Digital editing software and added "hardness" to the sound (LONG POST)

[David R. Modny]

I'm in the process of beginning a hopefully long project of transferring some vinyl, analog tapes, etc... over to CD-R just for fun and convenience, and would like to make some observations(and field some comments) on digital editing and what I perceive to be it's negative effect on sound (and I'm NOT just talking truncation and level changing distortions here). I'm hoping Steve might even weigh in, as someone who obviously doesn't use these tools. I also have it fresh in my memory as to the poster from a few days ago who had, what he considered, good luck with transfers he had made - in maintaining the original analog source sonic properties. Here's my most recent scenario:

I began with a notion of transferring over some old satellite-feed based concert recordings I had (from the 80's and 90's) of some Who/Townshend shows preserved on Beta Hi-Fi stereo. This would be my warm-up to some vinyl transfers I wanted to do. All of the recordings were very good to excellent, sonically, while maybe being a touch dynamically over-compressed for their intended radio/television medium.

I'm using a Pentium II/233MHZ MMX PC equipped with what I considered to be the best budget sound card I had heard for the money, the Turtle Beach Santa Cruz. Certainly not pro level, but a giant killer nonetheless. I chose to monitor with my trusty (actually a newer replacement) Fostex T-20 headphones. I also chose to tap off of the soundcards lineout jack so I could get some idea as to how the source sounded looped through the card while monitoring during recording, and then compare it to the level-matched digitally recorded output out of the card's same lineout.

My first impression is that the resulting WAV file DID sound different than the analog original. Drier, with not as much lushness. Yet, not really offensive at this point. Still, a slight degradation though. Where the sound really got harder, was when I made a simple edit in Cool Edit Pro, chopping off some extraneous beginning and end material before I planned on going into the CDWAV program to split the file and sequence it. Thus, the Cool Edit action was a simple delete command; no level changing, truncating, resampling, etc... which I had always been told was the no-no that led to DAW "hardness" of the sound. I should also mention that everything was done straight 16-bit, with NO 32 bit conversion/dither process.

This simple action in processing the file really hardened, to my ears anyway, the sound. I'm perplexed, as I would think that simply "cutting" something with a digital editor would be safe. But obviously in processing and saving the file...some sort of re-calculation or degradation must be occuring. The midrange/vocals now had that sort of "grit" that didn't exist in the analog original. Cymbals, also sounded like someone letting the air out of a tire, with all of the warmth and body of thier original sound degraded. What I'm saying is that the warm "goosh and smoosh" had been compromised.

All of this has me thinking that in the future, I may have to take a "Hoffmanesque" approach when transferring and do all of my moves "on the fly" during the recording - making my recorded wav file the FINISHED product (perhaps just using burning software for the "splitting" and sequencing of a long concert-type file).

I guess what I'm asking from the pros and hobbyists here: What methods are you using to get acceptable results? I've heard home-made transfers from other posters here, where I didn't get a sense of the hardness I experienced (boy does that sentence look funny on the screen!). Or maybe their processed files HAVE added this, but not having the analog originals at my disposal - I have no actual mode of comparison.

I would appreciate any comments, observations, or success stories.

[Grant]

I'm using the same card.

Now, what kind of cable do you use to run your tapes into the card? I use MIT Terminator 2 interconnect

You should not record to 16-bit and process. Record with 32-bit float. You will have to dither and noise shape back to 16-bit for the CD. I do this and get great results virtually indistinguishable from the source.

Are you recording as close to 0db without clipping?

Did you check the proper playback configurations on the soundcard's mixer? Don't forget that the card's output has 20-bit output. Do you have any of those surround sound or 3-D options turned on in the car'd configurations?

Software cannot impart any sonic imprint.

The soundcard is not perfect but it does rank just below the semi-pro/pro cards, which is pretty damn good for a consumer card.

[David R. Modny]

Yes, all of the configuration settings on the card are correct. The only differences between what we're doing is the 32 bit float/dither down to 16 bit process that you're using, and the straight 16 bit process, while briefly using CEP, that I am.

Again, my *biggest* problem isn't the initial WAV. It's the post-CEP edited one (which if levels aren't changing... shouldn't really be affected by keeping it 16 bit for a simple, single edit/delete...one would think).



Originally posted by Grant

Software cannot impart any sonic imprint.


