Freesound Forums

Thats fine matey. But you can't go around making remarks like this:

thetruwu
Tweeter, I don't care if you're recording at 16/44 or 24/96, you try to get the levels within 5 dB of 0. This is what they teach in school, this is what they do in studios. Maybe field recording is different, but I really doubt it.

You should care whether you're recording at 16bit or 24bit.

You should not try to get the levels within in 5dB of 0.

And nobody cares what they teach at school.

If you look what I said several posts ago:

mike-t3

I don't want to be the guy to say moderation is everything but it pretty much applies to this thread.

.....
However you should record at a fairly high volume to keep the signal to noise ratio low.

Mike

can you give me one good reason why should shouldn't get your levels between -5 and 0?????

Of course you should record at 24 bit, but that is not really what we are talking about. We are talking about recording hot levels rather than using levels below -12db.

I'll agree not many people know what they teach you in school, but without basic audio fundamentals it's impossible to do quality work.

What bit rate you are recording at is the whole point.

It was drummed into peoples heads for a long time that you had to record audio as loudly as possible because people generally worked at 16bit. You had to keep the levels up to keep the fidelity.

Now that people generally work at 24bit or higher it is no longer necessary. In fact unless you are going for a particular sound there is no need to push your gear by recording at high levels.

What I'm saying is don't get stuck in these set rules and sayings. I must record at a high volume is defunct. Record at a level that is comfortable and don't stress if ur levels are 12 dbs below 0.

If ur recording in a dodgy room and there's a computer fan buzzing in the background... it doesn't matter how loud ur pre-amp gain is set too. There is still an annoying buzz in the background.

Mike

I'm going to be uncharacteristically RUDE for a moment:

Sounds added: 0

Flame on!

Whether it's a good idea or not, I'll try to step in again and summarize the extracted constructive points from both 'sides' (I also read referenced material - but don't want to reincarnate the other thread from 2005);

In technical comparison between 16 and 24 bit, one 16bit sample has 256 finer steps in 24bit. This has direct effect on recorded noise floor (from the equipment) - if it would require a smaller value than it would using in 16bit.
Using this noise floor as a starting point, I have more dynamic range to my loudest recorded peak available than I'd have in 16bit.

In that sense I understand the point that one is not enforced to max the preamp, since the noise is still 'far away' (to use inapropriate wording).

On the other hand, from the 'viewpoint' of the recorded signal, it is (objective) better to use most of the available resolution as possible (Perhaps as a comparison: I wouldn't drive a lower resolution than my TFT has natively) -- if this is desired or not is subjective. The ADC wont be 'stressed' more in either bitrates, it just has more (finer) values available to choose from in 24bit - thus sampling the signal better. And here something essential is in effect which hasn't been brought up here if I recall correctly: Quantization errors - which are always 'less' in higher bitrates (if the full range is used)

Even if I can't tell the difference in the highly sampled raw signal, any postprocessing would generate noticable artifacts sooner if it were stored in less bits than available.

Thus I conclude the following (objective) points:
- If I want to 'avoid' the noise floor, when using 24bit I already manage to achieve this having a peak of less than 0dB.
- If I want to record the signal with the best quality available, I always try to have the peak of (close to) 0dB, regardless of bitrate.

Personally (subjective) I want to have both - and the second does not interfere with the first.

For the ongoing argument, perhaps this lowers the tension, but I guess I might even been seen as 'victory' or totally misunderstood from either side I just wanted to show you, that I think to have understood both of you, what I took from it and that it made think and realize more about what happens in the background. (Which might even be more than I was able to express in this post)

You've summed it up wonderfully. And let me say that I am sorry the original thread was highjacked by other arguements.

lonemonk

You've summed it up wonderfully. And let me say that I am sorry the original thread was highjacked by other arguements.

firefreak

- If I want to record the signal with the best quality available, I always try to have the peak of (close to) 0dB, regardless of bitrate.

ok ok fine. Hijacky hijacky. But it is a forum for discussion. :wink:

All I'm saying is that this comment is fairly void at higher bit rates. I'm not saying ur ADCs is stressed at higher bit rates. I'm saying that ur hardware such as preamps wont perform as well at higher volumes. - This in the overall picture is perhaps un-important but so is worrying about getting with 12dbs of 0.

Which is indeed, my point.

If you are going to argue a point please use the correct terms Mike-T3. The correct term for what you are talking about is BIT DEPTH, not "bit rate". Bit rate is used to describe the data compression on mp3's.

It seems difficult to take your argument seriously when you are not even using the correct terms in your argument. IMO it really hurts the credibility of the information you are trying to describe.

Your overall theory mike-t3 seems to be, (in agreement with the other posters) a lot nonsense. You say your method is based on experience, and not book learning? Well I can tell you from my experience this is untrue. Use your ears, and trust them. My ears are my most important piece of gear, I don't care what some pre-amp manual or some other documentation says, I know what works and what doesn't, and regardless of whether or not I am recording to a 16 or 24 BIT DEPTH I always push my levels to as close to unity as I can.

