digital volume adjustment is a lossy DSP, so "sound quality" is sort of the wrong lingo here.  (i.e. does bass boost help or hurt the 'sound quality'? -- you can't really answer that)

Agreed, sound quality is hard to answer, in the ears of the beholder.  It does affect sound accuracy, and therefore is a "distortion" of the original, which may or may not be pleasing to the listener.

Very interesting, I never knew that digital volume changes were not totally accurate, but it makes sense now that I think about it..  That implies that it's better to do volume changes in the analog world when possible.   I know I'd never hear the difference either way.

[Matt]

Very interesting, I never knew that digital volume changes were not totally accurate, but it makes sense now that I think about it..  That implies that it's better to do volume changes in the analog world when possible.   I know I'd never hear the difference either way

Well, it depends what's happening.  Some sound cards support decimal points so volume changes are perfectly lossless.

However, a CD is 16-bit integers (no decimal).  So in that case it's just like shrinking an image before emailing it -- there's no going back to the original.  

However it's a little like shrinking from a 60 megapixel image to a 30 megapixel image.  The output is still better than your brain can absorb.

[Bartabedian]

There is no higher quality way to digitally adjust volume than what MC does. (MC uses 32-bit precision)

Well, I certainly can't agree with that statement at all, and I've done high-end audio for over 20 years now, digitally for over ten. True, if all MC does is adjust a volume level, then it's "lossless" function is purely about volume attenuation. Now in Nick's case, he may be getting a CD with steady volume levels, but his CD is more than likely far from the 0 level it can handle, and noise will abound at the higher volumes he's using in a DJ setting because the floor is now much closer to the median than the ceiling.

When I master a CD, my volume levels "pump" between -6 and 0, rarely falling below that range (depending on the material, of course). Is what I'm doing lossy? No, but is it changing the file? Yes. Is it what all masters go through? yes again. When making a CD from various sources, would I do the same brick-wall limiting? Using as little as possible, yes, not RG, which is causing a lossy problem all it's own.

When I get ready to master my next project, should I use the $40 MC's RG ability, or my over $3K worth of software?

I'm not knocking MC Matt, I love the thing, but that's a hugely bold sweeping statement, and I'm sorry, but it's just not true.

[Matt]

Warren, you're talking about DSPs that alter the dynamic range of the music.  Replay gain intentionally doesn't alter the dynamics -- leaving that to the original studio engineer.  At that task, MC does as good of a job as possible.

[Bartabedian]

Ok Matt, let's clarify for ourselves and everyone else what the real issue here is:

Nick wants a CDDA at the highest quality possible with all songs broadcasting at a relative equal volume. RG creates a median, and follows that path as the disc is burned. He felt as though the disc was low volume, and I decscribed why.

Following my suggested path is certainly more tedious and cumbersome than MC and RG, but the quality of the CDDA is greatly enhanced. Now DSP or not, I'm only making the suggestion as the most quality path to a comp CDDA of songs from various sources. If burning a CDDA from a single source, I'd go without RP or my suggestion and do it straight up, no quality questions, DSP arguments or conversion concerns to deal with.

But basically, in a CDDA burning setup, I am suggesting staying clear of RG completely, as it is not intended for this use. It may not be altering the signal, but it's also not creating a HQ environment for a CDDA.

[xen-uno]

> It may not be altering the signal, but it's also not creating a HQ environment for a CDDA

I don't follow your logic here at all (and Matt has got it right). RG does alter the signal (but not negatively) on burning to some set level (89 dB). The sole purpose here is to bring ALL of the selected tracks up to this level (or drop them down if they're too hot), so that on average, they have the same volume level. That is the definition of Radio Gain RG (and is MC's default AFAIK). If you want relative differences between tracks kept, then you would apply Album Gain. I'm not sure what you would consider a better environment for a CDDA. Nick's disks may be too low of a volume, and in cases like that you would add some extra gain, but then you increase the chances of clipping, and it may or may not be possible in MC (I use WavGain, where it is possible). He could also just turn the volume up a bit...and he should only have to do it once. I'm at work so I can't put 100% thought into this, but RG is perfect for what he wants to do.

