Been out of DAW-land for a bit. Getting a new interface (SSL2+) to replace my 10 year old Focusrite and configuring Ableton this weekend. In the past I’ve just used 48k because the differences seemed negligible. These days though, with the advancing of time and tech should we be recording at 96k- or higher? And are people using 32bit? Does it matter with the end result?
Not using any VSTi’s, mostly recording analog sound sources.
6 Likes
My understanding is it really will come down to how much fx processing you end up using in your mixing/mastering process. 96khz should give better results as long as your processing fx are designed to take advantage of it. 32 bit is really just a safety measure to prevent clipping.
3 Likes
Usb streaming instruments yes. My Waldorf Kyra sounds a little better on the low end for 96k sessions compared to 48k.
2 Likes
…48k in 24bit resolution is all fine in most cases…
96k makes a difference, once u really record a lot via really good pre amps and top notch mics…but it’s really cosmetic and only suitable for truu classical approaches…
so, depends on what u wanna do…
for endless rerecording, bounce in place action treating and recording electronic instruments “only” in most cases, 48k will be best and solid choice…
essential is bit resoltution…don’t dither down to 16 bit too early…32 bit and 64 bit is only for internal, behind the curtain oversampling actions…ur daw and vst’s comunicate in those wordwidth…but not ur ears…for audible results 24 bit is all it needs…
never forget ur enddeliverance…if ur producing for streaming and in best case for cd or even vinyl, anything beyond consumers edge of 44k is final goal anyways…
not to mention that in THIS case ur audience will listen to ur stuff on earbuds in some reduced format most of the time…
highest prestine soundexperience u can have these days, is in a cinema…that’s also nothing but 48k…
as u might know, beyond 48k all ur programs have to handle the same thing but with way more amount of cpu and read and write processing…so don’t bother if not really needed…
and apart from the ot, all elektron boxes work with iternal 48k, too…
1 Like
I record in 32bit (48khz) as my xone:96 audio interface supports that and it s the internal processing bit depth of ableton. When using 24bit files in ableton it upsamples it to 32 bit anyway. So why not record directly at that if it s available.
1 Like
The most obvious difference is when you transpose down a lot. When you transpose a 96kHz recording down by exactly one octave you get an 48kHz result (in regards to real, not interpolated data). So in this case you get the same level of details as when you would have recorded an already transposed down signal with 24/48.
It’s like a hires image: you can simply zoom in more without losing details.
Technically this makes absolutely no sense to me. The low end gets already massively oversampled with 48k (in regards to the frequencies involved) and has nothing to gain from even more oversampling.
4 Likes
the information changes over the years. there’s an infamous white paper by Lavry about why 96khz is a waste of time. it’s almost 10 years old now though.
http://www.lavryengineering.com/pdfs/lavry-white-paper-the_optimal_sample_rate_for_quality_audio.pdf
basically, the clocks that run your AD hardware aren’t always very good so recording at higher sampling rates just gives them more opportunity to present clocking errors which essentially puts an error where a 1 or 0 should be. this means they create noise.
that being said, clocks and AD hardware in general is better these days. running at 96khz when using a daw and softsynths and plug in effects can be beneficial in some ways w/better sounding high resolution things. it’s why some vst processors have oversampling when in use and rendering.
i think high resolution recording really makes a difference when recording acoustic instruments like piano and stuff. all that magic captures well with nice mics and nice mic pres etc.
will anyone know the difference? probably not the end listener in most cases. with people mostly listening via streaming and mp3s these days only obsessives like us and the odd audiophile person will take some pride in higher resolution recordings etc.
i have recorded at 24/44 for ages. I few times i worked at 24/96 but only for the helluv-it.
that being said… if i had a really fast computer, doing a lot of field recordings with nice mics and recorders, doing lot’s of pitch shifting etc… i’d probably run at 24/96 if the system could handle it.
but if i had a consumer grade interface and typical computer i wouldn’t see any benefit to anything over 24/44.
edit: here’s a GS thread that lavry comments on regarding the paper.
