n_Track's 'clock'

How good is it?

So here’s the deal…

I’m (still) looking at getting a Yamaha i88x. Unlike many audio interfaces, the Yamy will only sync to its internal clock if you set the sampling frequency to 44.1KHz. So if yer using any other sampling frequency, you are dependant on the “clock generated by your computer”.

I asked a sales rep at Sweetwater how good that would be (i.e. it seems that there is significant attention paid to the quality of your digital clock…jitter and whatnot…for quality recording, and instinct tells me that a dedicated purpose-designed hardware-based clock would trump whatever is under the hood of a general purpose PC). The rep said that it is not a hardware-based clock with respect to the computer, but based on whatever software DAW you are using. He didn’t ask and I didn’t tell him what software I’m using as I’ve gotten NO respect from those people in the past when I confidently tell them I’m using n-Track (likely they realize at that point that I’m not going to take out a second mortage and join them in PT slavery so who can blame them I suppose…).

“As long as your software isn’t complete garbage you should be fine” he says. Well I know n-Track isn’t complete or even partial garbage, but how good is the clock?

Educate me. Anybody?

I’m fairly certain that it runs off your PC clock, or the soundcard, depending on wether you pick system timer or wav time respectively so it’s dependent on the quality of that. I’d probably go wav timer in most circumstances.

It would be interesting to set up ntrack to record with the i88 at 24/96 and set ntrack to use the wav timer instead of system timer.

Hi Guys:
This could become a fastenating Thread/Topic…

That’s a nice response, Willy…

Thanks for puttin this up, sweet_beats…

I’m of the opinion that the reselution of the audio of a DAW is dependent on the “Clock” of the audio Hardware… And the “Jitter” of the resultant audio is dependent on how well the the Multi-Track Editor resopnds to the combination of the DAW’s Clock and the Audio Card’s Clock as it is couple’d by the Editor…

Boy… Big-Words-There… :O ??? :p

Anyway, I got the Topic Tracked… I hope the “Geeks” respond to this one… My RADAR Buddy and I argue about this idea, all the time…

Bill…

I wouldn’t get a soundcard that didn’t provide its own clock for the rates you plan to use it. Clocking is critical to good sampling. Accept no substitutes.

Note: if you have one card that has a good time source and a digital out like ADAT or S/PDIF or clock output, and the new card has those as inputs, you can use the clock on one card to drive both – as you should if you’re using both cards at the same time anyway. What this means is the above statement about “I wouldn’t buy…” applies to a FIRST soundcard, but not a second.

The clock/oscillator circuit in a PC is notoriously CRAPPY! You will get better accuracy off almost anybodies decent soundcard clock (wave timer). That said, I’m sure some manufacturers “cheat” and base the sampling clock off the PC clock anyway. So whats a man/woman to do? Check that card out carefully before you plop down your hard earned Denari, Rials, Dollars, Rupies etc. Where specialized outboard clocks really shine is when you want to use multiple digital streaming devices. To insure proper synchronization, they MUST run off a common clock. Preferably a good, jitter-free clock. Google up some stuff on “Digital Audio Clock”. Facinating reading. Specially for geeksters! :D

D

Man but I am overwhelmed. I mean, thanks for the responses, but if I’m understanding correctly, my whole plan is blown!

I think I need some educating here…was Sweewater Man speaking gibberish then?? :O

<!–QuoteBegin>

Quote
The clock/oscillator circuit in a PC is notoriously CRAPPY! You will get better accuracy off almost anybodies decent soundcard clock (wave timer).


Wave timer…what is it? Am I understanding that, contradictory to Sweetwater Man’s statement, there is no such thingy as a software clock? That a software DAW software references an internal hardware clock in the PC? And the “wave timer” is the hardware clock on your PC’s soundcard (PCI or whatever)? THAT bites…

Why on Earth then would somebody like Yamaha waste well reputed converters and whatnot on the clock on your SoundBlaster or VIA AC97 card or whatever…?!

<!–QuoteBegin>
Quote
Clocking is critical to good sampling.


I totally agree, learjeff. That makes total sense. That’s what I’m shooting for.

