I need an oppinion

If you need to write about anything else please do it here..
jpn
Vip Member
Vip Member
Posts: 386
Joined: Mon Apr 19, 2010 9:16 pm
Location: Portugal
Contact:

I need an oppinion

Post by jpn » Sat Jan 24, 2015 3:08 am

Hi , I need an oppinion:
Sometimes in large projects it´s very difficoult to work in 88.200 Hz
so I have to mix in 44.100
What do tou think is beter? recording ,making some processes and mix in 44.100 or record in 88.200 make some processes ... then convert to 44.100 and mix?

Martinez
Member
Member
Posts: 293
Joined: Mon Aug 09, 2010 2:03 pm

Re: I need an oppinion

Post by Martinez » Sat Jan 24, 2015 5:04 am

You know, I don't think it makes a huge difference what sample rate you mix at. It does make a difference to be sure, just not a big one.

RJHollins
Expert
Expert
Posts: 4014
Joined: Sun Mar 28, 2010 5:53 pm

Re: I need an oppinion

Post by RJHollins » Sat Jan 24, 2015 6:41 am

hmmm ... I think you'll get a whole lot of opinions ... and all over the place with what others use ...

so here's mine :lol:

I would not rule out that bigger, faster could be better ... however, and this is NOT a justification but more of a running resultant.

I started working with digital sometime in the early '80s. Granted, it wasn't until several years into it that consumers were getting more involved. In those daze, when we got to 44.1, 16 BIT ... that was our format. Whether the early SONY or Mitsubishi [even the Fairlight] multi-tracks ... that was it. We later got digital 2 trks, like the R-Dat format [Panasonic 3500, 3700]. They were all 16bit, with NO room to play with ... it was the same spec as the release [CompactDisk].

Throughout that whole time we still had to release records [or CD's]. We were also expected that this releases were to, hopefully be hits :lol: and there were those :mrgreen:

Speed ahead and we have 96k+ and 24bit+. These are wonderful things as technology advances.

I've had the pleasure to work full sessions at the standard CD 44.1, and at the higher rates. I've also done combinations of both.

There in is where I feel the conflict :| In the studio we are allowed a unique perspective that most [or only a select few get to hear] and that is a direct, A:B comparison. The majority only get to hear the 'final' version [whatever it is].

I certainly have my personal preference ... but ya know, it's also one of the benefits of being an audio engineer ... we have the original sources :mrgreen: To that, I much prefer listening in a 'controlled' room. I'm kinda spoiled that way :lol: Most people have never had even that opportunity [sadly]. Maybe the closest is going to hear a symphony orchestra in a great listening hall. [or any style of music ... but in a room DESIGNED to listen to it].

Be that as it may ... I also do NOT subscribe to aiming for the lowest rung, at least NOT by intent :lol: Reality may dictate that not every project is allowed that [multitude of reasons ... none of which welcomed ... but again ... the demands of the real world can force us to only do our best.

Anyway ... we don't always get unlimited budgets, time, and stellar performances of spectacular arrangements of exquisitely crafted 'hit' songs ... and I don't find evidence that 'sample rate' had anything to do with determining the 'quality' of a recorded piece.

with that ... I never gave an answer to the original question, did I :lol:

I don't see a 'text book', 'one size fits all' answer.
i7-5820k, MSI X99A Plus, 16 GIG Ram, Noctua NH-D14, Win-7 Pro [64-bit], Reaper-64

NVC [Nebula Virtual Controllers]

jpn
Vip Member
Vip Member
Posts: 386
Joined: Mon Apr 19, 2010 9:16 pm
Location: Portugal
Contact:

Re: I need an oppinion

Post by jpn » Tue Jan 27, 2015 9:16 pm

Thank you for your opinions :)
I know what you mean RJHollins...
I have arrived also to my own conclusions some time ago , but it never hurts to confirm our thoughts .
Thank you again Martinez and RJHollins to share your time and your experience.

