Why modern recordings sound crap

[mosfet]

From the original article Over the Limit

But what is going on with Vapor Trails? The numbers quickly report the truth we were hearing with our ears. The average volume is a whopping 4.5 dB louder than Counterparts. But where Counterparts was able to achieve a 4.5 dB volume boost with almost no increase in audible distortion, in the Vapor Trails sample almost every peak is clipped, many of them severely so. And, as we shall see, the limiting is so severe that the songs have no punch, and just slam out of the speakers like a loud blast of white noise.

What you are looking at is a serious square wave. Note in the top swing of the clipped wave how the high-frequency harmonics have had all of their peaks shaved off? This is why clipping sounds harsh. Now note how in the lower portion of that wave the bottom is just perfectly squared off? That's a pulse of white noise. Now note how all of the peaks in this signal are all hitting the top of the scale? This is why the CD has no punch. A powerful blast of the kick drum should be louder than the rest of the signal and should have a sharp transient on the front of the wave. In the case of this song, however, it just produces a quick blast of distortion and a dull thudding sensation.

If you find something else, either objectively or subjectively, then you should email the author and let him know AK. editor@prorec.com

In any case if you or anyone else doesn’t hear any distortion in the first instance then all is fine. To my ears however a clipped waveform has always sounded like distortion because that’s what it is.

 

[jon]

I wonder to what extent clipping can be effectively used to add 'punch' to a CD? I'm not talking about having it on a whole CD - but can think of one or two albums which have (occasional, very loud) volume peaks that are heavily distorted. Could that be clipping?

Jon

 

[Guest]

mosfet wrote:

In any case if you or anyone else doesn’t hear any distortion in the first instance then all is fine. To my ears however a clipped waveform has always sounded like distortion because that’s what it is.

Mosfet, having studied the wave, I can see where the clipping is, analysis of the waveform in adobe audition suggest up to 0.725% (of whole waveform)clipping, in right channel and 0.526% this is from the track Ghost rider, are you telling me you can here distortion at this level? if it is between half and one percent of total 3 min+ track are these distortions (clips) of such a small duratiuon that it enables them measurable, but inaudible?

I don't deny the chap has looked at the waves, but like other measurements (speaker cables) the measurement doesn't seem to correspond to what you hear, yes the album is darker, yes it is more congested thatn other Rush albums, but reading this chaps stuff then listening back I can't hear ANY distortion, his assesment of it smacks of exaggeration, perhaps it is there, perhaps each clip is too short in duration to notice, PM addy and I'll send you rip of album I have, and see if your monitors can pick it up, I'll also send you audition rips of same track, BTW I'm using wav, not MP3

 

[Guest]

Been having a think about this, I have heard bad discs, I own some poor recordings, in the past I have chosen some discs over others, not for musical content, but because I liked the sound, or didn't.

As my system has resolved, discs which previosuly were unlistenable, due to compression or whatever, now have come to life, as my system has got more revealing so more of the congestion has been lost and music is coherent, so whilst recordings will allways be good or bad to varying degrees, some blame has to lie at door of kit.

I now choose the music I want to hear, from my whole collection, not just those discs that sound good.

 

[mosfet]

To take each of your points in turn AK.

The extent of digital clipping on Vapor Trails

You say the track you have looked at, Ghost Rider, has less than 1% peak amplitude clipping on both channels for the duration of the track. RR says, referring to instances of peak amplitude clipping, there are “thousands of ruined kick drum transients from the Vapor Trails CDâ€.

On balance your measurement is better because it is more specific. However respective findings do appear to be somewhat contrary.

Is this type of clipping audible anyway?

Considered objectively, a clipped waveform at 0dB full scale is the loudest possible sound a CD can record, so certainly loud enough to hear. The only question is one of duration; does the clipping last long enough to be heard?

Considered subjectively, if you can’t hear anything, you can’t hear anything and there’s no arguing with that. There are also reports of “crackling†noises from various sources in review of this recording.

Exaggeration?

Well, that's possible isn’t it. And something audiophiles are particularly prone to also! I’ll get myself a copy of Vapor Trails on loan when I’m next in the library and have a listen myself.

 

[Guest]

Do you know what track he recorded from vapor trails.

and yes when I said inaudible I was talking about duration, not level, the figures I got where from adobe audition.

I also agree with exagerration, |I have allways maintained that any changes I have heard with wires have been subtle.

As for Rush

I can copy you raw and processed through audition samples of all the rush tracks you want. for evaluation obviously.

PS i'm genuinely interested am not arguing for srguments sake, and given I have software he has tested I'm willing to try and replicate his findings, but need a little help.

 

[mosfet]

Do you know what track he recorded from vapor trails.

