96K is too sterile and digital. 44.1 gives me that 80's warmth.

 

The clue is in the sample rate..

 

That is incorrect. The plugin upsamples the audio input to a higher bitrate than the session, then uses that higher bitrate audio internally during it's processing, and then steps it back down on the output side of the plugin to match the bitrate of your DAW session again.

 

Don't mess up sample rate, bit depth and bit rate, otherwise it might get confusing.

The sample rate is 44.1 / 96 kHz or whatever, the bit depth (or precision) is mostly 16 / 24 bit integer for storing audio files and 32 / 64 bit float while processing.

The bit rate is the overall data rate of the audio stream, for example 1411200 bit/sec for a 44.1 kHz / 16 bit / stereo signal.

 

oh, true. i'm not sure how that would confuse anyone; but it is more accurate to use bit depth in relation to data; rather than sample rate, Nyquist and the rest of it which provides the number of samples per second. I was more focused on not causing confusion about the outcome of Oversampling actually having a benefit always, because that is not true in 100% examples.

 

This is slightly confusing. Oversampling is with regards to processing your signal at a higher multiple to your current project sample rate (i.e for anyone reading who doesn't know, that's what 2x 4x means in your plugins), hence OverSAMPLING. Bit depth, on the other hand, pertains to the number of bits in a sample, impacting dynamic range. These are distinct concepts, not one taking precedence over the other – just different technical definitions. But yeah, I agree not everything needs to be oversampled as aliasing is only bad if it bothers you, sometimes it even sounds good.

 

it shouldn't be. The comment mine was in reference to was "you cannot oversample in a session which is 44.1khz sample rate" when obviously, you can. The sample rate and bit depth are not intrinsically linked to one another. The digital word can be shorter or longer while the sample rate can maintain the same value, be increased, or lowered. If the word length at higher bit depth than what the input side of the plugin can accommodate arrives, the extra integer is digitally ignored, resulting in what we call Truncation. Conversely, enters Dither.

it's simply being pedantic in this situation for the sake of pointing out one wrong word in a sentence. That is actually the whole premise of most of these threads. In 2023 we're going to have a 6 page long thread about oversampling, because it is that time of the month again. A total circle jerk. Good luck with that.

Maybe we can all discuss Nyquist theorem again. That might be productive.

 

Last edited: Sep 28, 2023

That seems to be a good practical reason. I noticed that my audio chip on my motherboard defaults is 48k, among other devices.
I like competition and creativity, but I like standardized things, too.
Since it is not a technical difficulty these days, why not 48k?

I want it to sound as good as possible, so 48k...49k..whatever it takes.
https://tenor.com/view/mr-mom-220-volts-whatever-it-takes-gif-3478656

 

I don't burn CD-Audio anymore, just meaning 44.1kHz is the CD-Audio standard. I use 44.1kHz for final FLAC encoding. Also, I avoid as possible MP3 format when I buy an album, or single title. 48kHz is mainly used for audio in video.

 

Because the extra 1950 hz accurately represented in the higher sample rate recording is somehow detrimental to your signal to noise ratio. And the Easter Bunny is now here early, to help with the math.

 

Last edited: Sep 28, 2023

Who told you that nonsense?

 

Additionally, in which way is 32x OS relevant in a comparison to the SR? Do you use 1.4MHz?
I mean, the tires of my car can't stand 400km/h but ...

 

Because when you use the highest oversampling mode in Saturn (you have only 2 to choose from), the signal is internally upsampled to 1.4MHz, regardless of whether your project is in 44.1kHz, 88.2kHz or 176.4kHz. However, the higher your project sample rate, the lower the artifacts caused by the oversampling process. This basically means that the more processors you use that internally oversample and antialiase (i.e. clippers, amps, preamps, compressors, pitch, vocoder, partly modulation, delay and reverb, etc), the more you potentially benefit from a higher project sample rate.

