Demonstrating Sound Quality Loss from AD/DA conversion. Lesson 24-B
ฝัง
- เผยแพร่เมื่อ 4 ส.ค. 2024
- In this video, we'll be discussing sound quality loss from AD/DA conversion. This is lesson 24-B of the Audio Engineering Bootcamp.
If you're working with audio files that were recorded using analog to digital conversion (AD/DA), you'll likely experience sound quality loss. In this video, we'll discuss the effects of AD/DA conversion on sound quality and how to improve it. We'll also provide some techniques for improving sound quality in your recordings without having to resort to AD/DA conversion.
This experiment reveals how much the conversion effects the sound quality.
Join the Facebook community:
/ 636985553955376
Download the tracks here:
www.dirtcheaprecordings.com/l...
Headphones used in this video:
www.iskproaudio.com/collectio...
Analog to digital, and digital to analog conversion are an important part of the signal chain in any recording studio, and for audio engineers. This is important to understand when mixing using analog outboard equipment. I also show the difference at various sample rates, such as 44.1 kHz, 48khz, and 96 khz. - แนวปฏิบัติและการใช้ชีวิต
![NONT TANONT - ดอกไม้ที่รอฝน (spring) - Reimagined [Live Session]](http://i.ytimg.com/vi/nY2PsFZFEOg/mqdefault.jpg)
Everytime I get "converter itch" I remind myself of this video. Excellent job! Thanks so much.
For anyone curious, the added highs you are hearing is aliasing from the filtering of the converter. But 200 or 400 times is an insane amount of round trips but really good to show how aliasing sounds to people who struggle to hear it in small details. Now granted Focusrite has a analog color to their mic & line inputs so there is some of that as well. The Lavery white paper on sampling rate brings up that 64 kilohertz would be the actual optimal sweet spot but the math for the converters didn’t lend itself to the abilities of early converters but that is why some early machines like ADAT, DASH, and DAT had a 32k option with the idea that 32k would become 64k in the future. I remember DAT machines having a 32k or 441k option on them YEP I’m that old 😂 lol.
Before finding your video , i did this test and i though i was crazy to do "100 clone".
At 400 clone...you "take the cake" my friend !
Thanks for sharing !!
Thank you! I really needed this information
I recently got my hands on some old school ADAT tape recorders for the purpose of archiving a friend's older digital recorded material and have been pulling my hair out trying to find a cost-effective ADAT to USB interface for the purposes of avoiding substantial quality loss. The cost for one of those is through the roof.
This video just proved to me I have essentially nothing to worry about. Thank you for the peace of mind, sir!
Hi, I use the for my old a-dat tapes the m-audio profirewire 2626, really recommend it!
And I was concerned about going out to tape and back, then out to 500 series and back at 48k...Thanks for all that work.
Fascinating stuff. I have been meaning to try this experiment for ages so many thanks for doing all the boring legwork so I don't have to!
Ya. Except I’m such a nerd it wasn’t actually boring lol!
Very informative experiment. Thank you!
👍👍👍👍👍 Great video! Interesting to hear the same test out/in through the clarett! The converters are really good!
Thanks for completely debunking the myth of multiple ad da for mixing... sure makes more confident options for workflow ... thanks a ton
What a fascinating test! Subscribed!
Thanks, I’m glad you found it helpedful :)
Thank you for this very knowledgeable info! Even people that have been recording for years don't understand this concept fully! Me especially lol thank you.
so happy you made this video. this is a topic that needs to be talked about a lot more - thanks
Thank you for this awesome video and for your time! Cheers!
Agreed, could not have said it better.
Really interesting. Leaves me with more questions than I had going in!
Yes… I think sometimes the more we learn…. The more questions we have.
Great test!
Awesome. I might do an A-D or two in any situation. Great info.
Fantastic experiment. Thank you.
Great video. Great great.
Great video!
Wow. A lot to ponder. Thanks!