The only references in Rip Rowan’s article are to “samples†from the different Rush recordings. I have found another site with screen dumps using Sound Forge that may go some way to answering the question of duration and audibility of the clipping:

Some people, including myself, have copies of the Vapor Trails CDs that have many occurrences of clipping that cause the recording to crackle on playback. I've ripped the audio off of my CD to analyze it with Sound Forge. There are many visible signs of clipping in the waveforms that correspond to crackling.

In the above, taken from the track Secret Touch, you can see that ten divisions along the top axis represents 5 milliseconds or 0.005 seconds (00:04:47.570 to 00:04:47.575 and so on) therefore each smaller division represents 0.5 milliseconds. 0.5ms is an incredibly short period of time!

The question is are the instances where the waveform is successively clipped long enough in duration to be heard? Looking at the circled area it can be estimated in this example the duration is around 1.5ms.

This one’s a bit outside my field of understanding however as a pointer try downloading and listening to this MP3. (72KB).

What you should hear are four pairs of clicks. The first pair occurs concurrently (and so is heard as one event), the second pair of clicks separated by 1ms, the third by 3ms and the fourth pair of clicks by 5ms. I can hear the difference at 1,3 and 5ms and so can say, yes, I can detect an audio event lasting for only 1 millisecond. Does this lend weight to idea of being able to hear clipping that lasts for 1.5ms? Possibly.

What are the duration figures you’re getting in Audition?

I can copy you raw and processed through audition samples of all the rush tracks you want. for evaluation obviously.

Nope, this one’s for you to figure out AK now you’ve finally embraced objectivism!

When I posted the link to Rip’s article I was just presenting this as evidence and example of the increasing over use of audio compression (and hence reduced dynamic range and digital clipping) in modern recordings. As it happened the article was based on Rush recordings. Of course I was expecting you to post in some way (knowing you’re a big Rush fan) but I wasn’t expecting you to challenge the evidence objectively. All power to you for doing so.

As far as I’m concerned the increasing use audio compression is a real enough trend (now I’m beginning to sound like a subjectivist!). The original article I linked to written by Bob Katz has enough authority for me, given the man’s reputation as a recording engineer, for this.

***EDIT***

I can copy you raw and processed through audition samples of all the rush tracks you want. for evaluation obviously.

Then again.. if you’ve got ten minutes to spare AK? try this.

Rip a WAV file of Secret Touch.

From this cut a sample beginning at 04:44 and ending 04:51 (or thereabouts). Apply a one second fade-in and fade-out at the beginning and end of the sample (this makes repeated listening easier on the ear). This sample will cover what is shown above. I’ll have a listen - and look at the sample in an editor.

Email as an attachment to pigsinzen[at]fsmail.net

I’d be interested to hear.

Jon wrote:

I wonder to what extent clipping can be effectively used to add 'punch' to a CD?

Clipping cannot be used to add anything Jon. Apart from distortion. It’s a subtractive process that quite literally lops the tops of waveforms (see example above). When clipped waveforms are heard they generally sound harsh and unpleasant. I say “generally†because this isn’t always true. The sound of a valve guitar amp being pushed into soft clipping is anything but unpleasant.

For a recording to have ‘punch’ it needs dynamic light and shade, so the punchy bits really do punch out by sounding louder. This doesn’t happen if a recording has little dynamic range as a result of being compressed.

 

[Guest]

thanks for the comprehensive reply Mosfet, will gave a wee play and see how I get on, will take some screenshots if I feel it's helpfull, intend to test the idea with others tommorow, I'll put un touched and decompressed on same disc, to see how different results are... a copy could be made available to you, if your interested?

EDIT: Just seen your edit, will email clip as suggested, I'll do wav/32bit wave, and decompressed wave of the clip you sugest, mmmm, could I get used to this objectivism?

 

[meninblack]

analoguekid wrote:

mmmm, could I get used to this objectivism?

Free your inner Geek

 

[mosfet]

AK,

I’ve had the chance to look at a number of modern ‘loud’ recordings now and while I can see plenty of clipped transients (like those shown above for Secret Touch) I haven’t seen any I would think would translate audibly to ‘crackling’ noises. This is also borne out subjectively as I don’t hear anything that might be described as crackling when listening.

I think the best that can be said is the presence of digital clipping will make a recording sound less ‘clean’ generally rather than those clipped waveforms being identifiably audible as a noise of some sort.

So, the suggested crackling noises on Vapor Trails? I don’t know! :Not Sure:

 

[jon]

Clipping cannot be used to add anything Jon. Apart from distortion. It’s a subtractive process that quite literally lops the tops of waveforms (see example above). When clipped waveforms are heard they generally sound harsh and unpleasant. I say “generally†because this isn’t always true. The sound of a valve guitar amp being pushed into soft clipping is anything but unpleasant. For a recording to have ‘punch’ it needs dynamic light and shade, so the punchy bits really do punch out by sounding louder. This doesn’t happen if a recording has little dynamic range as a result of being compressed.

thanks, the visual example makes it clearer. I can think of a couple of albums (Alec Empire, say) where I'd suspect thatdistortion might be added by deliberately pushing it into clipping? anyway,just been listening to some PJ Harvey/John Parish and remembering how well the quiet/loud thing works

 

[Leonard Smalls]

When I was a Beeb bod we set zero level in a digital recording as -18dB, i.e PPM4 was -18. This meant that max level(PPM6) was 8dB above this - i.e -10dB.

