WHAT is the BEST SAMPLE RATE?

Read more about it here:
web.archive.or...
Thanks, wayback machine! 🙂
MERCHANDISE: teespring.com/...
EXTRA CONTENT: / whiteseastudio
STUDIO: whiteseastudio...
BUY GEAR WITH MY AFFILIATE LINK:
Sweetwater: imp.i114863.ne...
Thomann: www.thomann.de...

[@humanlikeee]

i find 0 Hz produces a really warm tone that you cant replicate with higher sample rates

[@rohinagrawal9727]

NICE ANALOG WARMTH

[@owenathanael]

432 is much better. It can potentially heal cancer.

[@iliatilev]

@@owenathanael The sad thing is some people believe that ;)

[@owenathanael]

@@iliatilev true.

[@NickBatinaComposer]

#noTroughsHereGang

[@trevorclover]

You are a person in the audio world I really respect and also the last one to say that recording at 48 kHz makes sense, I remember your old video where you said that 192 kHz was the best. To be able to rectify your ideas is praiseworthy. Cheers!

[@nomelocreo1979]

@@trevorclover read shannon es.m.wikipedia.org/wiki/Teorema_de_muestreo_de_Nyquist-Shannon

[@plk173]

high sample rate is really useful for resampling in sound design tbh, its nice having that high head room when you pitch stuff down many octaves down or really on weird distortion and other audio mangling

[@zac8670]

This is true, it's very noticeable.

[@KomodozGaming]

Ahh, I see this is makes sense now

[@MotoLemmon]

Thanks!

[@mrbanana779]

By far the most insightfull video I have ever watched upon the topic is the one on the FabFilter KZbin channel "Samplerates: the higher the better, right?". Very interesting video, very much een aanrader!

[@zonasound]

I've worked over the years at Universal, Enterprise, many big studios and the most common preferred bit and sample rate is 48K 24Bit, so i've always worked at these rates plus the plugins function much better unless you have an HD system

[@AdamSpade]

Video editing programs call for 48k/24bit and so I have always used that as well for the compatibility. Set it and forget it. Though I’m not a mix engineer. I’m a film composer and songwriter that does his own tracking, and I am working with a lot of samples that are already 48k/24bit as well. It’s pretty much the standard in my world.

[@zonasound]

No this is just solely music production

[@error8418]

@Lostie DeMonde Absolutely, bigger numbers are always a great selling point.

[@hoborec]

Nice to see you talk about this too! I made a video about this way back, and made sound examples with some cheaper converters. Some people got pretty upset and claimed that I did something wrong haha.

[@danielhornbeck6588]

Would you mind directing me to that vid?

[@insanebiscuit1]

I used to run at 48KHz, but recently moved to 96KHz since I now have a much more powerful computer. Pretty much all plugins support it now and any that do not internally oversample will benefit from 96KHz natively due to the reduced aliasing. Wont make any difference to those that do already oversample though.

[@WaterRiver777]

What format do you convert to for 96

[@insanebiscuit1]

@@WaterRiver777 WAV

[@WaterRiver777]

@@insanebiscuit1 where do you upload too

[@insanebiscuit1]

@@WaterRiver777 i dont upload, i mix for others

[@poetnprophet]

I went to 48k from 96k about a year ago for your reasons and also to save CPU overhead and drive space. It's been great!!!

[@tunemxr480]

192k yields some unruly file sizes to catalog, really only noticed a difference with a client playing classical music on a 9’ Grand piano with tremendous dynamic range in the music. Also it greatly reduces your access to how many plug ins can be instantiated. I agree-24bit/48k is the ideal tracking paradigm.

[@DragonboltBlastter]

Stock plugins are good enough, also manufacturers can always upsample their plugins

[@MichaelW.1980]

How can the dynamic range be affected by the sampling rate? The bit depth controls the dynamic range, by lowering the noise floor. And the hardware, no matter if it’s audiophile consumer hardware, a prosumer audio interface or professional gear, has yet to surpass the limits of a bit depth of 24 bits. And nobody would ever hear a noise floor of even 16 bit recording, unless the volume is at a level that will hurt and degrade your hearing capabilities quickly. So the advantages of 24 bits really only are of any interest in the field of recording and mixing. Or do I forget something here?

[@Str4t0sph4r4]

192k has nothing to do with the dynamic range, it’s bit depth that is responsible for dynamic range.

[@samyt681]

>really only noticed a difference stop lying mxr

[@michaelfarrow4648]

I have had the option of high sample rates for a long time, but the final format for the film scores I worked on was always 48k, 24 bit. My decision was to track at the sample rate of the final (dub stage), so I recorded and mixed everything at 48/24. My experience is that Sample Rate Conversion does much more damage to the sound than any benefit that may be gained using high sample rates. The 3 worst words in digital audio: Sample Rate Conversion. Your Mileage May Differ :)

[@Synth2000]

I track at 48/24, master is tape and then 96/24. The quality of the converters is more important that the sample rate. Depending on their clock and design, converters can sound better at specific sample rates. Find your sweet spot.

[@johnthecreative]

do you ever have issues converting a client's 44.1 khz project to 48 khz?

[@Synth2000]

@@johnthecreative I would keep it at 44.1 while mixing

[@johnthecreative]

@@Synth2000 tell that to many pros that do 48.

[@Synth2000]

@@johnthecreative My 44.1 comment was a response to the former poster. I currently work using 88.2 and DSD, and for many years I was on 48/24.

