Preferred Sample Rate for recording.

Phil O · Post by **Phil O** » Mon Dec 10, 2018 7:08 am

My question to those on the "I like 88.2/96K" side of the isle is:
(it's a real question - I'd like to know the answer to this)

Do you hear the improvement in audio that is recorded and mixed at 88.2/96K and then reduced to 44.1/48K in mastering or is it just in audio that remains at the higher rate? The reason I ask is that virtually everything I deliver is at 44.1K or 48K. So if I record at 88.2 or 96 I'm gonna have to SRC it down anyway.

Which brings me to the next point. Let's say there IS a noticeable difference between the higher and lower rates. I record a client's CD at a zillionK. It sounds great in the mixing sessions (we can hear the flea on the neighbor's dog farting, cuz it was picked up by my Earthworks microphones). We're almost done and in the final mastering I bring it down to 44.1K. The stereo field collapses, there are phase issues, the highs are gone and we can no longer hear the flea. Instant disappointment, right?...Right? I don't know - I'm just playing devil's advocate here. Help me out guys.

Phil

[edit]
Just for the record, I usually record at 44.1K/32bit, bwav for CD or 48K/32bit, bwav for those weird video people. If I don't anticipate any destructive editing of the soundbites, I'll use the 24bit depth instead, but lately I just keep it a 32 bits fp (because that's my mastering engineer's preference). Probably 90% of my work does not involve VIs.

EMRR · Post by **EMRR** » Mon Dec 10, 2018 8:26 am

The majority of the improvement remains after the final sample rate reduction. Heck, a lot of mastering guys are gonna upsample your 44K1/48K audio to a higher rate for processing, and downsample it again at the end. Many are gonna archive a high rate non-limited version of the master, in case you decide to do vinyl or a 'hi-res' release done the line. Supposedly many plugs and mix processing paths sound better at high rates. Most digital EQ's emulate the analog phase shifts of analog EQ's. Move the filters higher and out of the way, all the tools have more room to work within the audible band we perceive, keep the filters low, and it intrudes on those bands.

Phil O · Post by **Phil O** » Mon Dec 10, 2018 9:44 am

Thanks for the reply, Doug. I've got a three piece acoustic band coming in sometime in Jan or Feb. That'll be a low track count and relatively low CPU load. I think I'm going to give a higher rate a shot and see if I can hear a difference. Can't hurt.

Phil

FMiguelez · Post by **FMiguelez** » Mon Dec 10, 2018 10:29 am

I'm curious to know whether I have a misconception about this, or 96k+ lovers just prefer to ignore my questions, for some reason...

So I trust you guys have equipment that not only allows you to be able to hear these ultra-high frequencies, but at least it doesn't make music sound worse?
And I mean in your studio and at your house.

So, can it handle it, according to the specs? If not, can that "difference" you hear just be the expected distortion? How would you know?

Because I understand that's exactly what would happen when attempting to play 96 or 192KHz stuff on non-specialised-equipment-for-this-purpose... The converters, the amplifier and the speakers would distort and sound really awful. Worse than 48KHz!

So, do you have all that equipment, or how do you reconcile that?

Same question for those who say they can hear a difference between 32bFP and 24 bits.... Is your DAC giving you that bit depth? Or is it more like 20 bits?

If so, how can you say you hear it...

How do you know you are actually hearing what you think you are, let alone not getting things to sound worse with regular equipment????

Can you tell us what equipment you use for being able to record, produce, mix and master 96 and 192KHz stuff while still taking advantage of it and not making things sound worse?
The FULL chain, please...

Me no comprende... Just curious...

HCMarkus · Post by **HCMarkus** » Mon Dec 10, 2018 11:37 am

Hey FM... I think the argument in favor of higher bit rate recording is less about frequency response than it is about high frequency phase shift in EQ and DAC filtering. Higher bit rates lessen the phase shift curve slope. As EMRR has noted, newer converters have less of an issue in this regard than older ones, but the phase shift is still present. Whether and how the difference is perceived by human beings is another question.

And if a mastering lab wants a 32 bit float file, it would be to prevent any bit depth truncation in advance of its doing exactly that (in a very controlled manner, of course.) By my understanding, no DAC really gives us 24 bits; thanks to thermal noise, a bare wire passing an analog signal has less dynamic range than a 24 bit audio file.

The above stated, your questions are well-taken. However, because I fall in the camp of those who generally record at 44.1 or 48k, I cannot answer them.

PS: Although this thread will someday grind (albeit, nearly silently with nominal distortion) to a halt, we can expect the issue to rise again, like the Phoenix, from its ashes within less than 24 months.

Happy Holidays and New Year to all, no matter your preferred bit rate or depth!

mhschmieder · Post by **mhschmieder** » Mon Dec 10, 2018 12:17 pm

Good summary, HC.

The differences I notice aren't necessarily specific to anything innately superior about 96k also, similar to several of the things HC said.

This is why our staff physicist and I got into such a disagreement 16 years ago when he was hired.