10^-27

[gkerber]

> It may not be altering the signal, but it's also not creating a HQ environment for a CDDA

I don't follow your logic here at all (and Matt has got it right). RG does alter the signal (but not negatively) on burning to some set level (89 dB). The sole purpose here is to bring ALL of the selected tracks up to this level (or drop them down if they're too hot), so that on average, they have the same volume level. That is the definition of Radio Gain RG (and is MC's default AFAIK). If you want relative differences between tracks kept, then you would apply Album Gain. I'm not sure what you would consider a better environment for a CDDA. Nick's disks may be too low of a volume, and in cases like that you would add some extra gain, but then you increase the chances of clipping, and it may or may not be possible in MC (I use WavGain, where it is possible). He could also just turn the volume up a bit...and he should only have to do it once. I'm at work so I can't put 100% thought into this, but RG is perfect for what he wants to do.10^-27

All Replay Gain is doing is turning the volume up or down a little, digitally.  Which I'll guess is no different than me manually adjusting the volume on my B&K preamp (which I think does volume adjustments digitally).

So really, the only real issue with RG for burning is - is the listener willing to accept different volume adjustments for tracks from the same source album?  (all tracks of an albums should always have the same adjustment value to perserve the relative difference between tracks, as the artist intended).

MC really needs Album Gain - or at least allow us to manually set the adjustment value in Properties.

[Bartabedian]

First of all, where does 89 db come into play, or does this transfer to -11db? An audio CD ranges up to 0, beyond that is just plain ugliness. Secondly I'm talking about digital brick-wall limiters, no clipping occurs and if used correctly, no adjust to the signal is made at all beyond creating a "higher" decibel range to live in. It brings the signal up to digital 0, and "holds" it there in a range you create. It is far and away the absolute best environment for a CDDA, period. It's also a bit time-consuming and expensive, but "you get what you pay for" means a whole lot here. RG just doesn't compare. Sure, you can adjust the value up a bit, but you're still under the control of a headroom limitation, and you can only "guess" where that is.

And I personally have my own issues with RG's gain structure. I have very easily been able to generate considerable noise from using it's most basic controls. It does a decent job with a standard playback mix of songs that I don't want getting out of hand at any time, but only with the adjustment setting at 0 or below. Any plus values creates noise immediately. And whenever I use it @ 0, my volume drops at least 10db, causing me to find level increases eslewhere. My software mixer is maxed out, and at the amp, that's released resistance, which is noise. RG is far from the messiah of volume leveling.

[xen-uno]

What your talking about sounds a whole lot like "re-mastering" or just plain Normalization...especially when you say "It brings the signal up to digital 0, and holds it there in a range you create". Further, when you take the peak and hold at or near 0 db (the ceiling...or (+/-(2^16)/2), and then say that your raising the average volume (other than rescaling), then you are introducing dynamic compression. If your method doesn't raise the noise floor (like RG & Normalization does), then I would say there is some noise filtering going on...which has it's own problems, too (it is a lossy process).

10^-27

[Bartabedian]

Not what I'm saying at all. Mastering involves perimetric EQ, basic compression, multi-band compression and digital limiting. Normalization isn't that far from what RG is, finding a median to adjust around, but it does alter the actual file unlike RG, and I can't stand normalization. Am I re-mastering, no. Am I doing a very light CDDA mastering, yes, yes, yes. Indeed, a necessary evil when wanting to burn a quality CD of varying sources.

In this process, you set a threshold at which point the limiter will go for 0. It's not compression, it's limiting, a different process. It will only raise the noise floor that already exists in the original file, not add another one, and it gets it farther away from the -90db noise floor of a CDDA. A filter can be used for noise shaping, dither and bit rate quantization.

Look, it ain't for the faint of heart, and I've warned nick that it can be more time consuming and costly than using MC and RG. I just wanted him to understand he had other options should he want to explore further.

As far as lossless vs lossy, it's still (as we can all tell by this thread) a controversial subject. I have many colleagues who believe that a CD rip to WAV is a compromise of the original file, and there's no tangible proof either way. In the title of this thread, the answer is yes, but with a mighty asterisk next it, that's all I'm saying.

[xen-uno]

> you set a threshold at which point the limiter will go for 0

The limiter you refer to is analog hardware, correct? In that case your just keeping the peaks below a certain point, which is identical to it's digital equivalent...normalizing (except analog signals are naturally dithered). The limiter in digital software is the RG value or Normalization percentage that you set. FYI: WavGain will let you apply dither while gaining the file just like your setup.

>it gets it farther away from the -90db noise floor of a CDDA

Remember that we are starting with the same source...a file ripped off a CD in some digital format. Any movement of the noise floor toward the peak values or away from them would be considered compression for the former...noise reduction for the latter. Once again these are not lossless processes. I don't think any track off any CD you can rip is completely noiseless (that is, it is quieter than CDDA spec of -90 dB...meaning that it doesn't exist)

Nick asked "does it affect the quality of the sound". If by quality he meant...
"Does it change the integer values contained in the file?" => yes
"Does it decrease the dynamic range? Increase it?" => no & no
"Does it raise the noise floor?" => yes...but the average volume is increased by the same amount
"Is the RG'd file a faithful reproduction of the original?" => other than a scale change, it is as close as possible for digitally ripped sources

'Peace...we should probably just agree to disagree on this topic. It's made me think deep, though.