I give up.

I hate it when people start spouting cliches.

I made my point. You don't have to worry about volumes levels to an extent when working at 24bit. I backed it up with evidence. And I made side points about the preamps which you don't agree with.

fine.

Do what you've always done. And don't bother to ask why you're doing it.

Have you ever stopped to think... What is the noise floor I am trying to avoid? Or why would turning the gain up on my preamp lower the noise floor? - no.

All I've done in this post is try to make you think WHY you are doing something. Why do you push your levels close to unity?! What are you trying to achieve?

You can believe what you want to believe, and I'll believe what works. I guess you have no response to your incorrect use of terms and how it affects your credibility........

You've not answered my questions.

Go on try it. Question your technique. Why do you do it?

Re the hijacking: No problem -- there's much to learn in here
Re bit rate/depth: I did this error myself in my post

mike-t3
I'm saying that ur hardware such as preamps wont perform as well at higher volumes.

I want to understand you and must therefore question this for verification: Is this the core basis for your argumentation?

If this behaviour of the preamps is the case, then because of the implication I'd have generally a problem: Because this is in the analogue circuitry. (The digital recorder wont do anything as soon as the signal became discrete other than saving it to disk)
And spinning this thought further, because this is common analogue circuitry, I'd also have this problem when I plug in a microphone into a mixer and preamp its peak to unity - this is always done as the first step when using a mixer.

This goes hand in hand with the question as to why this is done: To have a unified peak level of all signals in my mixer (= unity) which is at what is labelled "0dB" and describes the 'best' operating level of the mixer according to the manufacturer. I would be very worried if the hardware does not perform as it should at what is defined as the best operating level - especially if it's a common recorder like the Fostex-FR2 which I have.
(If it were for the preamp alone, I'd also expect it to have a controlable range that controls its behaviour linearly)

Or, looking at it from another side: If this problem lies in the hardware before the ADC (signal flow wise), then I'd have it always, regardless of bit depth, right? Is this high-peak problem then 'lesser' than the problems of quantization errors and high noise floor in 16bit?

You're sounds are fine. Part of the reason I apologlized for various highjackings was this thread has become something differnet than the original intention.

mike-t3
You've not answered my questions.

Go on try it. Question your technique. Why do you do it?

Glad to answer your questions:

I push the levels hot in order to keep the noise floor low. It is just good practice especially if you plan on normalizing all of your recordings and mixing them to a soundtrack, musical composition, video game, etc...... Let's say you normalize a file that you had peak values at -12db. This raises the noise floor 12db for this particular sound. Now you add 40 or so tracks to a multi-track session, which you have used the same 12 db normalize. Even at 24 bit there will be audible noise present in your mix.

When you are recording at 24 bit and you have your levels peaking at -12db you are really only useing 22 bits of the signal. If you are going to record in 24 bit why not use ALL 24 bits!!!!!!

It doesn't work like that because you've already normalised it.

You're boosting the gain on ur pre-amp which is also raising the noise level. Where do you think the noise is coming from?

Consider this.. You get ur pre-amp and you record a singer. You think to yourself 'I must make this as loud as possible because I want to make use of my 24bits' So you turn up the gain.. find yourself clipping at the peaks... so you add a compressor so you can record to your DAW at a higher volume - permanently altering the vocal whilst ur at it.

What you've just done is boosted the noise floor... by raising the gain on ur pre-amp... you've then compressed the signal.. making the difference the loud bits and quiets bits smaller.. which has infact raised the noise floor again.

Just forget about the compressor, reduce the volume.. record in to the DAW. apply a gate removing the noise and don't worry that ur signal is representing 20bits because you now thats plently of dynamic range and you fine well know chances are its going to be mixed down to 16bit anyway.

yes, i know that the noise floor will be present from the preamp no matter what gain setting i use. the problem there is also noise created in the ACD. This is why if you record silence with no preamp running to your adc you can still get noise if you crank the signal up enough. So when I normalize a low level signal i am not only turning up the noise from my preamp but I am also turn up the noise from the ADC!!!!!! Double whammy when you are recording at low volume levels. This is why you don't do it. Use the dynamic range that 24 bits gives you.

NEVER NEVER NEVER use a compressor in your signal chain when tracking. That's audio basics 101!!!!!!! Of course clipping your peaks isn't good either but I would rather take a second run at something to get a better signal than to record at low levels or use a compressor during tracking.

ejfortin

NEVER NEVER NEVER use a compressor in your signal chain when tracking. That's audio basics 101!!!!!!!

What... never ever...?

:lol:

once you compress you can't undo. You can always go back and compress after the track has been recorded.

yes. i know. I was joking.

so you are in agreement now??? do you see how it can be beneficial to record at higher levels?