Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: What's worse, artificially amplify, or artificially quieten? (Read 3839 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

What's worse, artificially amplify, or artificially quieten?

Hey everyone, hopefully a quick question. There is an audio file, from a 1995 CD, where the source material had some issues, and the engineers that worked on the CD, had to do the best they could to ensure that the song was listenable.

They had to splice together, from two different analogue tape sources, the same recording, so that they would have the first half and also the end section. Through something known as a "cosine fade" I have been able to do what they couldn't, and ensure that there is no noticeable "cut" or "jump" in the song, start to finish, just using the actual song from the CD itself. Only problem is, the two different versions of the recording they used also weren't of the exact same volume. So in audacity, with my finished work, I have two sections of sound that fade into each other, and for the most seamless transition, and to ensure an even sound throughout, I found that one part needs to be +3 db louder than the other.

This means that, upon final export, either one section has to be 3 db louder than the original, digital, source material (artificially amplified), or one section must be made -3db quieter than the source material. Which of these is "better" from an audio standpoint?

I know that quietening things can reduce overall dynamic range, or "technically" bit depth, but amplifying things gets rid of headroom, so if I had to choose which of these two is the least destructive.. well what would I do? I know in audio production, and even things like EQ, the rule is always "do not amplify artificially", but this is sorta different circumstances, right?

The part of the file that needs to be louder, is like 10% of the total listening time, and a 3db amplification wouldn't make anything clip, so i'm leaning towards that, but any opinions would be much appreciated. Also, any technical input, or commentary on my thoughts, would be much appreciated!

This is a file I hope to distribute to friends and acquaintances, and it's no doubt gonna eventually get circulated a bit amongst fans of the band, so i'm trying to do the best possible job I can.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #1 –
Quote from: Andrei_Vaughn on
This means that, upon final export, either one section has to be 3 db louder than the original, digital, source material (artificially amplified), or one section must be made -3db quieter than the source material. Which of these is "better" from an audio standpoint?

You don't have to choose, you can let the software do the job by peak-normalizing upon final export. Indeed, it need not be the "either-or" you propose - it could be that one section is made 1 dB louder and the other 2 dB quieter. That depends on where the new peak is, and how loud it is becoming.

Before final export ... disclaimer: I don't know all the Audacity quirks, but at least it can be made to use 32-bit floating-point, and then it should do the job for you.
Floating-point (at sufficiently many bits) means that "3 dB up and then 3 dB down" is a lossless operation (because "3 dB up" won't clip) - and so is "3 dB down and then 3 dB up" (because "3 dB down" won't fade anything to zero bits). So it should take care of everything as long as you stay in the floating-point domain until you finalize it. Then you can peak-normalize, and the software will know whether the peak is in this section or that section.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #2 –
Hi, thanks for the reply!

I will admit, I read your message with some confusion. I think, the fault could easily be on me, for either not doing the best job at explaining, or perhaps my concerns are over-the-top.

So, I have the "resultant" track at a point where the "perceived volume" across both sections, and across the transition, is the same (or, to a point where I am happy with it), due to this statement, from my understanding of what normalise tool does, it is not what I am looking for, right?

The bit about 32-bit processing (I didn't mean to do that pun...) is correct, however I actually have not applied any volume change operations to any of the audio that I am working with, I am just using the "gain" sliders that all tracks come with, and so that procedure, where your gain settings are actually applied, occurs during export.

My goal, is to have changed the original track volume as little as possible, in terms of the final exported product, I am not talking about any of the "processing" I would be doing within audacity before export.

So the question really is: what's worse for audio, being reduced in volume, or being amplified? If you take an extreme example, and say... amplify a song that's at normal levels, with -70db, then export. Well, you can bring it back 70db afterwards, but you've lost some actual audio quality. Well, in the other direction, is there any actual harm in amplifying artificially, as long as you don't get to the clipping point? I'm guessing there is some problem with doing this, I just probably don't know.

You may start to realise that this is a seriously small concern, if i've wasted your time, i'm sorry.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #3 –
Quote from: Andrei_Vaughn on
So the question really is: what's worse for audio, being reduced in volume, or being amplified?

The question doesn't make a lot of sense.  You have to pick some output level (usually close to peak but not close enough to clip).  Once you pick that output level, the adjustment you make is already determined.  Short of going out and recording a new album, you can't do anything but apply whatever the difference is between the current gain and the target gain. 

As for what is better, obviously an album recorded as high as possible but not clipping. Or at least high enough that you fit the noise floor into the dynamic range of the recording.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #4 –
If you need to splice two recordings of different amplitude to one, adjusting levels so they align, it would in principle be slightly better to adjust the levels of the quiet part up, rather than adjusting the loud part down. As long as this does not cause clipping.

If the level difference is 3dB, this does not matter unless you are using an 8-bit file format. Which you should not.

Generally, it makes sense to do editing at the highest possible precision, e.g. 32-bit float, then render the finished product to whatever format is needed for distribution.

-k

Re: What's worse, artificially amplify, or artificially quieten?

Reply #5 –
Hey, sorry for bamboozling everyone that read the thread, and for the 1 day delay in my reply... but nonetheless i think there is still some sort of contention here.

A better question is this: If you have a track that's mixed, mastered, etc, and issued onto a commercial CD. You (to simplify *everything*) have to either increase the whole track by +3db, or decrease it by -3db, then save that in place of the original, and that's it, that's what you're left with...

Which of these two options could do more harm to the integrity, and quality, of the original file?

I know either way, whatever happens, will be very negligible, but that's just it, I just wanted to know, which was the theoretically better, more logical of the two options.