Nice work, thx!
Wow quite interesting, great work!
Thanks, I’m glad you liked it :)
Very interesting, thank you. But, I think all this prooves is that most of us don't really need to worry about ad/da conversion causing problems unless you have some insane back and forth going on. was interesting to see 96khz holding up better. defo worth recording like that if you have the space.
If the difference is only really audible after more than a hundred round-trips, it is definately not worth the extra space and cpu cycles.
Thanks bud, always was wondering about this now I don't really have to worry sending out a few busses to some analogue gear.
Other than slightly mismatched impedance.
Great video.
Thanks Andrew :)
Very good and helpful, thank you.
You are welcome!
really helpful to understand that spot
I’m glad it helped :)
Nice video. I have an Clarett8 Pre and knew the pres were good, but dang.
This was a fabulous video as a newbie, so does this mean I should tracking at 96k? or I can just stay at 44.1k and go in and not worry about conversion if i am going in and out less than 5x?
Wow! What a test! I have an Apollo Twin and I want to buy a BAE 1073, but I was worried than the Apollo’s AD converter would degrade the sound, but I think it will be just fine. Thank you.
Like I mentioned about the preamps not beyond bypassable… you might not get the full potential of the bae 1073.
@@Get.Beautiful.Recordings Thank you.
Likely your mix room would be a bigger problem/diffence.
thx mate, i was just about doing this myself, but then found your video. Muchas Gracias for your work doing these clone-cycles up to 400 times. Would you say it is save In a normal mixing session to run through outboard gear, as long as you don't overdo it? best regards!
I was wondering whether I should add a patch bay or not for my 5 pieces of hardware chain..i guess it's ok to hard-wire everything to the interface directly and use multiple conversions. 5-6 conversion cycle is not going to make much of a difference it seems with a high end ad/da converter
It’s aliasing distortion that you’re hearing.. 96k has a much higher Nyquist value so you’re not going to get that fold back into the audible range like you would for a lower sample rate. The difference between interfaces is probably due to the saturation each one adds. Apparently the Clarett is a lot cleaner so you don’t get much aliasing distortion. So it’s not necessarily the conversion but the saturation of the preamps.
The point here being that the converters used in the most inexpensive audio interfaces are SO good these days, that even after 400 passes the difference is minor ! .............In other words, don't worry about it.
@@ronanzann4851 yeah, that is his point but if I recall correctly he says that he hears some “trashy treble” and doesn’t know why. That’s what I’m explaining. That’s my point..
It’s cool to see a definitive reason to be recording at higher sample rates esp if you mix outside the box.
Ya, I believe there is a slight benefit to higher sample rates. I will be experimenting and doing a future video on this.
Another good reason is if you intent to transpose down or slow down your recorded tracks. All the harmonics shift down when you slow audio, so you might hear stuff that 44.1khz would've been cutting off.
Slightly off topic but there’s nobody else I’d trust to ask- I have an RME UFX+ interface at my studio. I use a MacBook Pro laptop. Is there a difference in exporting my mix/master at home solely through the laptop vs. exporting while connected to the UFX? Thank you
Hi! Thank you for this video. VERY interesting seeing this. I use to record back my mixes in 192 and then simply convert to 44100. And the sounds is way better that recording at 44100. You guys have to try this, I am sure you will improve the sound in a 20%. Nice experiment.
That's very similar ( in principle ) to the graphics world with photoshop to get a sharper image at lower resolutions.
@@DanglyLingham yeah man, thats exactly what I think, more resolution, more pixels, then more deep and sonic quality. thank you for replying!
Sorry i don t understand : you mean recording at 44.1 , then you do DA to go to hardware and AD at 192 , and then convert the mix at 44.1 ? Thx
@@MELOPSMUSIC no man, what I mean is basicly all the recording takes you make needs to be at 192, guitars, synth, vocals, ,and then your mix ideally should be recorded at 196 and next just turn your mix into 48 or 44 using izotope rx or some good src. You will need 2 soundcard, one working at 44 or 48 which will playback your normal mixing your pro tools or reaper, etc, and the other for recording which will be at 192, hope it helps you
@@YerayIbarria great, thx !