[@Rompler_Rocco]

My Tascam maxes out at 48/24, so I'll vote for that... Second place would be a Maxell xll-ii running at double speed 🤷‍♂️ ;)

[@RealHomeRecording]

6:08 it is worth putting reaper on the extreme setting for sample rate conversion. It has the least amount of aliasing according to that sample rate converter comparison website which I don't think I can link to without KZbin dumping this comment in the spam folder. Good video, Wytse!

[@noxbeatzproductions]

Always been a 48 kHz guy myself

[@DMarlow83]

Your favorite plugin manufacturer Acustica Audio recommends running their plugins @96k :)

[@rebelchaeper707]

Only time I find use for higher samplerates (96kHz), is when I do field recording-I do walk around with mic and headphones and capture interesting sounds, I can later transform in studio. Samples recorded at higer samplerates respond, sound better, when time stretched (usualy slowed down). And later do bounce them to my working asmplerate-48kHz, that I also use for recording bands.

[@STAR0SS]

Would be interesting to record some music at 192Khz and then spectral shit the top part into the audible range (without folding like in aliasing), would be the equivalent of a gamma ray picture.

[@tomaszmazurek64]

Higher sample rates on conversion are useful for creative sampling - if you plan to radically slow down or pitch down your samples, its a good idea to convert at higher rates, so you do not lose all the high frequencies.

[@kadavr0s]

Actually it does not work like that. If you tune down your sample, your low frequencies get pushed to high frequencies area, so your 100 hz becomes 200 hz, etc. So having frequencies above 20000 hz won't help - they will be translated to the inaudible frequency range anyway.

[@tomaszmazurek64]

@@kadavr0s I think you got it backwards. Maybe I've explained it poorly.

[@Skrenja]

​@@kadavr0sNah fam, there's definatley a huge difference in resolution if you're slowing down or speeding up high sample rate files. It's almost like recording at a high frame rate with a camera.

[@tmoblak]

what about sound design? aren´t higher sample rates more interesting there? i mean, in the terms of audio-stretching, slowing things down and stuff like that. (i am talking digital games and scifi movies here). in theory, you should be able to slow audio down without hearing too much artifacts when recorded in higher sample rates, right? (and you are golden when you have enough money to buy those sanken mics that go up to 100kHz)

[@phenixnunlee372]

Yes and no. No to the fact that a band-limited nature of sampled audio it will not help you extrapolate audio because, the signal is fully defined by the samples. However, higher sample for reconstruction and noise reduction higher sample rates help because, clips and pop can easily extend to 50kHz and do build to filters to remove them you need that content to exist.

[@Audio_Simon]

The one big reason to use 192khz is if you want to time stretch audio. More sample points, better result.

[@alesnovak2906]

Yes there are more sample points in general but the region,say,between 20hz-20 khz is represented with the same number of sample points in any scenario.

[@iammodus]

The Resampling feature at 6:00 is basically how accurate you want sample stretching to be. When you stretch a sample to be a different length than the original, you might run into these really digitized, computery artifacts (especially if you make the sample slower). The higher the resampling the better the stretching will be, and the less audible the artifacts.

[@mennovanderlaan6695]

That computery artefact is so nice of crunch to samples. Thats the whole reason i now use arturias emu sampler. 20 years back used the emu 1212 chipset for that.

[@Mefistofy]

Stretching is a concept not quite applicable to samples. They are defined as scaled dirac impulses which have a infinitesimally small width and cover a area of exactly 1. The option in reaper fives away all the information: 512pt sinc interpolation. You use a sinc function sin(x)/x where x is chosen according to the desired low pass frequency. A perfect sinc goes from -infinity to infinity though. You approximate it with 512 points here and call it good enough. The 512 points (if one side is meant) lead to 20*log10(1/512) = -96 dB which is the theoretical limit of 16 bit PCM. For each resulting sample you need to do 512*2+1 evaluations of the sine and one division though, both quite expensive operations on CPUs. The sum over all is cheap at the end. The amazing part about this technique is that you can choose any resulting samplerate you want. Fancy some 37kHz ? No problem. TLDR; samples can't really be stretched, sinc interpolation is great but computationally expensive.

[@iammodus]

@@Mefistofy I meant an audio file, not individual samples, sorry if it wasn't clear. Stretching something like a guitar to last longer, or be shorter. They are indeed pretty expensive on CPU, and that's why one should save higher point counts for offline rendering.

[@Mefistofy]

@@iammodus Did not think of the sample as sound recordings. I might bee a little to much into DSP sometimes. Everything you mentioned is exactly on point.

[@Projacked1]

I run on 88.2k, as it's the easiest for the converters to handle (to 44.1k) , unless there are 2 dedicated circuits for (doubling) 44.1 and 48k. 192k should be converted back to 48k if you want to be sure of matching bits and their calculations. Floating integer calculation at 64-bit will handle the conversion better, but I would stick to doubling/ halving myself.

[@EricOehler01]

It seems like it should be easier, but mathematically the conversion from 88.2 to 44.1 is just as complicated as 96. It's not a case of removing bits or samples or anything, it's still fourier transforms and curve fitting.