I brought up the fact that I was doing editing in audacity, as a preface for why I am dealing with this question, nothing more. People talking about processing done in 32-bit floating point, im not sure that's helping, maybe it is, i just can't see it. Seems to me i provided more info than the question required! I just didn't want people asking "why do you have to do this in the first place?".

Re: What's worse, artificially amplify, or artificially quieten?

Reply #6 –

In Adobe Audition 3 I use amplitude statistics to get the average RMS power in the -15dB to -16dB range for consistent levels.

G²

Re: What's worse, artificially amplify, or artificially quieten?

Reply #7 –
Quote from: Andrei_Vaughn on
A better question is this: If you have a track that's mixed, mastered, etc, and issued onto a commercial CD. You (to simplify *everything*) have to either increase the whole track by +3db, or decrease it by -3db, then save that in place of the original, and that's it, that's what you're left with...

Which of these two options could do more harm to the integrity, and quality, of the original file?

That is the question I was addressing above.  To reiterate, the question doesn't make a lot of sense, but neither is better or worse, it just depends what level you want the final file to be at.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #9 –
If I were you I would simply amplify.
By reducing volume you will reduce dynamic range and there is no way to get it back.
By increasing volume you will increase noise but it will be masked by music anyway so no worries.
Make sure to leave some headroom, I personally leave 1 dB so it will not clip when using lossy formats.
If there are only peaks clipping (or close to clipping), I would use limiter, you can find it in Audacity.
Using limiter is better than lowering the whole volume just for some peaks imo.

gold plated toslink fan

Re: What's worse, artificially amplify, or artificially quieten?

Reply #10 –
From theoretical standpoint, if the track is not clipping already, by raising the volume you do not lose anything.
if the track's peak is already at 0, by raising the volume you introduce clipping/compression which squishes the high volume parts (disregarding floating point for this)

Same goes for the low volume parts to the other end.
If the track has some quiet parts which are near -127db (or whatever the digital limit is, not science guy here), by lowering volume on that part, you'd lose information - if two parts of the track are -125db and -126db respectively, and you lower the volume by -3db, essentially both parts become -127db.

Since the silent parts are essentially inaudible, you could argue that losing that information is less bad than corrupting the upper end.

As far as your actual situation goes, i would just raise the single track by 3db and be done with, if it doesn't clip.

Re: What's worse, artificially amplify, or artificially quieten?

Reply #11 –
Hey, everyone. Thanks for the continued discussion! I don't want to seem like I am alienating replies, although I did receive a few that genuinely seem to target the specifics of my query (even if it wasn't worded all that well!)

To Glenn: this issue is pertaining to just one song, consider that it in a vacuum, and I am not trying to get it to similar levels as another song. Now, the song has 2 parts, at different volumes, the stage of the process where I make them the same volume as each other, had already been worked out (I can estimate very accurately the volume diff. between the two tracks), my question was, which way do i take it? Do i go up with one part of the song, or down (in volume), with the other part? (to get them to be even).

To saratoga: Yes, I acknowledge that it's very fair to say that anything that should happen as a result of *either* choices will be largely negligible, if not entirely theoretical. I just wanted to use this case, as a way to establish a "best practice", should I come across something else similar in the future.

To Markuza97: This was kind of along the lines of what I was thinking about, I am kinda relieved that i am not the only person taking such things into consideration, (even if it is a little extreme to worry about such a small thing). Since peaks are not my issue, i won't use a limiter, but the rest of what you wrote seems bang on (in comparison to what I was thinking, anyway!).

To fulfo: You wrote "From theoretical standpoint, if the track is not clipping already, by raising the volume you do not lose anything.", thanks for writing this. I *know* for a fact that the section of the file that needs amplifying has ~12db+ of headroom, it is much harder for me to determine what the quietest/lowest volume sections of the song are, plus the song fades out, so i'm guessing it would get quite quiet, etc. So therefore amplifying in this case is probably the better choice? (btw, your nickname is russian for "rubbish")

Well, that does give me a somewhat conclusive basis on which I can end my search for an answer, at least without worrying later on, if i've made the right choice or not. I once again appreciate this forum for putting up with my less-than-intellectual questions.

Thanks again to everyone that spent (or wasted?) their time on this, It is much appreciated!

Re: What's worse, artificially amplify, or artificially quieten?

Reply #12 –
Quote from: Andrei_Vaughn on
(btw, your nickname is russian for "rubbish")

Sorry but this made me lol so hard, haha.

Anyway, 12 dB headroom, whoa, that seems to be little too much.
I am now afraid that your song wont be loud enough for most people with standard equipment (without amp).

Try to use loudness normalization effect in Audacity.
Try to normalize it to 18 LUFS (Old ReplayGain standard) or 14 LUFS (Modern "standard") and see results.
gold plated toslink fan

Re: What's worse, artificially amplify, or artificially quieten?

Reply #13 –
Sorry but this made me lol so hard, haha.

Anyway, 12 dB headroom, whoa, that seems to be little too much.
I am now afraid that your song wont be loud enough for most people with standard equipment (without amp).

Try to use loudness normalization effect in Audacity.
Try to normalize it to 18 LUFS (Old ReplayGain standard) or 14 LUFS (Modern "standard") and see results.

[/quote]
In the early '80s I bought Vivaldi's Four Seasons on Telarc which had 12dB of headroom
which I guess makes it effectively 14 bits.

G²
 

Re: What's worse, artificially amplify, or artificially quieten?

Reply #14 –
Hm, your responses have made me look at things again... I think I misspoke.

That 12 number was not determined using sound logic, instead by me looking at an equaliser. Headroom is how much volume you can amplify before clipping begins to occur? See, by that metric, the section of audio that needed amplifying had ~5db of headroom (?), and the louder part of the audio which I was trying to match it with had ~2db of headroom.

So i guess i learned something, well, I learned how to correctly determine something I should have already known!