Come again? Any software that's manipulating or recalculating the signal sure as hell can. If I were to truncate bits in Cool Edit Pro I'm sure you would agree that the software, due to the added quantization noise, would most certainly leave a sonic imprint. An MP3 encoder, by it's very nature, certainly leaves a sonic imprint!

[Grant]

Again, it's all mathmatical, David. IUnless there is a bug in the software, and I assure you there is NO bug in Cool Edit, the data processed by it will sound the same as if you used Sound forge5.0 or Samplitude or any other 32-bit floating point software. If you are working with 16-bit the dither and/or truncation method may be different.

Of course mp3 will sound different because it is data reduction. Yes, it compresses the data but also discards what isn't usable for it's purposes.

Now, what you initially describe as happening to your sound is classic of what happens when recording at a sampling rate of 32kHz, 16-bit.

How about if you give us a detailed rundown of what you are doing and what gear/software you are using, right down to burn speed, and let's see if we can isolate this problem. Because, I don't have this problem, or, at least I quit having it when I stopped running my analog source through a converter before it went to the soundcard.

If you prefer, email me.

Why do you not post this on the Cool Edit forum? They made the software.

[David R. Modny]

Originally posted by Grant
Again, it's all mathmatical, David. IUnless there is a bug in the software, and I assure you there is NO bug in Cool Edit, the data processed by it will sound the same as if you used Sound forge5.0 or Samplitude or any other 32-bit floating point software. If you are working with 16-bit the dither and/or truncation method may be different.

This is simply not true Grant. It's a function of the software used that determines *how* the signal is processed and what calculations are used to achieved the end result.


Of course mp3 will sound different because it is data reduction. Yes, it compresses the data but also discards what isn't usable for it's purposes.


That's obvious. I was just perplexed by your statement that software doesn't impart a sonic signature. The functions (see above) of ANY software and how they mathematically achieve it will certainly determine the outcome sound. Not just with obvious data compression...but ANY calculations that are going on within in stuff like CEP. All things being equal, for example, CEP and Soudforge might use different algorithms to perform an identicle 3db boost at a particular frequency.


Now, what you initially describe as happening to your sound is classic of what happens when recording at a sampling rate of 32kHz, 16-bit. [/B]

Give me more credit than that. I'm certainly aware that I'm recording at 16/44. :)

[Grant]

Originally posted by David R. Modny


Give me more credit than that. I'm certainly aware that I'm recording at 16/44. :)

I wasn't trying to say you didn't kinow better.

But again, it is true. Software does not leave a sonic imprint. That's more audiophile mythology. If you download another program and record you 16-bit file it will sound the same. There is something else you are doing or using differently that causes us to get different results. My money says it's the 16-bit depth you used.

Interesting thing is, even when I make a simple edit on a 16-bit file I don't get the dramatic degradation you do. Now, I can hear a small change, but to the point you described? No.

[lukpac]

Originally posted by Grant
But again, it is true. Software does not leave a sonic imprint.

Well, in this case, it obviously does. David made a recording and listened to it. Then he made an edit of that recording. It sounded different. The edit was done via software. As such, the software changed the sound.

Simple, eh?

David, I will need to do some editing in ProTools and see if you can detect a difference.

Also, try this: take your edited and unedited files and place them side by side in CEP (or whatever). Then invert one set of tracks. If nothing is different, you should be left with *silence*. If CEP did indeed change the sound, you'll be able to hear something. Obviously, everything has to be lined up perfectly.

[Sckott]

FWIW, I've never had CEP change the sound via editing. That is, from just editing. If I did filtering or EQ, well that's on purpose.

[Grant]

Originally posted by lukpac


Well, in this case, it obviously does. David made a recording and listened to it. Then he made an edit of that recording. It sounded different. The edit was done via software. As such, the software changed the sound.

Simple, eh?

David, I will need to do some editing in ProTools and see if you can detect a difference.

Also, try this: take your edited and unedited files and place them side by side in CEP (or whatever). Then invert one set of tracks. If nothing is different, you should be left with *silence*. If CEP did indeed change the sound, you'll be able to hear something. Obviously, everything has to be lined up perfectly.

Luke, the software did not change the sound, as he believes. The action of editing changed the sound, and he did it on a 16-bit file.

Now, if editing did change the file, and obviously it does, the files will not nullify.