[@Projacked1]

@@EricOehler01 as far as I have heard dedicated to multiples was the way to go. It seems logical too when dividing 2 by 3 in integers. that's where the 'curve' fitting gets slightly affected, The bits are changed by the tiniest float points or something like that. I saw it in a presentation from Marantz in their high end stuff. It def changes the sound, butm at the moment processing is way improved, so I wouldn't worry that about it that much personally. That said, I record and mix in the same samplerate. Always. If I change the settings on both my interface or DAC there will be differences in the sound. And not always for the better. Steady is the way to go imho.

[@Dthebeatsmith]

@@EricOehler01 I thought the calculations are easier with 88.2 to 44.1 and vice versa because no interpolation needs to occur between samples. That vs something like 96KHz to 44.1 which would require some interpolation because of the non-integer division of the sample rate no?

[@pyratellamarecordingstudio1062]

Eric is right here. The “easier calculations” thing is really a myth. I always encourage people to do their own tests to see what actually works and sounds better. So I encourage you to try 96 vs 88.2 and see what you like best. At those rates it’s a slight difference and probably doesn’t matter very much though.

[@EricOehler01]

@@Dthebeatsmith it always requires interpolation, because downsampling isn’t strictly a divisive operation. My DSP math is weak and I haven’t done any since college a loooong time ago, but YOURE not just picking the path between samples, you’re refitting a curve to a new periodic rate. Basically saying “this equation described this curve at x sample rate, what equation does it at y” We’re used to the notion of halving and multiplying sample rates because it makes logical sense, and works for visual mediums and frame rates. But the math is complicated and fussy. If it were as straightforward as halving things, there wouldn’t be any differences between SRCs and we’d never have to worry about whose DAW did the best downsampling. :)

[@psilocyberspaceman]

The link is invaluable. Many thanks.

[@aviatedviewssound4798]

yeeeeees thx white sea studio i also work at 96khz and export at 192khz for aliasing and to apply oversampling built-in or use metaplugin for aliasing and convert it back to 48khz using dbpoweramp which is the most clean pristine sample rate converter available on the market.

[@yzmikh4570]

Good video! Totally agree, samplerate above 48 kHz is redundant unless you gonna do some pretty gnarly sound design, purely to reduce aliasing across the board. Really easy to resample to 48 too in a good DAW that has a focus on hectic processing. Something like a Reese bass with LOADS of saturation or a very smooth sounding FM bass that has multiple stages of frequency shifting (basically the situation described at the end of the video)

[@technodrone313]

96k/24bit is what I use and I like it. I record into my desktop because my pos laptop cant keep up like you said.

[@ManvendraShahBEE]

Absolutely love your videos!!! Great advice. Hats off!

[@inspyremusicnl5826]

finaly a Good explanation why you should(Could) cHOOSE for 48khz over 44khz . lots of people in youtube videos say:just work in samplerate whatever the final file sample rate will be. because of the fact that the conversion from one samplerate to the other could possibly create new problems. but very clear why u choose for 48khz. thats why im subscriber i need all this nerdy stuff. Keep on going!

[@eurz9188]

An additional consideration when choosing converter sample rate is latency. If you're doing "offline" work (record first, then work with the result) you may not care, but it can be quite impactful for "online" work where you want to minimize the delay between input at the microphone or instrument and sound out of the speaker (of course, converter rate is not the only reason for latency, just one of the multiple factors in the equation)

[@donttalktome4696]

Very very cool. I was just thinking and messing around with this last night.

[@TheBunkhouseStudios]

I have always worked at 48khz for many years, pretty much for the same reasons you explain. Well explained as well!

[@RadiAsian]

Been working at 24bit 96k since 2004. I hear a significant difference switching down to 48k, and only a slight difference moving from 96k>192k. 192k may very well kill my ancient rig.

[@RadiAsian]

@Joe Smith In theory, you may very well be correct. If you get a chance...try it on the old RME FF800. We hear the difference over here in the studio.

[@RadiAsian]

@Joe Smith most times yes as we track at 24/96. So provided the source files are recorded in high quality then yes. I'm not here to debate as I'm snowed under with radio work and need to prep for tomorrows show. Each to their own

[@RadiAsian]

Its a simple enough test..play a high quality Ultra HD song and from RME's Fireface settings switch between the sample rates as the song is playing. You should hear it for yourself (unless I have a defective unit of course)

[@Xylume]

I use 3 audio interfaces so I don't have to switch between 44.1/48 or greater sample rates. I feel 24bit/48khz is perfect for videos and most applications. When I hire a client for professional vocal audio-work, I'm perfectly happy with receiving studio-grade 24bit/48khz audio files.

[@AmberAge]

I didn't understand a word you just said but it sounds very interesting lmao

[@donttalktome4696]

Hahaha this was me two years ago! Keep with the hobby

[@fluctura]

Quality is often the decimal precision (how many digits behind the comma are used to describe the wave). Higher quality means that the algo needs to calculate much more to come to better precision. The opposite of it, is cutting away precision. This is also, what a bitcrusher algo does.

[@TheGurner1]

Do you remember I was talking about this, that some people were hearing the intermodulation distortion from the convertors, as an improvement in the treble response (One day you might get a plugin to give it that vintage I.D. sound, though I hope not!)

[@heavymetalmixer91]

I preffer to use plugins with oversampling on 44.1/48 not only because my CPU doesn't need to struggle as much, but also because of compatibility.

[@michaelanderwald4179]

After lots of testing and listening, I'm absolutely content with recording and mixing at 44.1kHz. The only plugins I'll run with oversampling enabled are those that produce a *lot* of distortion.