User avatar
ianc
User Level X
User Level X
Posts: 107
Joined: Mon Feb 18, 2013 4:17 pm
Location: UK

Re: I need an oppinion

Post by ianc » Sun Feb 01, 2015 12:30 am

http://www.trustmeimascientist.com/2013 ... n-it-isnt/

Read your post last week but it requires a kind of complicated answer that I couldn't really manage to put in words, there's a lot of depends to it.
Came across this today and thought it covers the subject quite usefully, as well as being a interesting read along the way.
Hope it helps add a bit more to make the suitable decisions for the task at hand.
And the other thing to bring into this discussion but only briefly touched upon in the article is "The McGurk Effect"
I wish I could drum half as well with two hands as one of Terry Bozzio's feet

TranscendingMusic
Expert
Expert
Posts: 1266
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: I need an oppinion

Post by TranscendingMusic » Sun Feb 01, 2015 3:15 am

I have my own opinions about this topic but why not try it for yourself? Mix down a simplistic arrangement with a good amount of non-linear processes. Do it first in 44.1 and then 88.2, mixdown and bounce out and compare.

For further personal enlightenment :D compare a dithered down version of the 88.2 as well as an un-dithered, un-converted version.
Cubase Pro 10.0.60 x64
Ryzen 2700x | Win 10 x64

jpn
Vip Member
Vip Member
Posts: 386
Joined: Mon Apr 19, 2010 9:16 pm
Location: Portugal
Contact:

Re: I need an oppinion

Post by jpn » Mon Feb 02, 2015 3:46 pm

Thanks iank and TranscendingMusic for your opnions...
TranscendingMusic , as you can imagine sometimes in different music "styles" the results of the same technics may be sound different , and we "can´t" test everything we need.
The reason I start this subject is .. if possible to get the maximum possible of oppinions and experiences.
Theres a lot of experts in this "place" ;)
and I think we' all will be grateful of the transmitted knowledge.

User avatar
Sheikyearbouti
Member
Member
Posts: 318
Joined: Wed Feb 16, 2011 6:49 pm
Location: London, UK
Contact:

Re: I need an oppinion

Post by Sheikyearbouti » Mon Feb 02, 2015 9:42 pm

TranscendingMusic wrote:For further personal enlightenment :D compare a dithered down version of the 88.2 as well as an un-dithered, un-converted version.
I guess a typo? Why would you dither? SRC doesn't need dither.
--
Nikolay Georgiev
http://georgievsound.com/

User avatar
Sheikyearbouti
Member
Member
Posts: 318
Joined: Wed Feb 16, 2011 6:49 pm
Location: London, UK
Contact:

Re: I need an oppinion

Post by Sheikyearbouti » Mon Feb 02, 2015 9:54 pm

As to sample rates... it is debatable.

I have found that when you record distorted signals higher sample rates push the aliasing above the human hearing range. You can see such aliasing also when you sample in NAT, even with the best of converters...

To me the point of higher sample rates is that your converter puts the digital anti-alisaing filter at higher frequency. Therefore, whatever damage it is that it does, it happens at higher frequencies.

Someone recently asked what is alisaing and if it matters. I work pretty much all the time at 96k with programs sampled at 96k and didn't think of the case when the SRC is done inside Nebula. I will not be surprised if the SRC in Nebula isn't so great [I may be wrong though!]. So this could be a factor.
--
Nikolay Georgiev
http://georgievsound.com/

TranscendingMusic
Expert
Expert
Posts: 1266
Joined: Sat Mar 27, 2010 7:01 am
Location: USA

Re: I need an oppinion

Post by TranscendingMusic » Mon Feb 02, 2015 11:32 pm

Sheikyearbouti wrote:
TranscendingMusic wrote:For further personal enlightenment :D compare a dithered down version of the 88.2 as well as an un-dithered, un-converted version.
I guess a typo? Why would you dither? SRC doesn't need dither.
Hi,

I mention dither as to reference a typical, final format such as Red Book. So 16-bit, 44.1. You remove other variables and hear the sound as you would in a real-world situation. For self-knowledge, that's why I say also test the "rawer" form without dither as well.
Cubase Pro 10.0.60 x64
Ryzen 2700x | Win 10 x64

jpn
Vip Member
Vip Member
Posts: 386
Joined: Mon Apr 19, 2010 9:16 pm
Location: Portugal
Contact:

Re: I need an oppinion

Post by jpn » Sun Feb 08, 2015 1:07 pm

Thank you all for your oppinions.
Thank you ianc for the link (very nice article).