The Yamaha i88x is a firewire interface…so if you can’t set the i88x’s internal clock as the master (unless one is recording at 44.1KHz), then you’re stuck using your computer’s soundcard clock as the master…or worse, the clock on the MOTHERBOARD! I’m having memories of MUCH more desparate times…passive bass guitar plugged directly into the 1/8" mic input on a SoundBlaster 16…monitors were a really nice set of generic PC speakers…relax everybody. I knew it was wrong…its all I had. The sound was SWEET!

Again, just for clarification, the i88x is a multi-channel firewire audio interface, not a soundcard per se. It features 8 channels of balanced analog I/O, S/PDIF I/O, and 8 channels of ADAT Lightpipe I/O.

<!–QuoteBegin>
Quote
I’m fairly certain that it runs off your PC clock, or the soundcard, depending on wether you pick system timer or wav time respectively…It would be interesting to set up ntrack to record with the i88 at 24/96 and set ntrack to use the wav timer instead of system timer.


Where in n-Track do you designate “system timer” vs. "wave timer? Does the wave timer generally blow compared to the internal clock of a reputable multi-channel audio interface?

If this horrid calamity is true, then I’m toast. I’m using a laptop so the option of installing an affordable decent PCI soundcard is not an option. Is the clock on most onboard soundcards decent or doth its lameness stand in the light once you start using a more serious interface?

Why would Sweetwater Man insist that a software DAW generates a syncable word clock? Instinct told me differently. Software doesn’t generate anything…it merely controls or is controlled by hardware…arg

Instinct led me here to you rowdy chaps.

<!–QuoteBegin>
Quote
Thanks for puttin this up, sweet_beats…


You’re welcome, Bill. I’m pleased to post something that is of interest while at the same time helping me learn and maybe keeping me from making a mistake!

Bring on the comments! :p

<!–QuoteBegin>

Quote
Where in n-Track do you designate “system timer” vs. "wave timer?


Go to “Preferences” and the “options” tab. In there is a section titled “Use system timer for”. Untick both options to use the Wave timer.

There is a similar setting on the “midi settings” tab too.

<!–QuoteBegin>

Quote
Does the wave timer generally blow compared to the internal clock of a reputable multi-channel audio interface?


You’ve got it a bit mixed up. The wav timer is on the soundcard, system timer is the pc clock.

Some wav timers are known to be very good - the emu cards for example. Dunno about the yamaha, maybe they planned on it being used with other mlan devices.

Just to clarify… the software has nothing to do with the clock. The clock is a hardware thing. If you have a crappy clock that you syc to, it makes no difference if it is ProTools Stenberg, or n-track, the result will be the same.

What Great Replies, So far…

Most of us here are aware of this in some way/respect, with our set-ups…

The CBC here, where I live, has what is known to be a (and I stand Corrected)… A “Master Clock” that all their “Editing Suites” and “Production Rooms” “Sync To”… when they are used… This… I believe keeps all the electronic Hardware in “Step” with each-other…

But all of these “Master Clocks” are “Slaves” to a single “Master/Master” Clock somewhere, in the Nation… I have no idea where it is… That’s not important… But when a “Feed” is switched from one part of the country to another there are NO-LOST FRAMES… Well…

I just wish I could “Connect” to IT… :p ??? Would IT help my Bass Playing?? :p Whoooooppppps…

Bill…

First, when we say “soundcard”, we mean “audio interface”, and it doesn’t matter whether it’s Firewire, USB, PCI, or ESP (that’s the “extra sensory plugin” interface … :wink: ).

While there might be a nerdy technical wimpy justification for what the salesman said, basically he’s blowing smoke. You just can’t do a clock purely in software, you need a timing reference of some sort. Without a soundcard clock, the only thing left is the real time clock on the MOBO (“system” clock). Which is notoriously bad on a laptop.

However, note that the i88X would not use the system timer unvarnished: it would still need clock discipline circuitry to divide that clock into smaller bits and regulate it suitably for recording. Still, I wouldn’t trust it with a 10 foot pole. There are a number of timing & synchronization issues with DAWs, and I’m not in favor of adding yet one more.