I have the Apollo twin with the neuman tlm 103 what do you think I should get to upgrade my sounds
The gear you have is great, and capable rivalling a big studio if used correctly. If you’re not getting that level of sound, it has to be due to room acoustics, or your own skill level.
If you are getting that level of sound, but still want to upgrade, a microphone like a Bock Audio 251, or a good tube preamp, like a DW Fearn VT-2, but these aren’t really upgrades for you, more like just a different character.
Also, you didn’t mention your monitoring, but this is one of the most important aspects.
I run a Facebook group called Dirt.Cheap.Recordings where I give free advice for studio upgrades, and I’m always posting tips on saving money, and getting better recordings, feel free to join the group.
Is It usefull for sampling material?
Thanks so much. BTW, which mic did you use to record this episode?
I used a prototype microphone that hasn’t been released yet. It will likely be released later this year, and it will be called the iSK Stellar
@@Get.Beautiful.Recordings Wow it sounds great! Can’t wait to see it on your channel!
uhh the focusrite test may be flawed?? at 10;02 in the vid the focusrite track has 2 wave forms like duel mono or stereo? and the orig single it looks that way on your screen.? and your output shows 2 bars not 1 . please clarify this . that would make the focusrite track sound more pronounced .allot of people think incorrectly including me the first time I saw this on a small phone vrs a panel
What is that silver microphone? Looks so cool
Great Video and I agree in all your conclusions. But The AD/DA conversion is also related to the quality of the converters. In your video you are you using Hi End Converters Interfaces. Me too, I have the Lynx Aurora m. But when you record vocals and acoustic instruments happens when you use the under 400€ Interfaces the quality change. Can be great to make a video about that.
The same idea runs for mixing (my main task) When I upgrade from the Apollo 8x to the Lynk, the difference into de de DA conversion was really noticeable.
I have done this test with lower cost interfaces, but the challenge is that most of them don’t have dedicated line inputs, but rather, pad down the signal and run it through the preamp, which causes much more noise.
Of course the higher sample rate will sound better. Its taking more snapshots (per second) of the waveform it's converting into a digital signal.
I wonder if the dithering as an impact on these sounds differences. :)
Interesting video, thanks! The sample rate comparison is a bit skewed the way you did it, though, bc in-the-box downsampling does way more harm to your signal than a couple of clone cycles. And if you "clone" the downsampled file, you'll just amplify the downsampling artifacts ;) This is why in the end 96kHz always sounds way "better".
That’s an interesting point. Have you done testing to confirm that the downsampling does more harm to the signal than a couple of clone cycles? You are not the first person to have proposed this, and although I believe the degradation from down sampling is far less than that of the clone cycle, I have not tested it.
There are 2 ways I can test this.
1- redo the experiment with the original recording at 44.1 kHz, and upsample it to 88.2 and 96 kHz for the respective tests. And compare these results to the original results. If you are correct, then the 96 kHz one won’t sound significantly better than the 44.1 kHz.
2- I can take the original recording and do several hundred, or even thousand, clone cycles of just down sampling and up sampling. I would go from 96 to 44.1 to 88.2 to 48 so that I’m not undoing and re-doing the exact same file, and this would introduce the most artifacts in an attempt to maximize the degradation.
I’ll do a video in the future that incorporates both of these tests.
So to avoid audible losses, while still enjoying the benefits of external hardware, all you have to do is stay away from hundreds of conversions?
To me, Focusrite sounded clearer in the top end, with little to no degradation, BUT it sounded colored across the whole spectrum, as if there was some small amount of saturation applied.