Some times asking oppinions is a shortcut to increase "personal enlightenment" especially in a delicate matter like this...
all information count .... at least for me.

User avatar
Sheikyearbouti
Member
Member
Posts: 318
Joined: Wed Feb 16, 2011 6:49 pm
Location: London, UK
Contact:

Re: I need an oppinion

Post by Sheikyearbouti » Sun Feb 08, 2015 2:58 pm

Than there is also the question - how do you filter (LPF) and how is this or that particular Nebula program made?

From what I know, usually you will have 2 filters at your ADC stage (one analog for the oversampled signal and one digital linear phase for the actual sample rate at which you record for when the oversampled signal is than downsampled to go to 192, 96, 88, 48 or 44). Please correct me if I am wrong or missing something, but it is what I remember from reading books back in the day... Quite a lot of thoughts below and as you can see a bit chaotic.. and a lot to discuss and check really.

Some developers swear in native sampling (so separate sampling session for 44, 96, etc.). I personally don't see the point of this. Least to say, you waste a lot of time sampling the same device more than once. Than you also lose consistency.

OK, if you sample at, say, 192 and than downsample, you would have the (digital) filtering twice (in fact 3 times if we count the analogue filer in the converter too). And if you native sample, it will be 1 filter (2 with the analogue). So in theory it may be better to native sample. But than in practise we are talking about a linear phase filter applied at a very different frequency range of the audio spectrum (so if there is ripple from the first filter, the question is - is it not something we will remove with the second completely?). And also you miss the fact that when you downsample offline (SRC), you can chose the best configuration of the filer for the particular program, and not use the default fixed filter in your ADC (so you can chose to have steeper filer, or to lose some of the VHF content and make the filter less steep by adjusting the cut-off frequency, or adjust pre-ringing parameters, and this depending on the actual aliasing created when you sample - so for a clean program you can have one setting, for a dirtier another).

You also have the well known Lavry argument that at sample rates higher than 88/96 the converter delivers worse performance. If you check the Lavry Gold DAC you will find that the highest sample rate is 96k. And it is a 5k device for 2 channels!

There is another argument, which I kind of favour - and it needs more testing on my side - that at 88/96 your converter works better than at 44/48 and thus you think that the recording is better at 88/96 than at 44/48 (well, in fact, it may also be that the recording slightly better, but not only). This can easily be tested in a blind listening test with HF rich signal that is record at say 48 and 96. Than the 48 is upsampled to 96 and the two 96 versions compared with the original 48 one. If the upsampled 48 sounds better than the original 48, than clearly it is your converter that sounds better - not the recording.

Also to put it this way - when I record, say, distorted brass (yeah, I love putting ribbons on brass and than cranking a copy on another channel on the N**e console to distort it and do parallel distortion to add some highs) I am 100% that I get cleaner sound at 96 than at 44 - my explanation is aliasing. But when I record clean piano, this HF content is not there to create aliasing.

For now to me 96k is a very good safe area to work at. I am doing tests to see if 192 is better for NAT sampling with additional offline SRC.

If I have to downsample to 44/48, I would FIRST process in Nebula (so preamps, some basic EQ, etc.) and than downsample. The idea is that you use the higher res Nebula presets on the higher res audio and than you go down to 44/48 to finish your mix. But I also like to commit to processing and process to tape when I record, so this approach works well for me - for others it won't.

Than there are people who go for DSD (or modern equivalents of it) to avoid the filtering at all (no need for anti-aliasing filter there - at least in the first stage of recording). But in the end, you still need a filter to make the program usable in NAT/Nebula. So it is not such a massive benefit, if there is any at all - but again, I've no real opinion, as I have never tested it myself.

And than the last option is to SRC in Nebula, which I have no opinion on too, but would generally avoid, exactly because it is an unknown process for me that I'd rather test first.