This gave plenty of room for stupidly high peaks, and also meant that we didn't have to use Brick Wall Limiters (i.e cut off any signal above the set level). Therefore we had properly dynamic recordings with headroom.

But we noticed there was a tendency for outside (oh yes, it was them and us!) recordings to come in with zero level set at -10, or even -8dB giving no room at all for manouevre. Amazing how many we had to send back for going over 0dB - in other words they clipped... And those very same outside companies berated us for not using the full dynamic range!

C***s.

 

[jon]

Just been playing with foobar, and noticed it has settings (in Preferences, under Playback) to apply replaygain to moderate playback levels and also a tick box to reduce levels in order to stop clipping. Seems to make Oasis albums sound much easier on the ear

 

[mosfet]

and also a tick box to reduce levels in order to stop clipping

Not quite sure what you mean Jon, but to correct recorded clipping you need a program that re-draws (if you like) the tops of waveforms that are missing. Some audio editors can do this. The tick box in foobar is perhaps to reduce recorded levels to prevent clipping in the soundcard. In any case if you’ve found something that makes an Oasis recording sound better then that’s a result!

I’d suggest you download a copy of Audacity and have a play. Rip a few WAV files and you’ll see the difference in recorded dynamic range between different recordings. By zooming in you’ll also see digital clipping (if indeed the recording is clipped).

Also try applying a low-pass filter set to 15 or 16kHz. Open the WAV file then Edit > Select > All to select the complete track. Then Effect > Low Pass Filter and enter 15000 in the Cutoff box (to filter out frequencies above 15kHz).

 

[jon]

yeah, does seem to bring down recorded levels... Somehow does seem to make it sound less harsh, but don't think it was loud enough to send my soundcard into clipping

Apparently the processing it does to lower the levels can also lose some detail - so maybe that's the 'improvement' I heard... but yeah, if it works I'm not complaining

Will try to have a play with audacity, too...

 

[Guest]

been a bit busy mossy, expect email in next couple of days

 

[SSM]

Just got the Carrie Underwood CD. Nice material but the recording qualifies as 'crap'.

Geez sound levels are so loud and relentless, my ears are cringing. It makes Underwood seem like she's hollering all the time. She even thanks the engineerson the liner notes for "making this album sound amazing!" Women:roll:

I can't listen to this disc all-through again, can only pick one or two tunes for quick blasts.

 

[mosfet]

Remember the Blur vs. Oasis chart war? Here’s the Blur vs. Oasis loudness war!

Probably the best example to see this volume rise in action is to pull out a few albums from the 80’s and early 90’s and compare the music with that released in the last few years. I did not believe this loudness issue until I loaded up a few songs in a sound editor and seen it for myself:

Most music from the late 90’s to the current charts has a fairly maximised graph. This 5 second clip to the left is just an example from ‘Hello’ by ‘Oasis’. With older songs, the rhythm is clearly visible in the graph. This 5 second clip to the left is an example from ‘Country House’ by ‘Blur'. While listening, the beat clearly 'jumps out' unlike in the above song.

 

[mosfet]

Just noticed this recent letter from the directors of Griffin Audio USA on the online pro-audio site Mix.

Further self-criticism from those actually involved within the recording industry; “We are professionals. If we don't care, who will?â€. So long as objections like this are raised within the recording industry there remains hope standards will improve.

Personally I blame it all on bling and R n’ B!

The technology available today offers all opportunities to create some of the best-sounding recordings ever made. There are higher bit depths and higher resolution in the entire recording and mastering chain. Why do all the “new†records sound like they are recorded on 12-bit equipment and run through a fuzz box? I'd say we can thank the “Loudness Wars.†There are artists and record company folks telling recording and mastering engineers to “Make it louder!†Guys, there's a great device invented decades ago known as the volume knob that will easily make things louder without causing the THD of my final product to exceed 10 percent! When the whispered voice is just as loud as the electric guitar through a full “stack,†then what is louder?

The NARAS Producers Wing is involved with CES in an attempt to ensure that the consumer equipment manufacturers maintain a high standard of quality for audio playback. Why bother if the CDs we produce today will play back with 10-percent (or higher) THD? It's not the playback equipment's fault; it is designed to play back musical waveforms. Not the “flat-topped†distorted signals on today's CDs

link

 

[ClassikFan]

Kate Bush album sounds pretty well recorded to me.

I heard her discussing mp3 etc on the radio and how keen she was to keep her music HiFi.