10^-27

[Bartabedian]

Agree to disagree, no problem, it's what makes life interesting...

The L2 is, ya know maybe you should read this yourself...

http://waves.com/htmls/prods/hardware/l2_ultramaximizer.htm

and

http://waves.com/htmls/prods/masters.htm

...basically, music you've heard in the last few years, it went through this onto a glass master...

[gkerber]

Agree to disagree, no problem, it's what makes life interesting...

The L2 is, ya know maybe you should read this yourself...

http://waves.com/htmls/prods/hardware/l2_ultramaximizer.htm

That "machine" totally explains some of the crap recordings I've heard recently.  Maybe there should be a law againt it...

I see and hear the trend, but until now I didn't fully realize how bad it was.

[Bartabedian]

That "machine" totally explains some of the crap recordings I've heard recently.  Maybe there should be a law againt it...

I see and hear the trend, but until now I didn't fully realize how bad it was.

Don't blame the tool, that "machine" as you put it is invaluable to CD mastering as we know it, and as unfortunate as a medium that it is, 16bit-44.1k CD is the mass market distribution media.

Blame the people using it and the reasons they over-use it, and know that the day that will come when "CD quality" isn't necessarily considered a good thing.

[gkerber]

Don't blame the tool, that "machine" as you put it is invaluable to CD mastering as we know it, and as unfortunate as a medium that it is, 16bit-44.1k CD is the mass market distribution media.

Blame the people using it and the reasons they over-use it, and know that the day that will come when "CD quality" isn't necessarily considered a good thing.

The machine is advertized to make recordings loud, that's what it's for, and described in their literature.  The CD can sound d-a-m-n good, bad sounding CD's are the result of bad  production and mastering (and of course the "content").

It is clearly over used..

CD Quality only really implies something about playback accuracy.  The old Garbage In, Garbage Out rules applies.  The CD will faithfully reproduce (within it's limitations) the input, whether that input is "good" or "bad" quality.

The first Bad Company album is a good example, it sounds terrible, not the CD medium's fault, the input recording was  bad.

Buddy Guy / Sweet Tea is an example of over compression and loss of dynamic range.  That album sounds terrible.

[Bartabedian]

The machine is advertized to make recordings loud, that's what it's for, and described in their literature.  The CD can sound d-a-m-n good, bad sounding CD's are the result of bad  production and mastering (and of course the "content").

It is clearly over used..

CD Quality only really implies something about playback accuracy.  The old Garbage In, Garbage Out rules applies.  The CD will faithfully reproduce (within it's limitations) the input, whether that input is "good" or "bad" quality.

The first Bad Company album is a good example, it sounds terrible, not the CD medium's fault, the input recording was  bad.

Buddy Guy / Sweet Tea is an example of over compression and loss of dynamic range.  That album sounds terrible.

Well, a lot can go wrong in a recording, from the sessions, to the dubs, to the mix, past the label execs a few times after redoing sessions, dubs and mixes and oh, a few thousand cups of coffee. Maybe the Producer is good for the project, maybe not? Does the engineer really know what he's doing or is he too stoned to know the difference? Don't those vocals seem awful loud to anyone? Oh, the label exec says turn up the vocals more? And compression at this stage is vital and crucial to the outcome, and in danger of being over used immediately.

Then the ominous mastering, which I feel has become a forgotten art in itself. Mastering used to involve a whole lot of mathematics and diamond cutters, a true science. Now it's about something else completely.

My point about the CD and it's relation to this "machine" is that it can take any source and get it to CD in the most HQ possible chain. All the things you're dissappointed with happened before this process, it just makes it loud on the CD, the "sound" doesn't change at all. That was compression from mic to tape (A or D), then there's A tape compression, tube compression in the mix and most likely the master.

I've said this before in another thread, there is a canyon of difference between what I hear in the studio and what the public hears on a CD. Since I rarley listen to the stuff I've worked on, it was years before I actually heard it. The first time I experienced it though I was truly alarmed. Now I don't listen to CD's much anymore at all.

