Going 96k (after a gear upgrade; not sure when)

[mhschmieder]

OK, I've made a definite decision to go all-96k, for any new projects that get started from this point out (I'm in the middle of several projects right now, for myself and others, but will leave those be).

The bottom line is that I am tired of all the hassles in going back and forth between sample rates, such as when using stuff like Sample Modeling's Sax Brothers libraries that require 88.2 kHz or 96 kHz to work properly 9without glitches, noise, distortion, etc.). It's a real PITA from a workflow point of view, doing SRC in RX2 Advanced, eliminating the weird "empty tail" problem in BIAS Peak Pro, restarting DP and setting sample rate everywhere (including hardware interface, master vs. autosync etc.) and in the right order so that DP doesn't get confused when I try to trim tracks that were SRC-ed elsewhere.

Mostly, I am also finding more and more that 96 kHz makes a bigger difference for ITB than for instruments brought in through A/D conversion (except stuff with long decay tails, such as celtic harp and cymbals, where 96 kHz has an audible impact). For synths, this is due to less aliasing. For stuff like Sax Brothers, the algorithm depends on oversampling. And disc space is cheap these days, so it seems better to consolidate at 96 kHz and just use top-notch SRC for the final master.

Latency is lower at higher sample rates. Thus, overdubs (especially bass) can sound more organic.

Also, most gear operates at its highest sample rate and uses SRC to get to the desired sample rate anyway. Well, maybe not most, as many vendors don't say, but of those who do say, most seem to use SRC internally. Thus, we are subject to the quality of their SRC (usually lower than iZotope's excellent SRC algorithms).

Anyway, now I need to make sure my gear can support it. It can't currently as I depend on S/PDIF to connect from my Mytek AD96 to my Fireface 800. Upgrading to Fireface UFX would give me AES/EBU, so then I'd have to check whether I can go 96 kHz between the OctoPre II and the FF800 with a full eight channel count (which I need -- at least during ghost tracking). I can figure this out quickly once I set aside time to study the manuals. Hopefully it is doable.

Well, I just thought I'd bring this up, because previously we haven't really discussed 96 kHz from the point of view of practicality vs. sheer quality (which many claim they don't perceive). It also seems better for future-proofing as new media comes into play.

[labman]

So MH, are you still doing 96k whole way thru your process? Please update us if you would be so kind.

[mhschmieder]

Well, it depends on whether I'm the person who starts the session or not.

I am happy to report, at least, that I have been successful in getting more people in the SF Bay Area to switch from 44.1k to 48k as their basic rate during tracking.

The 96k rate is not as important for most rock contexts -- especially further down the mix -- but I notice a big difference in 48k vs. 44.1k, even with great converters. My clients do too; even if they're not told in advance and are given blind tests. The music breathes better and sounds more organic. I know in the bigger scheme of things, it's subtle, but it help people connect to the music better, and that means more than anything technical related to the rate.

I am indeed sticking with 96k for projects I originate -- ironically especially in the case of MIDI or VI projects, because at the tracking stage one gets less aliasing issues as most synth algorithms don't really solve that problem until the sample rate has been bumped up a bit (some say even 192k makes a difference, whereas up until recently I had thought that was more for further reducing latency and also improving the exactness of film cues).

The metal project I am currently recorded started away from me initially, initiated surprisingly with the guitar parts (it's been a HUGE challenge to get the other musicians to lock in and keep it sounding live, together, and energetic, as though all were in the room at the same time, but we're finally there and everyone became a better musician and a better listener for it and it also helped with their live act). I inherited it at 48k, but it's not a problem as there isn't much top kit so the long sustains of cymbals and acoustic guitars aren't part of this music.

[Sean Kenny]

Hi MH

Not sure if this helps any http://www.motunation.com/forum/viewtop ... =4&t=57942

I understand why people don't think they can hear a difference, cos for me the real big difference is when you star to really process the audio. Eg heavy vocal tuning, which can yield very subtle high frequency graining on some vocals @ 48K, no such problems at 96k. I also found when generally when you have a pretty heavy session with loads of plugs things just sound clearer.

Cheers

Sean

[mhschmieder]

Sean,

Since the industry and academic consensus is that it isn't possible to tell the difference, I've tried to accept that it might be a placebo affect or something, but here's something interesting: I can hear up to 22 kHz, and have been able to do so my entire life (as can my dad). It's not too well-known that the "20 Hz to 20 kHz" range is an average, but also "easy to remember".

When I was six years old, I remember being in pain when anyone would use a dog whistle, and couldn't understand what the "extra" sound was that humans can't hear, not realizing it was the sound I could hear that most others couldn't.

When CD's first came out in the 80's I couldn't stand to go into record stores because there was some sort of pilot pitch that made me feel like I was being tortured. People in the stores thought I was mentally ill, so I stopped shopping in order not to freak people out. They thought I was "hearing voices". I think those pitches were probably closer to 18 kHz though, but most people can't even hear that.

[Sean Kenny]

I think those pitches were probably closer to 18 kHz though, but most people can't even hear that.

Glad it's 'most people', otherwise I might be out of a job

[EMRR]

Phase linearity through multiple conversion cycles has to make a difference at lower rates. A single ADDA cycle is already close to 60 degrees phase shift near cut-off frequency with several MOTU converters, and I doubt anyone else gets much of a different result. Mix analog and we then have 120 degree shift close to cutoff, with the 60 degree point down around 12K. Switch to 88.2K, and it's under 10 degrees for one cycle.

I've always heard an obvious improvement at higher rates with the converters I've used (2408mk3/24io/HD24XR) with any kind of music. When I drop back to lower rates the top always becomes less clear, and I doubt I still hear 20k like I used to.

[mhschmieder]

Just remember, the so-called falloff of high frequencies with age is yet another over-simplification so as to not intimidate people with complexity or hard-to-remember details.

The truth is far stranger: we lose specific frequencies as we age. There are tests, available on the internet, if anyone is interested (especially people whose livelihoods may depend on being aware of their particular hearing loss specifics).

[stubbsonic]

I use a $5 app called AudioTest which is useful for quite a few tasks. It will give a readout of the frequencies as it sweeps so you can get the good/bad news about the sound vanishing at some specific point.

I've also used it to get a rough read on the freq specs of some headphones I may be trying out.

http://www.katsurashareware.com/pgs/audiotest.html