[@Whiteseastudio]

Interesting... 44.1kHz is a difficult setup for me since AVB is not really stable/non existent in 44.1...

[@JAMStudiosIE]

Same. Unless I’m working for picture, 44.1khz. What about the debate about whether recording at 48khz but printing your mix to 44.1khz causes audio degradation - theoretically or audibly?

[@JAMStudiosIE]

@@Whiteseastudio what’s AVB?

[@SteveStockmalMusic]

@@JAMStudiosIE Audio Video Bridging (I actually just looked that up, and the first definition that came up was “already vaped bud“ LOL)

[@DerekPower]

44.1 is great for music alone. 48 is what is used for film/video production (probably because it synchronizes with a 24fps ... I can be corrected on this).

[@MichaelMoore-bx6st]

I write electronic music at 96khz because I can get lower latency on soft synths, at the cost of more CPU. For mixing purposes 48khz with plug-in oversampling like you said it's the way to go.

[@monsirto]

The vids are getting better! 96/48 here. 48 is "common ground" and "plays nicely with others".

[@delvenhamric1200]

Over the years I have found that upping the bit depth from 16 to 24, sounds better than doubling the sample rate. Also, converting from 48k to 44.1k sounds better than going the other way. I can also see the advantage of plugins, even mixing itself at higher sample rates. So, I record at 2448 and let my tools up sample if they sound better and live with it. But that's only my opinion!

[@EricOehler01]

The measurable improvement from going 16 to 24 is VERY noticeable. I always tell my clients that if they can improve one thing with their recording chain, it's that.

[@ScotBontrager]

I've been running my AD/DA at 48kHz for 2 years, after years of 96 and experimentation with 192. My Lexicon reverb's SPDIF only runs at 48kHz, which dictated the sample rate I used for several projects. I found that I honestly couldn't tell a difference. I quit worrying and have not changed since. With a few soft-synths (NI Reaktor) and distortion plugins (as you've discussed many times) I can hear a difference in very contrived circumstances. But I don't think these would be audible in an actual track.

[@johnthecreative]

if you have clients do you ever have issues converting a client's 44.1 khz project to 48 khz?

[@datutturugang666]

well in single tracks it’s different, depends on the mix, when i submit the final mix to the record labels i usually go 96khz 1411 kbps 24 bit, if requested 32 bit. the average music consumer doesn’t give a honk on immaculate audio quality, they care about their phones and storage, thus preferring aac or mp3 to lossless formats, live flac, alac or wav, mainly because the highest quality they gonna have is spotify most times, which tops out at 256 kbps. i did an experiment, with several people, making listen various file formats of the same song, through the same equipment, and to the untrained ear, it all sounds the same, slightly more detailed when playing lossless, with high bitrates.. plus bluetooth, is now the main portable headphone output, which tops out at max 320 kbps (LDAC excluded due to sony/lg proprietary software) through aptixhd, so yea

[@johnthecreative]

@@datutturugang666 thanks for so much info. I heard a pro say he prefers the sound of 48 to 96 because 96 actually has TOO MUCH headroom to respond to what he is trying to achieve. I have only ever worked in projects in 44 and I need to start up-sampling these to 48 for compatibility with gear I just bought. I wondered if anyone has hurdles or glitches in this conversion process, like plugins and samplers that don't do this reliably, etc. I use logic and I am crossing my fingers hoping it works okay. another issue with 96 is many plugins don't work in it, like older waves plugins.

[@datutturugang666]

@@johnthecreative i also use logic mainly, like 80%of the time, i usually have no issue in opening different bitrate files. it’s easier to open 48khz files in a 96khz project, because its exactly double, it gonna sound like a 48 file anyways, but i mean, who aside pros or audio enthusiasts even gonna care about the difference between a 44.1 vs a 48 file, if they enjoy a song they’ll listen the shit out of it even if it’s downloaded on a shitty mp3 downloaded from a sketchy website.. it’s just us visibly mentally unstable people who prefer comically large file sizes for a 5 minutes song lmaoao

[@johnthecreative]

@@datutturugang666 well thanks for the info. Like I said I only know about 44 but now i Have to learn about others in order to use new gear requirements of 48. Crossing my fingers that I can easily switch sample rate of on an entire project in Logic Pro.

[@antiHUMANDesigns]

OK, holy crap. I just took an old mixing session I did at 44.1kHz (because all those audio files are 44.1kHz) and just increased the project sample rate (in Reaper) to 192kHz, and holy crap, the dynamics improved like crazy...! I did nothing else, just increased the project sample rate to max. So, I did some quick testing. 1. I rendered a fresh 44.1kHz version, with the project at 44.1kHz. (Just to make sure it's this exact mix I'm listening to.) 2. Then I rendered a 192kHz version, simply setting the project to 192kHz. 3. Then I loaded this 192kHz render and put a brick-wall lowpass at around 20kHz (analyzer showed some action above that, so I wanted to manually remove it), and rendered it back to 44.1kHz. Versions 2 and 3 sound more or less the same, with very much improved dynamics, especially the snare. Simply allowing it to process in 192kHz, even after later downsampling back to 44.1kHz (native resolution of the audio files), made a huge, huge difference. Obviously, the true 192kHz version has a tiny bit better quality in reverbs and such, but that's not the point, here. The point is that the dynamics improved so much that the mix would need some adjustments to adapt to this change -- it's not a subtle difference at all! The snare now gives me a headache, as it seems to shoot out from the mix and punch my forehead... The upper range of the distorted guitars is also more defined in versions 2 and 3, though this is more subtle. (All renders were still native 24bit, and I used lossless FLAC format. Only the sample rates were changed as stated above.)

