Fixed 48k Sample Rate Impedes Recording Studio Integrations

Status
Not open for further replies.
I literally spent months complaining about the 48kHz locked sample rate and spent even more time and a lot of money trying to remain in the digital realm. What I found was that the only time that digital was necessary for me was with recording the dry signal for reamping. It's too low for analog; even balanced analog was too close to the noise floor. Remaining in digital for dry and reamping is pristine. Recording wet signals with a half decent DAC (balanced) is going to sound great for nearly everyone.

Good quality outboard SRC's start at about a grand. I shopped around for a long time and tried several. I'm pretty sure that they aren't a single chip. That's what you have inside of most devices that allow multiple sample rates onboard. I've owned some of them, done testing and there is a difference in sound quality between sample rates with those devices. I am pretty convinced that it isn't because 96kHz inherently sounds better, but because the native sample rate of those devices is 96kHz and then it's down converted on the fly to a lower sample rate using a $1-5 chip. And this notion that you can't get good quality at lower sample rates like 48kHz is wrong. I've recorded at multiple sample rates using a high end DAC and there is no difference. The real cost is in the filters of the DAC. It is expensive to have quality components that have a sharp rolloff at the higher frequencies, but it is very cheap to just throw a faster clock in there and oversample it and then filter it. If you want good quality real time SRC it is costly. Even then software conversion is still better and a hell of a lot cheaper.

The other issue is that it may add latency to the signal path. So we'd have many of us who have hundreds to even thousands invested in IR's that would be rendered useless, lesser audio quality to integrate real time SRC cost effectively and potentially higher latency.

In the end it would be nice to have multiple sample rates, but at this point I've come up with a workflow to get around it and it isn't going to bother me. Bottom line is that knowing FAS I am pretty sure that if there was a viable path to multiple sample rates that it would have been done.

Wow, this is the most informed posted I have ever read on the subject. I had wondered if it was a hardware restraint. It makes a lot more sense now why Fractal would be resistant to using multiple sample rates, both because of additional cost and potential compromises in quality.

The sample rate issue doesn't affect me, but I still find it interesting.
 
I think Fractal has the right balance here.
One of the things I appreciate about FAS is their focus on function first. I would rather leave conversion rate out of the box, if it leaves room for more amps, pedals, patching etc.. There are dedicated boxes and DAW that can convert sample rates as needed. I don't like the typical Swiss army knife approach by most other companies to music electronics, because you end up with watered down products with overlapping/redundant capabilities, and it's going to start overlapping, and you don't get the value - just lots of 'do all' boxes that don't do anything really well. So I'm happy to get a separate rate converter box or use a DAW that converts on the fly for projects that are not 48KHz.
 
I think Fractal has the right balance here.
One of the things I appreciate about FAS is their focus on function first. I would rather leave conversion rate out of the box, if it leaves room for more amps, pedals, patching etc.. There are dedicated boxes and DAW that can convert sample rates as needed. I don't like the typical Swiss army knife approach by most other companies to music electronics, because you end up with watered down products with overlapping/redundant capabilities, and it's going to start overlapping, and you don't get the value - just lots of 'do all' boxes that don't do anything really well. So I'm happy to get a separate rate converter box or use a DAW that converts on the fly for projects that are not 48KHz.

Following that approach consequently would mean to stick with an AX8. One amp block, no bells and whistles.
 
I could be wrong, but I think the 44.1Khz sampling rate is an unfortunate relic of early adherence to Nyquist's theorem that says a sampling frequency a little over double the sampled frequency will return an accurate reproduction of the original waveform. In audio, we really don't care about anything over 20Khz, so 44.1 is "a little over" double that.

Equally unfortunately, it's an oddball float number, so it adds processing time, while 48Khz, 96Khz, 192Khz, etc. are evenly divisible and easily derived in hardware. No processing time, which leaves processors free to do other fun stuff.

I seriously doubt anybody can hear the difference between 44.1 Khz and 48 Khz, but the electronics involved are dramatically different. Again, unfortunately, since the industry started off in a bad place and it settled in as dogma, we're stuck with this goofy sampling frequency and now there's a bajillion dollars of gear and habits out there that would have to be overcome and replaced to get to a sensible place. Not gonna happen. At least, not this week. But, if smart/determined guys like Cliff keep pushing for sensibility, perhaps one day the status quo will realign.

But, I'm not gonna hold my breath while waiting. It's just a theory of mine.

There's also filtering and aliasing issues you have to be aware of when considering the Nyquist theorem. Cliff has been noted as saying he'd personally prefer 64 kHz sampling rate because of this. But I'm sure Fractal devices were designed with 48 kHz performance in mind.