I let it go (like I do most things, and sometimes regret it as I'm usually proven right over time), and we get along fine. But as someone who works in manufacturing, I have my feet firmly planted on the ground and tend to be disinterested in theory and only focused on reality.

As I said, I only use 96k on projects that are sparse and intimate with a focus on delicate string work (for example) -- something I haven't done much of since the 80's (tape days anyway) -- and projects that start as MIDI and will primarily use Virtual Instruments and libraries. There's no uniformity in how they're implemented or recorded (if sampled vs. modeled), but synthesis based stuff universally exhibits fewer artifacts at higher rates.

As a mathematician, it befuddles me that people aren't seeing the difference between these two scenarios. Every time I see the argument again about why high sample rates aren't needed, it tends to go back to the ideal and theoretical. Once reality catches up with those ideals and there's no longer any difference between what can be delivered in software or hardware, and what shows up in textbooks, then maybe we can write off the high sample rates.

EMRR · Post by **EMRR** » Wed Dec 12, 2018 12:21 pm

FMiguelez wrote:I'm curious to know whether I have a misconception about this, or 96k+ lovers just prefer to ignore my questions, for some reason...

Can you tell us what equipment you use for being able to record, produce, mix and master 96 and 192KHz stuff while still taking advantage of it and not making things sound worse?
The FULL chain, please...

Me no comprende... Just curious...

There's no special equipment required to take advantage of the improvement, it shows up most places and the extended bandwidth does not cause 'bandwidth overload'. I can only imagine it would be worse if you had loud sounds at 40kHz or something, that folded back as IMD, but the stuff up there is much lower in level, the energy content is very low by nature. Bass is what eats all the power energy in an amplifier. Additional distortion you might notice could only be IMD since the 2nd harmonic of 25kHz is 50kHz, and can't be heard, and IMD would require an unusually loud 50kHz signal, which just doesn't happen. I record with 1940's tube preamps and ribbons pretty frequently, one of my home systems is 1930's tube amps into Klipsch La Scalas, and the improvement is there just as much as it is on my JBL LSR28P and 705P studio monitors, just as much as if I'm using a Sennheiser MKH800 into a modern wide bandwidth transformerless preamp like a Sytek. I have an Audio Precision test set for doing equipment assessments and Spectrafoo Complete open in another window at all times, so I've been looking at frequency and phase response at varying sample rates for a long time.

With the 2408mk3 you see the noise shaped dither peak forced into the top of the available bandwidth for each sample rate, so it's in the audible band at 44K1 and 48K, and above 32kHz at 88K2 and 96K. This is an implementation aspect of the overall question. I don't have a screen capture available for the 16A at the moment, but recall that the dither peak gets smaller at lower rates, being simply cut off by the lower rates, and greater in size as rate goes up.

Bit depth; easily swamped/masked by power supply noise in a converter. The 16A is vastly quieter than the 24i/o and 2408mk3 because of the power supply implementation. AVB communication with the computer shows up as a 32 bit path so the AD noise floor meters (RMS) as -137dBFS to -154dBFS in the audible band. 2408mk3 has multiple AC hum spikes above -116dBFS.

I can't recall, have you tried it using audio you've captured yourself through AD converters?

cuttime · Post by **cuttime** » Wed Dec 12, 2018 12:42 pm

mhschmieder wrote:and projects that start as MIDI and will primarily use Virtual Instruments and libraries. There's no uniformity in how they're implemented or recorded (if sampled vs. modeled), but synthesis based stuff universally exhibits fewer artifacts at higher rates.

You are referring to the recorded audio, correct? I was wondering if the project sample rate affects the internal sample rate of a VI AU or VST? Or is this something that is always fixed by the coding of the software plug?

MIDI Life Crisis · Post by **MIDI Life Crisis** » Wed Dec 12, 2018 4:12 pm

C’mon kids. My head was already exploding score all this?

HCMarkus · Post by **HCMarkus** » Wed Dec 12, 2018 5:20 pm

MIDI Life Crisis wrote:C’mon kids. My head was already exploding score all this?

Won't that apparatus stuck in your ears hold your head together Dr. MLC?

MIDI Life Crisis · Post by **MIDI Life Crisis** » Wed Dec 12, 2018 7:24 pm

Musta posted from the iPhone. What does "score all this" even mean? Duh! Of course, it meant before all this. As for the stethoscope, it's actually a pipe. I had my ear drums removed so I could smoke and have it go directly to my brain. So far it's woking. I mean WORKING (I meant ahhh...)

HCMarkus · Post by **HCMarkus** » Wed Dec 12, 2018 10:19 pm

MIDI Life Crisis wrote:Musta posted from the iPhone. What does "score all this" even mean? Duh! Of course, it meant before all this. As for the stethoscope, it's actually a pipe. I had my ear drums removed so I could smoke and have it go directly to my brain. So far it's woking. I mean WORKING (I meant ahhh...)

Which "brain" are you talking about?

https://arstechnica.com/science/2018/10 ... pe-liquid/

MOTUnation.com

Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.

Re: Preferred Sample Rate for recording.