He also did not tell us exactly what he did to produce such a radical degredation of sound that he initially describes. I mean, did he record at a low level? Did ha have the "dither all transforms" options checked in Cool Edit's options? What dither depth is he using? Luke, I know you don't have much if any experience with Cool Edit, but all of these things, and tons more are all configurable in the program and will affect the sound of whatever he does within it.

[Grant]

How about this?

[lukpac]

Originally posted by Grant
Luke, the software did not change the sound, as he believes. The action of editing changed the sound, and he did it on a 16-bit file.

The software was doing the editing. It changed the sound.

Editing via a razor and block doesn't change the sound. The fact is that CEP *did*.

What's so hard to understand?

[Sckott]

How about the fact that it shouldn't. I have to side with Grant in the fact that it *shouldn't*.

If it did for you, then something's up. I've never personally had CEP change the sound. Sorry to hear that it did in your experience, but maybe something WAS turned on like Grant points out? Donno.

[Grant]

Originally posted by lukpac


The software was doing the editing. It changed the sound.

Editing via a razor and block doesn't change the sound. The fact is that CEP *did*.

What's so hard to understand?
No. What''s so hard for YOU to understand? If you edit that file in any editor it will change something. Whether the action is audiable depends on many factors.

And just a few weeks ago YOU were arguing with SH and everyone else on the DCC forum that an edit made on a DAW doesn't change the sound. Make up your mind.

[lukpac]

I didn't say any editor. I said ProTools on the Mac. CEP is not ProTools.

David has said the sounded changed from simple editing. I'm not sure *why* that's the case, but it seems to be the case. It shouldn't matter if he was using 16 bit, 32 bit, whatever.

A razor and block is an "editor". That doesn't change the sound.

Where is the confusion here? David did an edit in CEP. It changed the sound.

[David R. Modny]

[QUOTE]Originally posted by Grant



But again, it is true. Software does not leave a sonic imprint. That's more audiophile mythology. If you download another program and record you 16-bit file it will sound the same. There is something else you are doing or using differently that causes us to get different results.


It isn't true and it has nothing to do with audiophile mythology. It's a simple fact of algorithms being proprietary to any given software. I guess I'll have to give you an example to try and prove it to you...yet I doubt I'll convince you.

I made some simple 30 second recordings today of the same analog source using both Dart Pro and CDWAV. I made several passes using both software programs to make sure the result would repeat itself. NOTHING was touched during recording; input levels on the sound card weren't changed; nothing was checked in the software that would alter the signal (e.g. normalization), bit depths were the same...EVERYTHING was equal.

On all passes, the resulting wav files had a 1/10th of a percent peak level difference displayed when analyzed in Steinberg...a real MEASURABLE DIFFERENCE. Thus, something in the algorithms of the programs used to record, made their own sonic imprint on (i.e. altered) the input signal. Simple as that. So you see, the "different programs recording a 16 bit file and sounding the same" argumment that you're trying to cram down everyone's throat is merely wishful thinking.


Interesting thing is, even when I make a simple edit on a 16-bit file I don't get the dramatic degradation you do. Now, I can hear a small change, but to the point you described? No.


The original point of this whole discussion is that these are subjective hearing differences that I heard. You may not hear them at all. When people speak of a "hardness" to the sound in mastering on this board it's going to be subjective. Steve, himself, as stated that the reason he doesn't use DAW's in mastering is because he perceives an added hardness to the sound. You don't have to agree with this, but the fact that people do hear differences simply make some of us want to investigate further. It might be even be measurable if we knew exactly what we were looking for and had the advnaced scientific tools to do it (e.g. things such as jitter).

BTW, to answer your other thread. CEP only allows the option of dithering during an amplitude change-dependant edit, or during sample conversion (eg. 32 to 16). As I mentioned earlier, my file was recorded and the edit was made at 16 bit. So dither had nothing to do with this particular case.

Frankly, this whole debate has gone on further than I wanted it to with one person. I can see we're locking horns at this point. I'm done.

[David R. Modny]

Originally posted by Grant

in reference to Luke:



And just a few weeks ago YOU were arguing with SH and everyone else on the DCC forum that an edit made on a DAW doesn't change the sound. Make up your mind.


Considering how just a year or two ago you were telling people on the BSN Board that not only do the sonic signatures of various brands of blank CD-R media sound different - but the "sound" of the recordable drives themselves as well, I'd say the above comment...and your seemingly current transformation to "numbers is numbers"....is the pot calling the kettle black!