[@nomelocreo1979]

imposible ,si una grabacion es a 44.1 siempre sera de esa frecuencia, si lo procesas por un hardware y lo grabas a 192 es otra cosa ,si grabas una guitarra a 192 por ejemplo o 96 etc , si se notara ,aunque yo no noto nada ,ni en numeros ni en oido ,uso analogico , si tienes algo en 44.1 siempre lo sera aunque lo subas ,saludos

[@antiHUMANDesigns]

@@nomelocreo1979 It's about oversampling during processing.

[@nomelocreo1979]

@@antiHUMANDesigns imposible """No hice nada más, simplemente aumenté la frecuencia de muestreo del proyecto al máximo"" eso dices ,si lo procesas con pluginsde 192khz que apenas hay puede que mejore ,te habra pasado otra cosa ,uso ssl the bus+ y fusion hardware ,grabando a 192 y 44.1 el master ,no hay diferencia audible ni medible ,una eq solo llega hasta 20khz y el oido igual ,a no ser que seas un murcielago no notaras nada, seguramente ha cambiado el pannig law a triangular ,entonces sube el volumen ,la dinamica ,es la forma en que se trata el estereo, la gente lo confunde con calidad muchas veces ,saludos

[@antiHUMANDesigns]

@@nomelocreo1979 The point is that I must have had some plugins on that were causing aliasing, for example, which need to be oversampled to remove the aliasing. I don't even remember, this was an old comment.

[@nomelocreo1979]

@@antiHUMANDesigns ok entendido ahora ,saludos

[@The-Vay-AADS]

REAPER's Resample Mode - it has to do with the amount of samples the DAW uses to re-create a new waveform. I'd like to try to explain. Please correct me, if I'm wrong. Imagine a file being played at 0.5x speed without pitching it up 2x. So just "slow it down" in the most literal sense. In the DAW the file's samples are now twice as far apart. But the DAW still needs to play at YOUR OUTPUT's sample rate. So it has to re-sample the audio and create a new waveform on the fly. It has to create samples at your output's sample rate inbetween the original file's samples. How does it know how to set the amplitude, the height of the sample? The easiest method is "linear interpolation". The DAW just draws a line from the two surrounding samples and creates a sample at the line's amplitude in the middle of the line. Sounds fine on paper but in reality it sounds bad, because the waveform between the two samples might contain more complex frequency content, which will be reduced to "whatever is in the middle". It's a bit like aliasing. Lower frequencies get introduced. So anyway, any higher quality resample mode goes to where it needs to re-create a sample, looks at X samples AROUND that spot and THEN recreates a waveform that better matches the original. The more points you look at, the higher the CPU cost per new sample. That being said, I also can not hear any difference between "good" in REAPER's resample mode or anything higher. However, I had to deal with this problem above in game engines, because in order to run fast, they often use linear interpolation. Imagine a car engine going up in pitch as it's going faster, this is often done by dynamically pitching the file - resampling it. In my example my high frequencies of a bat squeek were lower and lower, weird sounding ones were added, after I put the file from the DAW into the game. So yeah. Cool to know audio nerd stuff. :)

[@trollkonto4313]

2:00 In fact, the reverse claim is also true. Some more or less ancient low- to mid-end interfaces only deliver good sounding monitoring when forced to operate at high SR. Everything now relies on delta-sigma and a proper digital filter algorithm should do the job, but some devices have a nasty roll-off in the audible part of the higher frequency range.

[@hannes1734]

If you got oversampling on the plugins where aliasing is a problem like Saturation, Distortion etc, 48 kHz is enough. You don't need any more.

[@MuzdokOfficial]

By just working in the box I like the precision and the less harsh highs (treble) in the rendered master.

[@SingularityMedia]

Also 24/48 in the mastering room and production room here.

[@laynehoward2870]

I run Studio One at 48kHz. The problem is that the rest of the world is typically 44.1kHz. When I leave Studio One and go to say, Soundcloud, I have to remember to change the Apollo to 44.1kHz. Maybe I'm just doing something stupid....it wouldn't be the first time.

[@nomelocreo1979]

44.1 forever and ever

[@lastdaysguitar]

At the end of the day, its somewhat system dependent - recording 96k sounds a bit better than 48k on my system, but I do not know if that is plug in related and I've not been able to tell the difference on other systems.

[@Beatsbasteln]

5:59 resample mode seems to be a way to select which interpolation method is used to convert from one to another sample rate. check out the wikipedia article of sinc filter for example to read about the best ones in this list. little spoiler: the more points the sinc filter has the steeper the anti-aliasing filter can be. btw i think it was a good solution to just tell your client to send you everything at 192khz, because didn't have to explain anything that way and just had to wait for new files. but essentially the real problem is probably much more that the client doesn't know how to use oversampling features and it would be better for the client in the long run to just learn how to do that.

[@articmobile]

Don't higher sample rates make for better time stretching?

[@melissabell585]

Theoretically yes, but the sample rates that we’re talking about and the ability to over sample during that process render the benefit more or less imperceptible. It’s not like video frame rates, where we’re talking about huge and extremely obvious differences.