5 Likes
I heard that Noisia records at 32/48.
yeah I usually record 32 also… I figure better safe than sorry, it basically just gives you headroom above normal clipping value, so if you do clip you can always pull back the gain and have an unclipped recording. It doesn’t give you better sound quality to my knowledge though.
Also when I play higher notes it’s brighter but in a good way in 96k. Like you said this maybe just the over sampling compensating for what I am hearing.
Other than storage space and computer resources there is no reason not to use 32 bit.
However, 32 bit recording is a bit hyperbolic for most uses. AD/DA performance is ultimately limited by the performance of the analog circuitry, where 24 bit translates to 144dB signal to noise which is a lot, and in practice, a much wider range than any audio content is likely to use. 32bit/floating point is used internally by DAWs/Pluggins but the benefit is only in the ability to process signal that has already been digitized without digital clipping and artifacts. Basically it insures that once you got a clean signal into Ableton the program will not clip internally. The program can process in 32 bit float even if the recording was done at 24 bit.
When the Xone:96 plays back the output from Ableton, the DA is likely 24 bit.
This is a good explainer I came across:
4 Likes
…actually, i hear the first time of people stating they record in 32 bits…
as far as i know, there is no such thing…
32 and 64 bit of wordlength resolution is taking place all the time within daw/plugin processing internally…“behind the curtain” as i like to call it…for realtime but background oversampling/anti aliasing purposes ONLY…to underline that’s nothing u could print into any kind of reasonable and existing audiofile as we know it…
i mean, who would need such a range of dynamics in any realworld auditioning situation anyways…
stop the myst and ferry telling…if u swear for high res operating audio…use 88k…
then u need at least no dither efforts to end up in 44 again…where u end up, no matter what…
and no matter how loud neil young once was promoting this yellow 96k hi rez forever and only ipod music player device thing a few years back…boah, it sounds soooo clear…u should listen to the 192k version…that’s crystal CLEANed out for real man, i can tell…
but ok…that’s just me…who can hardly tell any difference between a 320mp3 and a proper wav file on most listening devices out there…
truu hifi enthusiasts listen to vinyl ( that ends far below that official dead end range of that nyquist 20thousand hz boarder) via passive amplifying and passive speakers in passive rooms…
while most of the rest enjoy music with their earbuds via bluetooth…
and sometimes they hold their breath when experiencing 48k/24bit in full dynamic range mixes while going to see movies…
This seems like a substantial advantage to me:
My reasoning is to keep it all at 32bit 48khz from recording to export so no conversions are required. The masterer gets a 32bit file and does the conversions so i can t mess it up.
(Re: vinyl. I have my first one coming out and boy am i disappointed in how much of the high end sizzle is gone… vinyl is nice as a prestige thingy but i ll be playing my digital files instead)
1 Like
The 32 bit floating point processing is a substantial advantage indeed.
Using it as a file format for consistency also seems perfectly logical.
Regarding the advantages of 32 bit recording and playback…the finished track will sound no better than 24 bit fixed-point on Spotify, CD or even directly from your DAW
Congrats for your first vinyl. When you are 85 years old and your ears have also lost all the high-end sizzle, your vinyl will preserve all of its prestige-thingynes and your grandkids will think they have the coolest grandpa.
2 Likes
real-world difference, both in hardware & software: DSP/CPU load.
so i prefer 48.
Unless techno is regarded as an anachronism by then 
2 Likes
those 32bit files will sound great streaming over spotify through someone’s $29 blue tooth speakers.
it’s a shame all these improvements and amazing technology have no benefit for the average listener other than convenience.
other than our niche of audio workers and producers who appreciate it it seems lost on the masses.
as tech improves and computers get faster perhaps the bandwidth for streaming will improve to include higher resolution options like 24/96 etc.
there’s no rules in all this except to make the best music you can. a good song can survive anything including shitty streaming codecs.
taking it seriously and making the sound best as possible at the source will pay off eventually and immediately for the person making it who gets to enjoy it in all its glory at home or on a large PA etc.
1 Like
Well, besides hobbyist techno producer, i am also a professional audiovisual artist and i run a stage tech rental and production company (live sound, events, theater, etc). So unlike many on the forum i do make my living from electronic sound and media, and have good reason to get the most out of the tools at hand.
1 Like