For example SADiE was notoriously famous for bad SRC (fixed in the latest version). And SADiE and Prism are one company really. So even the they had it wrong until recently.
In any case - all these will be relatively minor differences and of course there are lot more important factors to consider - what is the source you record (player, instrument, room), where do you put that microphone, what is the gear you use for the recording. Only than comes Nebula and its sample rate and the way the programs are done...

And BTW, see, I do not mention anything about hearing above 20k, which I think is marketing bull. Although recently I have been hearing poeple from serious companies talking about interaural time differences in localisation that can be detected above 20k time span - i.e. the effect of sample rate on transients (how come we can hear if we can't hear??? ahaha). I think that if implemented perfectly 44k should be just enough for us humans (the questions I ask are related to - is it implemented perfectly?).

Yes, there are tons of amazing records released at 44k/16bit. Still, my job as an engineer is to deliver the best I can. Thus I try to stay at 96 for as long as I can in the production process, ideally until the final point when I have to prepare the audio for CD quality release. I even bounce, freeze and commit to achieve this.
--
Nikolay Georgiev
http://georgievsound.com/

User avatar
david1103
Expert
Expert
Posts: 517
Joined: Wed Mar 31, 2010 12:26 am

Re: I need an oppinion

Post by david1103 » Sun Feb 08, 2015 7:19 pm

Thanks for the great post Sheikyearbouti!

I have just done a whole lot of tests and am certain that my soundcard sounds worse recording at 44.1 than 88.2 and 96! I am sure it has nothing to do with higher rates being 'better' but how the ADC works in my sound card. I would also expect a better ADC to sound better at 44.1.

Conclusion, there is no escaping the fact you need to try all the sample rates and decide which is sounding best ON YOUR SETUP!

I have proven beyond a doubt that I get a much better sound recording at 88.2 or 96 (doesn't matter which) and then using Sox at highest quality settings in 100% linear phase mode to downsample to 44.1

http://sox.sourceforge.net/

Its not that subtle, I am only testing on my recorded voice. At 44.1 native recording it sounds boxed in and lacking air.

It's important to note I tried using Reapers SRC and the results were only a fraction better than native 44.1. You need to use Sox downsampling in the right mode to get the better sound.

richie43
Expert
Expert
Posts: 5049
Joined: Fri Feb 04, 2011 8:47 pm
Location: Minnesota, USA
Contact:

Re: I need an oppinion

Post by richie43 » Sun Feb 08, 2015 7:43 pm

david1103 wrote:and then using Sox at highest quality settings in 100% linear phase mode to downsample to 44.1
I am a smart guy (I like to think so at least...ha ha), but I have never been abke to wrap my head around SoX. I even am familiar with command line, Linux, etc. But I have never been able to make SoX work.....
David (or anyone else): Would you be willing to help me at least to the point of getting SoX to do some simple conversion? Once I get that far, I am sure I can dig deeper on my own.....Please.......?
The Sounds of the Hear and Now
http://soundyaudio.com/

User avatar
david1103
Expert
Expert
Posts: 517
Joined: Wed Mar 31, 2010 12:26 am

Re: I need an oppinion

Post by david1103 » Sun Feb 08, 2015 7:57 pm

richie43 wrote:
david1103 wrote:and then using Sox at highest quality settings in 100% linear phase mode to downsample to 44.1
I am a smart guy (I like to think so at least...ha ha), but I have never been abke to wrap my head around SoX. I even am familiar with command line, Linux, etc. But I have never been able to make SoX work.....
David (or anyone else): Would you be willing to help me at least to the point of getting SoX to do some simple conversion? Once I get that far, I am sure I can dig deeper on my own.....Please.......?
Hi Richie,

Are you on a PC? I am using the Foobar Sox plugin:

http://www.hydrogenaud.io/forums/index. ... opic=67373

It's very easy to use. If you are using Reaper then this reascript might be the best solution:

http://forum.cockos.com/showthread.php?t=145240

You need Python installed on your machine to run the script.

Hope one of those options works for you :) I am sure there are many different ways to use Sox other than the command line.

Post Reply