[@articmobile]

@@melissabell585 I see. Thanks

[@jaco1368]

This guy is so cool and he has such great knowledge about music production, hardware and sh1t that I would be happy to have such a friend. it is like adding saturation into my life.

[@s1gne]

I usually use 192KHz 24 bit but in some games (Cyberpunk 2077 for instance) the audio clips and you have to use 48KHz (24 bit) to prevent that. But i'm no producer or musician, only an enthousiast.

[@noahaguilar8180]

Oversample does matter in any plugin, thats how we can hear that some daws or plugins sound "better", its oversampling (aside aliasing)

[@TerryFlatEarth]

Good info. Thank you very much!

[@phiprion]

People check the upper mids and high freqs, check saturation and stuff like cymbals, comparing 44/48 and 96. Try to tune vocals in 44/48 and 96 or pitch some samples. There's a big difference between those.

[@DnBLand81]

ok is a old video but work at 96khz sample rate is best option to avoid aliasing introduced from saturation plug in (especially) that in a full mix in the end will make it sounds so bad in hi frequencies

[@KrachWerke]

Completely agree with the workflow. I do most basic stuff and recording in 48. I even get a basic mix and levels done this way. Once I am happy with most things I set Reaper to 96 and stem ALL the tracks to 96. Then I do the last tone shaping and compression. At this point it does not tax the system too much as the main fx like amp sims etc have been applied. I have noticed that the mix and master is a lot more clean and defined as well as the instruments are more separated. I guess it just gives the daw more points to work with. I also try to make it clear that resampling higher does not add information it just ads more data points. I have a video about this here if anyone is interested: kzbin.info/www/bejne/n2GUq3-FfZqdb7M

[@famitory]

i work at 44.1 whenever possible and disable oversampling because I enjoy the sound of aliasing. I often also add bitcrushers on delays and reverbs to crunch them down to 22kHz for extra grit.

[@famitory]

@espoir inconscient happy hardcore, jungle, nu-metal, and generally anything that takes my fancy. i tend to blend together elements that aren't usually seen together like mixing new-age pads with bluegrass banjo or such.

[@bobsykes]

Another case similar to your example at the end, is if you receive (or create intentionally) a file that is clipped, you’ll discover that the clipping creates frequencies way above the expected cut off, and any processing at all, even a downward gain change, will produce distortion that can be 6 dB or more above the MSB. Converting files like that to higher sample rates sometimes alleviates this problem. Separately, the best thing to me about high sample rates at time of the initial AD conversion, is that there is FAR less phase shift in the audible band than what you get from the extremely sharp cutoff antialiasing filter required for 44.1 or 48. If you’re recording something like an acoustic piano, it can sound far more open, spacious, and natural due to having far less phase shift on all the harmonics that the piano produces. At least a compromise of using 96 for an application like that will be audible.

[@jeffrosen2010]

Interesting! I have never thought about the phase shift from the cutoff filter. If the filter is constant shouldn't it create a repeatable phase shift and therefore multiple sources would still be in line? very curious.

[@bobsykes]

@@jeffrosen2010 You're right, in that if you use the same converters for all tracks in a recording session, the phase shift from A to D antialiasing filters will be identical in each track. It is not a problem of track to track phase matching. What I hear and don't care for in low sample rate recordings of acoustic instraments is phase shift between the midrange fundamentals, like a key struck on a piano, and that note's higher harmonics, of which there are a lot, when converting the analog signal to digital.

[@rossdonald5026]

Food for thought!!!........... Thanks

[@rogercabo5545]

The most producers in the world using 44.1KHz and 24bit. But your video is very interesting Wytse, Thank you!

[@hanisiblini]

iLOVE Your videos a lot you are a source of knowledge Thank you

[@rolithesecond]

Using an ultrasonic microphone used to then slow the sound down for some cool effects needs a lot more than 48khz, as an example.

[@boulevardsound5137]

If you are working with something like pitch bending sine wave, 192 vs 48khz show massive differences. Since you're no longer limited to what your microphone can record.

[@underwoodstudio1821]

Running 384kHz here ......sounds fantastic ...! Also tracking DSD and really love that sound ...depends on the music of course...

[@acidcube6967]

Hi dude, what Sound Card & DAW so you use? I once saw an interview with Rupert Neve who when questioned about his opinion on this matter also mentioned that 384KHz bandwidth being required for DAC’s to faithfully reproduce the highest quality Analogue signals. Cheers Marlon

[@underwoodstudio1821]

@@acidcube6967 I work in a couple of DAW's mostly Pyramix. I use a DADAX32 and Merging stuff. I also have some custom gear. Lots of people work in DXD and 384kHz . We a just dont shout about it ...it does not matter at all if you are working in modern pop music and hip hop ect ...it only has meaning if your recording incredible instruments played well...then it has meaning . If its samples and files its really no piont at all......

[@acidcube6967]

@@underwoodstudio1821 🌟✨🤛 I totally agree great Musical performances should be captured accordingly, as one can never meaningfully upscale this loss thereafter. And in my opinion its a travesty to constrict harmonics denying humans have the ability to feel and sense. Most producers love to blindly preach the maths being exact and beyond question. When really the only unquestionable exactness of this Maths is that it perfectly demonstrates its formulaic ability to arrive at band limitation it had been derived from? And aside from these exuberant band widths I’m totally amazed that so many fellow musicians cannot simply conclude CD’s outputting 44.1 KHz is totally inadequate just by using their ears 😂 Anyhow, cheers! 〽️🕶

[@ShortCircuitProductions]

Hey Wytse. thanks for some always great content :) The alternative link you provided for the article seems to lead to a chinese unsecure website. Just so you know...