If you or anyone wants a quick primer about sampling rate and such...
https://wiki.xiph.org/Videos/A_Digital_Media_Primer_For_Geeks#Sample_Rate
https://xiph.org/video/vid1.shtml


A general response to the thread... Kemper gets used in lots of studios, fixed 44.1 kHz sample rate from S/PDIF is all the digital output it has. So while I see how it can impede recording studio integrations, seems to me like there's an easy way around it... no?
 
It's nice that you all have blind faith in Cliff. Unfortunately, he doesn't understand professional digital interfacing.

A professional digital device must handle whatever clock it is given.
Look at any DAC, ADC, digital equalizer, or digital reverb.
They all accept 44.1, 48, 88.2, 96, and none of them place any constraints on the clock.

People have suggested work-arounds:
* Use analog
* Use 48khz
* Use a sample rate converter
* Change your workflow

But the fact remains:
The Axe III does not have a professional digital interface.
 
Last edited:
5 pages regarding clock frequencies and not one mention of "jitter" and/or precision of clocking. I find this to be concerning.
 
A professional digital device must handle whatever clock it is given.
Look at any DAC, ADC, digital equalizer, or digital reverb.
They all accept 44.1, 48, 88.2, 96, and none of them place any constraints on the clock.

Not entirely true. Just take a look at the all-new Eventide H9000 which has limitations in this area. SPDIF is not supported above 48 kHz, and certain algos won't work above 48 kHz.
 
This comment is exactly why I'm concerned. Stability of clock (jitter) is far more important than chosen clock frequency.

Importance of jitter is just a different topic, IMO. Frequency is used here in the context of interoperability, first and foremost. And converter noise and resampling artifacts are much more noticeable than jitter. Impact of these is huge, immediate, easily heard and measurable.
 
Unfortunately, he doesn't understand professional digital interfacing...
But the fact remains: The Axe III does not have a professional digital interface.

Could it be that this is because it is 'not' a professional digital interface but rather a professional guitar pre-amp? Does a professional interface have a guitar pre-amp built in? Therefore they don't understand guitar preamps? I had a Yamaha MOX6 keyboard with a built in interface that would only do 44.1. A non-professional keyboard?
 
That was my thought almost exactly on this point. Turning into a full-featured audio interface at the level Cliff does things would most likely raise the cost to around the 4K range if not more, and put it out of reach for most of us. I already have my RME, and I like having an interface that is fully dedicated and capable of very low latency when I work with soft synths and use the software monitoring for other things. For my Axe-FX III, I just want a nice guitar device for recording direct and I'll most likely use it for my live rig too. It sounds great recording the analog out into my interface, and I don't think anyone could tell the difference in a double-blind test vs using digital connections.
 
Yamaha, A&H, and dozens of others have very expensive professional, modern, digital gear that tops out at 48k. Granted, for live rather than studio use. But to say they aren’t professional devices because they don’t clock higher than 48k is ludicrous.
 
Topping out at 48k was the norm in pro studio gear until about maybe 15 years ago when 96k slowly became a studio standard as well. But these devices had both 44k and 48k usually and that is the point. The problem is not sound quality issues where 48 vs 96 makes the gear pro or not, it's that 48k is by far the least used rate industry-wide to clock your room full of gear at and is quite rarely used so being stuck at 48k is the showstopper when integrating your gear into the digital side of your studio kit.

Traditionally, most pop, rock, hop, R&B, country etc (noisier musics, plus often with 50-150 tracks running simultaneously) studios clock their entire rooms at 44 and still do because they don't need the extra quality of 96 (especially in this ghastly age of mp3's/Youtube) and moreover, their computers can't handle the double-load of 50-150 96k tracks running at once. Classical, jazz, etc recordings (and mastering) have less tracks usually and listeners expect a higher level of sound quality, hence they jumped to 96k (even 192k and DSD) as their standards asap. In recording these types of music, the engineers nearly always seek sound quality and accuracy as their goals. In recording classical and jazz, you want to "document the performance" where in contrast, pop, rock, hop, R&B, some country, etc, the recording studio is a creative instrument in itself so it’s part of the paint going onto the blank canvas.

With all this said, I believe that if Axe-Fx users had the option to clock at 44, or even if 44 was the only option, that the you would satisfy the majority of the complaints. Why? Because Axe-Fx is far more scarce in classical, jazz and mastering audio work where 96k+ is more prevalent as the clock rate in their studios settings.

Locking the Axe exclusively at 44k as Cliff mentioned would force a sacrifice on the vast majority of Axe Fx users and therefore should be ruled out if there is any audio degradation at all. Then it becomes is adding 44k as an option along with 48k easy (low cost in regards to FAS’s time, money and/or Axe performance) or very costly to pull off. I can only assume from FAS hearing almost 10 years of moaning that adding 44k to 48k is costly.
 