We frequently get queries here about timing problems in recorded tracks. The top answer is “install the latest soundcard drivers”. The second is probably “Use WAVE timer rather than system timer”, in n-Tracks “Properties -> Options”. This cures so many ills that I would NOT want a soundcard where it’s not an option.

I say choose a different soundcard. There are other Firewire interfaces. I use the MOTU 828 myself. Not cheap, unfortunately – but I got mine used on ebay at a reasonable discount (and it’s the original 828, not the 828mkII which is much better).

I have a question though: why wouldn’t you be happy recording at 44.1kHz? Is it the “faster is better” thing? While there are technical reasons why higher rates can give better results, the difference is rather marginal in most circumstances, and doing a good job of engineering is FAR more important to overall quality of results than 44.1kHz vs 96kHz. FAR, FAR, FAR more important. I’d say that you have to a pro quality, and perhaps maybe even a better-than-average pro quality engineer before the difference would be discernable in a mix.

<!–QuoteBegin>

Quote
I have a question though: why wouldn’t you be happy recording at 44.1kHz? Is it the “faster is better” thing? While there are technical reasons why higher rates can give better results, the difference is rather marginal in most circumstances, and doing a good job of engineering is FAR more important to overall quality of results than 44.1kHz vs 96kHz. FAR, FAR, FAR more important. I’d say that you have to a pro quality, and perhaps maybe even a better-than-average pro quality engineer before the difference would be discernable in a mix.


I believe Jeff speaks the truth here. At least that’s been my experience as well. My card can go up to 192K. Does my stuff sound any better at 96K or 192K? Not really. I have a LOOONNNGGG way to go in this game before that kind of difference in sampling rates will really help.

D

This is awesome.

Okay…

Learjeff, I’m TOTALLY with you on that. I know people that gracefully place me in the position of galactic neophyte with their worst productions. There are three reasons really:

1. Flexibility. Unfortunately, from a marketing persepctive, job opportunities are lost these days if you can’t tell them you can capture their stuff at 96KHz (and for some its 192KHz…goodness). If I had a reputation like REAL engineers, clients wouldn’t be asking/caring what I’m recording at, but until then…
2. I’ve found that with my own stuff, if it is an effects laden production it seems that the result is better down the line when the master takes are at something beyond 44.1Khz. The bit rate is definitely the key here though.
3. Hard-headedness…I can’t stand the though of paying for an interface that “goes up to 11” (i.e. 96KHz…does that make 192KHz ‘22’?) and me not being able to take advantage of that unless I’m enjoying using a “notoriously bad” darth hideous MOBO clock or soundcard clock.

Some of you, or maybe all of you have read a great article on this topic. It can be found here: http://www.lavryengineering.com/documents/Sampling_Theory.pdf

That’s where I’m coming from. I’m in total agreement with all of you. Scads of incredible music has been captured at 16-bit 44.1KHz, and I’m not even opening the analog discussion here. Timeless classics recorded using (by today’s standards) unacceptable equipment, and we are still trying to duplicate the result. There was also some incredible equipment to augment those great recordings at the front end, but it takes somebody who knows how to use it. n-bit n-KHz won’t fix lousy preamps and engineering.

LATE BREAKING NEWS

I might just be confused after all! No suprise to some…

Here is a link to the i88x owner’s manual: http://www2.yamaha.co.jp/manual/pdf/emi/english/synth/i88XEFG1.pdf

Look at the inset on page 10 titled "About the OPTICAL SELECT switch, MASTER CLOCK indicators, and Wordclock"

In there there are two references to the i88x defaulting to its internal clock running at 44.1KHz. I read that and I had been following some threads on gearslutz.com with users frustrated that the i88x would only operate as the master clock at 44.1KHz, but as I look further I think the user was using Mac OSX, and there is/was a driver issue that Yamaha hadn’t worked out in OSX where the OS would not allow the i88x to be the master clock, but OS9 and Windows XP are okay with it. If I’m reading page 10 correctly, if you set the i88x as the master, and the OSX computer is saying “No, no, no” (but probably really, really fast), the i88x parks at 44.1KHz. Am I reading that right?

The kicker is on page 14 about halfway down. It says “The i88x can operate at a sampling frequency of 44.1kHz, 48kHz, 88.2 kHz, or 96kHz as either the wordclock master or slave.” So I’m okay right (running Windows XP)???