[nickharambee]

Well thanks everyone for your contributions to this debate, it has certainly moved on my thinking with regard to the best volume levelling process.  Perhaps inevitably though it has raised as many questions as it has answered, and so I thought that I would post this message, a summary of my understanding of the different software and the pros and cons.   May someone wants to let me know if I’ve got the basics right, or answer some of the questions raised………

Sound quality issues/processing of original file:

In both programs (L2 or Wave Hammer plug-ins and replay gain) the original file is unaffected, but obviously it is processed by these programs when burning to cd.  The best description I have found of what L2 (or Wave Hammer, I will use L2 only from now on, but could of course be referring to both plug-ins) is actually doing is from the user guide:

“the maximum level of a digital signal is governed by the highest peak in the file.  simple normalisation finds the highest peak, then raises the entire signal so that this peak is at the maximum value.  However, many of these peaks may be of a very short duration and can usually be reduced in level by several dbs with minimal audible side effects”

so L2 is actually making changes not only to the overall signal level, but also to the range of the signal, limiting its highest levels, so that the db level of the overall signal can be increased.  To put things simply then it is effectively changing the character of the signal as well as its amplitude.  Now of course the aim is to do this without changing the perceived sound or quality of the original source.  The danger is that music processed through L2 will sound ‘flat’ because the range of the signal has been limited or reduced.  I presume that the skill here is to apply settings that boost the signal without changing the sound.  Personally I have not found my processed tracks to sound flat, but I have heard this complaint levelled against Wave Hammer on forums.

Now in replay gain, because there is no function to limit the very high levels, it needs to find a lower overall gain level in order to avoid clipping.  The disadvantages of this process are relatively higher noise levels as the signal is (nearly always) being pushed down relative to the noise level.  Is there also a second disadvantage here, because I will not be getting the best out of my amplifier by putting in a lower signal (i.e. the quality of the output of amps is better at lower settings on the volume dial), or am I talking about the same process here?  The advantages of replay gain is that it only alters the amplitude of the signal, and not the character.

Time issues:

Obviously given that I need to make individual settings for each track in L2, Replay Gain is much more time friendly.  Furthermore, because all of my music is compressed as .ape files, I need to convert to .wav format before applying L2, which is more time.  

Storage space:

Because L2 does not change the actual file, I cannot re-compress to .ape and keep the L2 information (is this right?).  So this has implications for storage capacity – all of the tracks that I want to use for burning compilations will need to be stored as .wav files, so that they can be burned accurately using CD Architect.  I am aware that there is a lossless compressed format for Sonic Foundry software (.pca), but this is not supported in MC9, which I want to use as my media player.

Fine adjustments to sound levels:

From my experience of using both replay gain and the plug-ins, it is rare that these processes leave all of the tracks in a given playlist exactly the right volume levels.  Maybe one or two of the tracks still sound louder.  At present I am not sure how to make fine adjustments to tracks after processing (and even if it is possible to do this).  Obviously I cannot reduce gain level in L2, so is there another plug-in/process for making small adjustments (to either increase or decrease the signal)?

DJing on my laptop:

OK, here is another potential spanner in the works.  Up until now I have been using CD mixers to DJ.  With this in mind I am tempted to use L2 just for my DJ tracks where a high signal level/punch is important, and replay gain for stuff I’ll listen to at home (because of the time saving factor).  However, now I have some rather tasty DJ software, and I am going to experiment using my laptop for DJing.  Now, if I want to have my DJ tracks at a high and relatively equal volume level, both L2 and replay gain will not help me as the information from these processes will not travel with the file to DJ software programs.  What I need is a process similar to L2 that will permanently change the original file.  Can L2 or replay gain do this?  I am aware that MP3 gain does this, not sure about Wave Gain.  What do you suggest?  I don’t mind changing the files for DJ tracks permanently because I can make copies of files first.  Of course I can change gain levels within DJ software, but it does make life easy to have tracks about the same level from the start.

Well, I think that’s about it, a bit of a missive I know………

Nick.

[xen-uno]

Nick > Now in replay gain, because there is no function to limit the very high levels....

Not true. RG is based on peak values. The very high levels you mention ARE the peak values. The RG setting becomes the limiter.

> The disadvantages of this process are relatively higher noise levels as the signal is (nearly always) being pushed down relative to the noise level

There is no relative dB change as long as clipping doesn't occur. There is two types of noise your likely to hear...1) Inherent (and encoded) in the source material or 2) Noise from the amp & pre-amp stages in your equipment. #1 has a linear relationship with RG (again...assuming no clipping). #2 in non-linear to and beyond RG's control (well, sort of...a strong recording minimizes that by either being RG'd or by being simply a strong recording). The equipment noise floor could be considered a constant (it produces X db's of hiss when the attenuator is set at Y).

10^-27