BTW This problem is not due to FF Saturn. Saturn is in the midfield as far as artifacts are concerned. There are plugins that handle this better, but also many that do far worse. I used Saturn for this example because it is a popular saturator, not because it has bad oversampling.

 

Last edited: Sep 28, 2023

I understand your point. But if the plugin in question upsamples to a static 1.4MHz in either example; wouldn't the delta be a higher value for the project with the lower sample rate? The amount of possible improvement would be lower.

 

It's complicated and I think my answer is a bit beyond the scope of this thread. When I have time, I will create a separate thread on this topic. Then I can go into the problem in more detail, show more examples on specific case scenarios, and put my findings up for open discussion. You are then very welcome to attack my arguments if you want to. Okay?

 

Every time i see someone saying that oversamples a 192khz session i want to purchase a Akai Sampler to make it my DAW

 

I honestly can't remember I believe it was here that someone said if you oversample while using 44.1 that it will either not do anything or make the audio worse...which I thought was kinda odd because I would try it with apogee free limiter/ clipper which is a amazing plugin by the way and it would sound better sometimes .. not to mention it didn't faze my cpu .. I know that's normally not the case .. but still cool nonetheless, also Redrock plugins run really well with oversampling on.

 

If memory serves, there's a standards organization called the Moving Picture Experts Group (MPEG - established jointly by ISO and IEC) that sets standards for video (and by extension, audio) encoding, that adapted a French-Dutch-German algorithm called MUSICAM (this would have been in 1989 - when DAB was the new exciting thing in Broadcast) into the the MPEG-1 Audio Layer II codec, aka ISO/IEC 11172-3.

All audio linked to video, DVD, TV (commercials), etc - was subsequently set at the MPEG-1 Audio Layer II codec specs, i.e., 48kHz.

MUSICAM stood for Masking pattern Universal Sub-band Integrated Coding And Multiplexing - a fancy way to say: an algorithm that uses psychoacoustic models (namely but not only: masking) to perform data or bit reduction.

Layer II was commonly spoken to as MPEG-2, hence the little brother's moniker of mp3 when said codec was released into the wild, in 1991, alongside its senior brethren. MP3 was ridiculously unprofessional and strident sounding (never mind the bass range) and was never considered a serious codec, unlike MPEG-2 which was widely adopted by broadcasters all over the world (from ABC to YLE and every one in between - public or private).

So all the music or interviews some of you older ladies and gents have been listening to on radio since the early to mid 90s have been encoded on MPEG-2 - and still are (most of them, to this day, including KCRW in Santa Monica, but also the BBC, CBC, DR, DW, SR, YLE, etc).

And that's just Audio. Now, on to Television: same as above. DVD: same as above (all DVDs). BluRays were PCM I think (they've got the storage space).

My point being : 48kHz audio has nothing to do with copyright or the greedy film studios (or RIAA), and everything to do with technical standards (if you've traveled the world with electronics you know how each plug is different? that's because there wasn't an international body setting standards).

Back to MPEG and 48kHz... the number is important because an MPEG block (the smallest size sample - which cannot be cut) is exactly 24ms. Each irreducible unit of audio is 24ms. If you need to cut 1-23ms, tough luck, say goodbye to all 24ms. Therefore, it made sense to film/video editors (with their little Texas Instruments or Casio calculators handy) to be able to multiply or divide in MPEG blocks, or audio chunks, or multiples of 24. And what's 24 x 2?

 

With the state of music production and people's ears these days, after 20+ years of loudness war, you can record your music at 32kHz/16bit, practically Roland W'30/S-330 quality, and nobody would notice. My RME card and TC SK48 are dead set on 48kHz for everything. I don't even think abut it anymore. I am a bit tempted to use 96kHz at times, but then I just say "naah, nobody cares, nobody can hear it, and plugins will use 2x CPU, so it's not worth it." I'm not saying 96kHz doesn't give you better quality audio, just that nobody will be able to tell the difference. Just relax and make music. That makes a difference.

 

It's a nice number, easier to remember.