However, I also want to add that what you said hold true only if you have good modern gear. On cheaper and/or older hardware the degradation will be more noticeable. But you shouldn't buy a lot of analog gear to your studio if you're using first gen Scarlett 2i2 anyways, and cheap analog gear can and likely will sound worse than just using plugins, conversion degradation or not.
Yeah.. I totally agree :) thanks
can you do a video about , how you do tis 400 times please cause maybe there is a hack to do it, cause ihave not the passion to make this with my RME
Hey nice video, but my queestion is:
if i use a piece of analog gear, it will just have that one conversion way. So D->A to my gear and A->D back in my DAW again
wouldnt this be just like one clone cycle?
So is there a big difference?
Correct, when you use analog gear, it’s just 1 cycle. The main reason I did so many cycles is to demonstrate just how little 1 single cycle makes.
@@Get.Beautiful.Recordings Ahh i see. So with the numbers of cycles you basically wantted to exaggerate the effect which takes place with one in order to show it to us. I got it now. Thank you. Nice technical video!
You asked the question that I was thinking the very first minute I started listening. Who does more than a cycle or 2. I did however run into problems with my gear when I was in the middle of a recording session recently. I had 3 people standing around while my focusrite claret 8pre x took a crap during my efforts. Started getting glitches and pops and one of the pre amps said bye bye and an over head for the drums was lost. All my gear in 10 years + including the octopre. Was a nightmare when the outputs wouldn’t work when I could only use the octopre as the master clock. Bought the redline 16 and ditched ADAT completely and everything is line in now. A dream come true. Everything is solid and clean. My macpro is a 2013 and struggling but it’s getting the job done. One cycle is all we need and it’s beyond amazing that he took the time to record 400 cycles. Darn.
Thanks for doing this! Very interesting. Must have been tedious doing all those passes :-) Was this a 3D LIO or an older generation?
Yes, I used a 3d lio. Thanks for the comment:)
@@Get.Beautiful.Recordings Wow, I'm shocked the Clarett sounded better!
@@ozzy3ml I was shocked as well :)
Who the hell is cycling back & forth 30 times! OR 400 TIMES! THE PRB IS THE PERSON DOING IT!
What converters are you using? Or which interface?
I use a Metric Halo ULN 8, and an SSL alpha link Madi as my main recording rig.
My question is who re processes audio in and out of a converters 40 times in the real world? Mixing or Mastering Engineers only process the same signal a couple of times with analog gear.
That's correct. This is not a real world scenario, it is an experiment designed to push the limits, and learn from the results.
It would be interesting if you put a hard low pass filter on the original mic recording so there was nothing above 20k (would have to set it to something like 15k or so) since these harmonics above our hearing range affect harmonics in our hearing range. My guess is there would be little or no difference in sound and the focusrite may already be doing this when you record at 44.1k.
That’s a good idea… wouldn’t hurt to test it
Interesting test. Theres an old saying I read from Yogi Berra that applies here. " In theory, theres no difference between theory and practice, but in practice, there is.
In theory your test shows up the degradation but in practice, how many people ever send any audio out of their DAW 400 times? In practical real world scenarios, nobody is gonna be able to tell any difference as to whether your mix is all in the box or that some of the tracks use outboard gear. Your recordings are gonna be assessed on whether people like the tune and if the track sounds good or not. Converters these days, give better dynamics range and lower noise than the Studer 2" machines and analog mixers ever did all without the constant maintenance headaches involved in using tape. Do I like tape and the sound it provides? Absolutely but Ive abandoned that because nobody is gonna get the full benefit of using that old school recording chain, when the end user is listening to mp3 tracks off their phone using a pair of earbuds.
so whats the point ? im not recording the same sound more than 3 x let alone 200 x. can somone explain what was the point ?
Yes, nobody records the same sound more than a few times. However, if I’m this test we only went through 3 generations, the signal degradation would not have been enough to be audible, so I did it 200 times to exaggerate the effects.