<!–QuoteBegin>

Quote
However, note that the i88X would not use the system timer unvarnished: it would still need clock discipline circuitry to divide that clock into smaller bits and regulate it suitably for recording.


Learjeff, could you go into this a bit further? It sounds like you are saying even if I can set the i88x as the master clock, I’m still at the mercy of the system timer.

Woxnerw, I’ve found a great way to improve MY bass playing…I’ve got a friend who plays bass really well and when he plays I sound much better! :D (not that you need THAT sort of assistance…I’m speaking purely of myself)

I really appreciate all the input, everybody.

New thought: let’s say we track something at 24-bit 44.1kHz. Is it important/valuable to convert that to 32-bit and a higher sampling rate for engineering/production? n-Track (and many others) have these utuilities, but do you muck things up in the conversion process there and back? What are your experiences?

Also, what on Earth does “floating-point” mean?

I’m disappointed that the person at Sweetwater that’s supposed to be expert is convinced that a software DAW generates word clock, and that, in spite of the fact that he stated he was very familiar with the i88x, he agreed that it would only operate as the master at 44.1kHz in spite of the info on page 14 of the manual.

<!–QuoteBegin>

Quote
Learjeff, could you go into this a bit further? It sounds like you are saying even if I can set the i88x as the master clock, I’m still at the mercy of the system timer.


Nope, so long as you set N-track to Wave timer, the clock in the Yamaha will be used.

<!–QuoteBegin>
Quote
New thought: let’s say we track something at 24-bit 44.1kHz. Is it important/valuable to convert that to 32-bit and a higher sampling rate for engineering/production? n-Track (and many others) have these utuilities, but do you muck things up in the conversion process there and back? What are your experiences?


Ugh, is this in the wiki yet? Okay, in n-track (or most DAWS for that matter) you record your tracks in 16 or 24 bit. Then, when the DAW goes to read those tracks and send ti through its mixing matrix, it adds additional zeros to the source on the fly taking things to 32 bit. What this does is cuts way down on any rounding errors. (We’re talking billions of colculations here, so it makes a difference.) Then on the way out the software has to cut the output back down to 16 or 24 bit depending on your sound card. What most folks will do is render their final mix to 32 bit and master this 32 file. As a last step in mastering your convert to 16/44.1 and add dither usually. Nothing actually records in 32 bit (yet).



<!–QuoteBegin>
Quote
Also, what on Earth does “floating-point” mean?


It is a math thing. There is fixed and floating point math. Fixed means the decimal point never moves… there are so many bits going through the CPU everytime no matter what. Floating point lets you do things like 23^-8 and stuff like that rather than 0.0000000023. From Wikipedia:

<!–QuoteBegin>
Quote


A floating-point calculation is an arithmetic calculation done with floating-point numbers and often involves some approximation or rounding because the result of an operation may not be exactly representable.

As an example, a floating-point number with four decimal digits (b = 10, p = 4) and an exponent range of ±4 could be used to represent 43210, 4.321, or 0.0004321, but would not have enough precision to represent 432.123 and 43212.3 (which would have to be rounded to 432.1 and 43210). Of course, in practice, the number of digits is usually larger than four.



The precision issue he mentions is why we use the 32 bit float above. It gives a lot of room to compute things while reducing rounding errors to an almost insignificant level.

Bubbagump, don’t mean to be a burden. Thanks for the info. That clears a lot of questions up.

<!–QuoteBegin>

Quote
What most folks will do is render their final mix to 32 bit and master this 32 file.


So am I understanding correctly that most people don’t convert their individual tracks to 32-bit floating-point, but rather engineer them in their native format and then render those tracks to a stereo 32-bit floating-point file?

Anybody else wanna weigh in and tell me if you think I’m interpreting the statements on the i88x and its ability to be the master clock?

Thanks again, Bubbagump.

sorry I missed this thread…

I have an i88X. You sort of already figured it out, but you should be able to use as the i88X’s clock with any sample rate, although most guys are gonna tell you that 44.1/24-bit is good enough.

I also think the i88X is junk and I’m sending it back. Its a nice piece of hardware but the drivers suck. I wouldn’t recommend it.