[@voag1344]

damn steven wilson thanks a lot! you got yourself a new sub :)

[@Mirandess1]

Domagoj Vida changed profession, from football to producer 😎 P.S. Tnx for video 🙂

[@aaroncline86]

24-bit @ 48 Here!! Only a few time I went to 88.2. It was all string and percussion. I believe I heard and felt the subtleties a hair bit more. ... You know because someone's going to be watching this back on their phone 😂

[@siriusfun]

96k makes good sense. Blu-Ray quality, labels request that for archiving, etc.

[@linasmak9199]

sound recordist use high sample rates with mics that have extended frequency response, there is huge difference when you design sounds from these recordings then, but for anything else i don't think there is a reason to go over 48khz at 24bit.

[@Beatsbasteln]

this sounds useful for recordings that are intended to be slowed down later, as this would bring supersonic frequencies down into the audible range

[@Buunshin_]

I would like to add that working at a higher sample rate allows for more creative options when it comes to resampling (from a producer perspective, not mixing). There is a great difference in the high frequency content when downpitching samples at different samplerates because for lower samplerates frequencies higher than 22.05khz or 24khz are being cut off by default. With higher samplerates the ultrasonic frequencies suddenly become audible, which for some people is interesting to investigate. This way you can literally hear things you could not hear before :)

[@adoremotion]

FINALLY! 44 or 48 for recording. Oversampling for mixing.

[@nowheretoshower]

If a subjective difference in sound quality is not useful to you, maybe a massive reduction in latency is! If your buffer stays the same number of samples, doubling the sample rate will reduce both your input and output latency by half! Quadrupling the sample rate will cut latency to one fourth! Can you imagine playing through amp sims that feel really good because the latency isn't getting in the way of your faster phrases?

[@Methar39]

1:34 But I thought that the sigma delta converters that we use for audio didn't need a steep analog antialiasing filter, since it's the digital FIR filter that does the "real" antialiasing...

[@Hahejo]

If my system supports it I'll gladly run my sessions at 192. But sadly not every one of us gets to play with the big toys. So 48 it is and oversampling whenever possible (for processing). I recorded a traditional instrument the other day and 192 did make an audible difference when compared to even 96 (I will not be discussing how it sounds since I haven't fully tested it and I might try your method to see if the converter can actually handle it well or not). Also, I would say if I had the privilege to mix completely analog (the computer would then be more or less like a tape machine, storing and editing data only) then 192 all the way all the time it is (the CPU wouldn't be hit as hard since it doesn't have to run any plug-ins)

[@offthisworld]

Why .... did it sound so much better importing the 192Khz files into the daw? Define 'better'? Could it be that the conversion from 192Khz to 48Khz suffered from 'bad settings' ? And how did it sound once the final project got exported in the 44.1Khz or 48Khz master. Also better? Or did 192Khz treat your ears in such a way neither 44.1 or 48Khz can ever do? Hmm. Questions.

[@lxm2600]

Spot on!

[@petter9078]

From mechatronics class at university - we learned that the optimal sample rate is slightly above the nyquist frequency. So, humans hear up to 20k, this means that 40k is the nyquist frequency, so 44.1k is fairly safe, however, 48k is even safer with a reasonable tradeoff in terms of performance for making sure signal is replicated as its supposed to..

[@ZeroDividesByYOU]

Well specifically with audio, the gentler the filter the better because of possible phase shifting and latency with designing filters. So 48 kHz is my preferred just to be extra shape

[@osm013]

if we try to think about raw AD DA sampling without any interpolation or filtering from the pure technical wiev 20k sine is sampled about 2times, so from sine you will get square with 20k frequency and somehow shifted phase. Also 10k sine will be sampled 4.8 times that will not keep original sine shape and phase.

[@droningbrightness3403]

been running at 88 for a couple years now but likely going back to 48 for the next projects and onward.

[@urssounds]

This is really informative! Of course Reaper's way of handling "internal oversampling" is quite unique I think! Is there a way to do this in Cubase??? You can allow audio import with multiple sample rates in a project. But I am not sure how Cubase then handles the whole oversampling, aliasing etc.

[@aloisjolliet2760]

Thanks for the great video! When you said: Some plugins have it and some doesn't because they don't need it - can you explain which type of processing needs it and which type doesn't? Like distortion, eq, ... ...

[@DarkBlackReaper]

Can you make a video about audio 32-bit float depth vs 24 bit?

[@DJayFreeDoo]

48khz allows for me to use the highest buffersize on my system. And im not able to hear aliasing from the synths like i can at 44.1khz. and if i need higher samplerate somewhere i just use a plugin with oversampling. If i was running a tracking studio i'd probably run 96khz.

[@TokyoSpeirs]

"ali-assing" 😂 but for real this is all super solid info.

[@okaybenji]

I had designed a guitar tone I really liked using a couple of different plug-ins. Then I decided to go from 44.1 to 192 in Logic. Suddenly my guitars sounded horrible. I did tests at 44, 48, 96 and 192. The difference between 44 and 48 was subtle, but overall it seemed the higher the sample rate, the worse the tone. I decided rather than redesigning my guitar sound, I would just stick to 44.1 for now. But still not sure why this is the case.