Last edited:
  • Like
Reactions: cn7
...I am pretty convinced that it isn't because 96kHz inherently sounds better, but because the native sample rate of those devices is 96kHz and then it's down converted on the fly to a lower sample rate using a $1-5 chip. And this notion that you can't get good quality at lower sample rates like 48kHz is wrong. I've recorded at multiple sample rates using a high end DAC and there is no difference. The real cost is in the filters of the DAC. It is expensive to have quality components that have a sharp rolloff at the higher frequencies, but it is very cheap to just throw a faster clock in there and oversample it and then filter it. If you want good quality real time SRC it is costly. Even then software conversion is still better and a hell of a lot cheaper...

Well stated. Also my experience, and my reading of published tests: If 96k performs better: its attributable to the particular hardware (and software) implementation and performance, not the sample rate.

My crystal ball shows 48khz as the future standard in sample rates.

Music recording studios aren't the big kid on the block. Most professional sound is done at 48khz. These are the markets of broadcast TV, radio, movies, and streaming video. This has been the defacto professional standard for decades (adopted for its compatibility with video frame rates) and predates the 44.1 CD standard. Media and streaming industry pressure is now underway to standardize 48khz as the delivery format for everything - even iTunes. IME, recording studios that do even occasional work on soundtracks or for broadcast media run at 48khz (or 96khz) and only convert to 44.1 at the end as a delivery format for CD or MP3. Now might be a good time to consider eventually moving the workflow to 48khz as its likely that 44.1 is on its way to being a legacy format.

The Nyquist theorem has been empirically demonstrated: There are no audible frequencies that can't be represented by a 48khz sample rate. As further refinements to digital filtering and converter performance at 48khz are implemented, I won't be surprised if 96khz or 192khz eventually no longer offer any sonic benefit and are not called for.

Seems to me the weakest link in the chain is now at the audio reproduction end: The speaker (monitoring)!
 
Last edited:
Well stated. Also my experience, and my reading of published tests: If 96k performs better: its attributable to the particular hardware (and software) implementation and performance, not the sample rate.

My crystal ball shows 48khz as the future standard in sample rates.

Music recording studios aren't the big kid on the block. Most professional sound is done at 48khz. These are the markets of broadcast TV, radio, movies, and streaming video. This has been the defacto professional standard for decades (adopted for its compatibility with video frame rates) and predates the 44.1 CD standard. Media and streaming industry pressure is now underway to standardize 48khz as the delivery format for everything - even iTunes. IME, recording studios that do even occasional work on soundtracks or for broadcast media run at 48khz (or 96khz) and only convert to 44.1 as delivery format for CD or MP3. Now might be a good time to consider eventually moving the workflow to 48khz as its likely that 44.1 is on its way to being a legacy format.

The Nyquist theorem has been empirically demonstrated: There are no audible frequencies that can't be represented at a 48khz sample rate. As further refinements to digital filtering and converter performance at 48k are implemented, I won't be surprised if 96khz or 192khz are eventually only used in processor oversampling, and no longer offer any benefit as intermediary or delivery file formats.

Seems to me the weakest link in the chain is now at the audio reproduction end: The speaker!

Higher sampling rates than 48k don't seem to make too much sense to me, with the availability of DAC level oversampling for digital filtering and anti-aliasing. There is something of an argument against producing content at higher sampling rates, as intermodulation distortion can occur from higher frequency content through speakers, amplifiers, etc. that were never meant to reproduce those frequencies...

It seems to me that the audio pros know this, as you've described. Dunno, only a hobbiest, I don't work with this stuff! :p
 
the all-new Eventide H9000 has limitations in this area

How embarrassing for them!
it is 'not' a professional digital interface but rather a professional guitar pre-amp

Apparently the word "interface" is ambiguous.
We're talking about the AES and SPDIF digital interfaces, not the USB computer audio interface.
I had a Yamaha MOX6 keyboard with a built in interface that would only do 44.1. A non-professional keyboard?

Yes, a non-professional keyboard. I hope the Motif and Montage do better.
My crystal ball shows 48khz as the future standard in sample rates.

Perhaps it will be. But right now, the standard is to support 44, 48, 88, and 96.

In a digital studio, the master clock determines the frequency and phase.
It's the Axe's job to follow whatever clock it's given.
It's like an orchestra -- the players follow the conductor.
I can only assume from FAS hearing almost 10 years of moaning that adding 44k to 48k is costly.

A sample rate converter chip is like $3.

The DSP and cabinet IRs can run internally at 48khz.
Only the AES and SPDIF I/O needs to be flexible.

In a digital studio, every single device is a slave.
Slaves place no requirements on the clock.
 
Last edited:
Status
Not open for further replies.
Back
Top Bottom