@@Get.Beautiful.Recordings so what was the point ? That the interface don’t matter. At this point ?
Weird... With more cycles, it sounded like compression to me. It sounded as though the dynamics were being squashed with levels increased.
So for sweet crunchy highs convert 400x's got it!
I very clearly heard all of the differences on ordinary home stereo speakers. But, the non-audiophile scientific types will always say we are imagining it.
Ok thank you, it's useless to record in to same computer whey bouncing a mix, it will better to capture into different computer with 96K to AD. I was thinking this a lot, if you use 48K project on your mix it make sense to capture in different computer with AD on 96K capture. Thank you.
so I see the difference is that the digitized version is being converted. over and over a copy of copy. so effetely digital version being copied over and over again, what if yo go back and copy the oringinal again and again and not the copy
This is hilarious 🤣
What is the practical application of this exercise in a real life audio engineer scenario?
Everyone on earth not converting a sample 400 times:
This guy ⬆️
Hi Mateo,
The practical application is for people who use analog gear to have confidence that they’re not losing sound quality due to the conversion process from just going out and in once or twice, and it also demonstrates the value of recording at higher sample rates.
Thanks for the comment :)
I kind of liked MH artifacts more than what Clarett did to the sound TBH
Wow
And just how many people are going to cycle the same track more than once or twice . ?
Which makes me wonder what is your point?
The point is to demonstrate that for most people who cycle through the conversion once or twice, the sound quality loss is negligible.
192 and DSD all day I just invest in hard drive space I don’t play around with audio. We do it as best we can and when you play it back on the computer, it’s not the same as tape real to reel less is better this is why we have a mastering engineer
Ya, if you’re computer ca handle it…. Why not?
I don't see the point going above 96 KHz unless you're doing stretching of pitch shifting...
We need a plugin to simulate dac re recording now 😅😅
This is very telling. It tells us 1- modern conversion is fantastically accurate. Most people use only a few different “clone cycles”, if that. I dont even use outboard gear.
2- im not sure how to interpret the samplerate affects. I suppose you could say higher samplerate yield better “sound quality”. Though ive always been opposed to saying that without qualifying the statement. I assume the anti aliasing filters on this device (Clarett???), arent the greatest at 44.1. So at 96k youre hearing less of the aliasing in the top end. Again, nobody is clone cycling 200+ times in real world use. So a few times should yield ZERO difference at any samplerate. I suspect even better results would be had on higher end gear like RME or Lynx converters etc…..
this makes me think that people who are swearing by their incredibly expensive converters probably have devices that they never really needed. or maybe the expensive ones add a pleasant sounding saturation.
Yes… I think it’s a bit of placebo effect too.
this extra high pitch and the clock leaking after 200/400 cycles
96 is fuller and true to the original
So don't buy 10.000$ converters guys ^^ If you want to add some nice harmonics you can still buy a cool color box or line preamp
exactly!
i think its time you change your own converters. and nobody really needs to add anything 200 separate times into a single mastering chain so that it sounds f as much as 400x lol ))
It’s a shame he didn’t change his guitar strings..
Obsess much? .) I'm joking , cool experiment
So basically..anyone using under 40 pieces of outboard gear wont hear a huge difference
Lol… depends on the converters, but yeah… that’s it :)
@@Get.Beautiful.Recordings But in real life there no such a thing as Clone Cycle, every cycle is different is like starting again , because you are adding to the signal, so then you have a new sound not a Clone. Even if you go out and in lot of time , every time you have a different sound that include the added analog gear, so there is not going to be degradation . But the experiment for education purpose is great.
it s not 0 quality loss, added noise and that shit.
Now do it 40000 cicles
Lol…. I don’t think i could
there is no difference the same guitar playing dom dom dom
If there was any real worry he'd only be doing one cycle.. there is No real worry in the real world. This kinda shyte is a joke to actual engineers.
Exactly… there’s no real worry about adda conversion. Thanks for your comment:)