[@interru_io]

48khz everywhere except during non linear processing (everything which changes the frequency spectrum). But that is something you almost never have to worry about because nearly all plugins will oversample when it matters. Bitrate 32-bit float in the DAW for processing because of the advantages of float: No worrying about clipping because of near infinite headroom until you save it as 24-bit int. I also highly suspect that the problems you had with the client were because of 48khz vs 192khz. More likely that you got clipped integer files and later 32-bit float.

[@preciseaudioblog]

I go for 48 as well. Cheers!

[@DaveChimny]

4:50 By the way: That "Sandstorm" song is awesome! Love it! Props to Darude!

[@danielvernonlee6781]

Dude that’s not Sandstorm by Darude. That’s actually Darude, the song is Sandstorm.

[@Audio_Simon]

I design audio equipment and take a lot of measurements. One very important point is that hardware re-sampling (by an ASRC chip) is able to remove a lot of the distortion products introduced by a less than ideal clock signal. For example, if your DAC locks on to the AES/EBU input and detects the sample rate, this is using something called a PLL (phase locked loop) to generate a clock from the input signal. This can often lead to less than ideal jitter performance and introduce distortions (usually side-tones of a fundamental). If the DAC has an ASRC chip it can use it's own internal clock and resemble the input stream. The result will be cleaner. In fact it is much like quantizing a badly timed drum piece in midi. Software resembling is a mixed bag IME. Im not really sure why it should be any different than a hardware resampler, but if you resample a 1khz tone in Adobe Audition and look at the FFT plot, you can see that the various resampling setting availible make a huge difference to the distortions introduced. Maybe it just looks worse when resampling an ideal 'generated' wave rather than something captured by an ADC.

[@jeffrosen2010]

My understanding of this is that digital depictions of the waveform are in fact flawed and if you send that exact signal out to an analog oscilloscope you will still get a perfect sine wave. Although I could be misunderstanding what you're saying!

[@Audio_Simon]

@@jeffrosen2010 Yeah I wasn't meaning look at the waveform, but the FFT analysis. This shows the frequency content of the waveform. You will see the original tone plus spurious distortion products introduced by the resampling. I'll grab some examples.

[@Audio_Simon]

@@jeffrosen2010 Here are some examples. This is a particularly bad example of a DAC that has poor PLL recovery of the incoming clock stream VS. same DAC switched to resampling mode at 192KHz. Note significant reduction of harmonics (sinusoidal jitter) as well as noise floor (random jitter). imgur.com/40ZL7iZ

[@Audio_Simon]

@@jeffrosen2010 Here is the same waveform from the poor PLL DAC simply resampled in software to 192KHz. The result is the worst of both worlds. The original jitter is already encoded in to the waveform and can not be removed at this point. You also then add the artefacts of resampling. imgur.com/6TLyban So I'd suggest running software at native sample rate of your ADC, let your ADC provide clock to the system. Then use a DAC with hardware re-sampling when an analog output is needed.

[@Audio_Simon]

@@jeffrosen2010 One of my replies went missing! Here is a software generated 'perfect' waveform resampled in software to 192KHz. You can see the base of the tone becomes wider and the noise floor increases. imgur.com/FQI2Qej

[@littlederk652]

Are you going to review the new Cymatics plugin? I'm interested in hearing the opinion of an audio engineer instead of random bedroom producers

[@OneandOthermusic]

Its so hard sometimes to marry science with something as subjective to the ear as sound and music... on one hand you have a standard set of bullet points that seem to need to be checked with in industry, and yet also at the end of the day ya must bring something pleasing to the ear... I mean thats if ya want others to listen... So what I have determined for me is if it sounds cool and i like it , I will listen to it over and over again at any sample rate. In fact sample rate has very little to do with a sound that is pleasing to the ear... this is what is more the art choices in music I reckon... I enjoy these more theoretic videos and the opinions of diff topics . these are very interesting things to explore....always a thing to learn :)

[@cbrooks0905]

I’m confused. I work in logic, and if I record at 48 and then change my project settings to96 or 192 after the fact it changes the speed of the playback. Plus, I’m not really understanding how you can process at 192 when you recorded at 48. Where is the computer getting the extra data from if it wasn’t recorded in?

[@MichaelW.1980]

I currently use a MOTU M4 USB audio interface. Technically speaking, my signal becomes hugely more noisy, because above 28khz the signal continuously rises. 96kHz is already worse than 48kHz tho still below anything worrying in the ultrasonics, but at 192kHz it’s plainly ridiculous. My combination of preamps and ADC is the least noisy at 48kHz or below.

[@leoelias77]

i can mix longer when on 44.1.... sometimes i switch to 48 but i keep getting back to 44.

[@singerfromhell666]

What I noticed is that is seems to me that the Better mode when rendering is better for uploads on the web. I at least had issues when putting my projects on soundcoud when using the extreme mode

[@cheekoandtheman]

I record crazy modular synth patches at the highest possible sample rate as the modular synth makes noises well out of the range of human hearing and when I střech the audio and pitch it down I find sounds I wouldn't get at the normal sample rate

[@kevinlentz7604]

Stumbled across this just lately,track in 96,32 bit float mix down in 192❤