Good replies all, Bubba! Rumors to the contrary notwithstanding, Bubba does indeed have a clue. (The rumors were started by Bubba, of course.)

Beats, inside n-Track, everything IS in 32-bit float. The tracks are how you recorded them, but as soon as N reads them it converts them to 32-bit float because that’s its internal standard. It’s a good internal standard.

Now, on mixdown, you get the choice of what format to use: 16-bit fixed, 24-bit fixed, or 32-bit float. If you choose 16 or 24, you lose some of the information – some of the fidelity. So, if the song isn’t entirely done being processed, e.g., you’re going to send it to be mastered, there’s no point in throwing away any information: just leave it in 32-bit float. When all is said and done, of course, we dither and convert to 16/44 for CD format, or 24/96 for DVD (IIRC – never done it myself).

There’s another advantage to 32-bit float. “Clipping” doesn’t realy clip. OK, that’s a contradiction in terms, so let me explain. The word “clip” comes from going over the equipment’s max signal range, so the peaks get leveled off. (With analog gear, it’s not a flat level, but something more interesting. With early ADAT, really nasty things happened like wraparound. With current digital, it just gets leveled off perfectly flat.) Now, that kind of clipping only applies to fixed-point formats.

With floating point formats, the maximum codable signal value is over 700dB – which is probably how loud an A-bomb is. I mean, you won’t get there. However, there is still a “nominal” max level, the level you’re not supposed to go over because it WOULD clip when converting from 32-bit float to any fixed-point format. So, it’s CALLED clipping because it WOULD CAUSE clipping if converted to fixed point. And it IS clipping if you’re listening to it, because that’s through your soundcard that uses fixed point. But there’s no degradation of the encoded signal. Now, why does this matter? Well, it just means that when using 32-bit float, you can “get away” with clipping as long as you manage to fix it at some point before you’re done with the project.

So, use 32-bit for any intermediate stages.

BTW, I didn’t realize you had paying customers. That’s a good reason to use the best, just so you can say you did. Funny how customers are, but that’s how they are. Also, I agree with you that even in theory, certain kinds of plugins are very sensitive to sample rate. (Most of all, the kind that use fariable delay-lines, like most chorus/flanger plugins, and some pitch-shift doublers. The difference in resulting sound quality for those is obvious even to folks with mediocre hearing.)

So I think you’re good to go with the Yamaha.

Cheers!
Jeff

PS: We probably should put something in the Wiki about these things!

That’s a very interesting link, btw, to davry engineering. He explains why rates higher than 44.1k might sound better, but going over about 60k is a waste of time. (I confess he goes way over my head with the math, but I think I get the point …)

Interestingly, so far I’ve only seen one credible scientific experiment that claimed there were any benefits to sample rates higher than 44.1k. (It involved gammelon music and people who were used to listening to it – and man, that stuff does have high frequency content!)

One omission in the paper, though, is that it is only concerned with recording and replaying the original waveform, and doesn’t take into account any issues related to signal processing like using FX and mixing. I’m sure if he’d seen the algorithms used by variable delay line FX, i’m confident that he’d say that yes they’d sound better in higher rates, and that it wouldn’t matter whether they were recorded at higher rates or upsampled to them.

Hey SW,

Yeah, that article from Lavry engineering really helped me understand a lot about digital audio. I found that site several months ago. That’s why I recommend folks to Google “Digital Audio + Sampling Rate”. There a bunch of good resources on the web to help folks understand a little better. Lavry is also why I record almost exclusively at 24/44.1 or 24/48. My card can go higher but at a reduced physical I/O count. If I remember correctly, in one his articles Lavry points out that sample rates at or above 96K do improve the audio to some degree but it takes specialized equipment or “Golden Ears” to REALLY hear/see the difference.

I did a test at home with just strumming some chords on the Strat going direct through my Korg Pandora with a little reverb and a rotary speaker effect. Same stuff recorded at 24/44.1 and again at 24/96. I could hear a difference… but just barely. If I only had the time, I’d like to do a full song at 24/44.1 and again 24/96. Just to see if it makes a noticeable difference. I doubt it will with the gear I have.

D