Audio Issue

[dbijoux]

Is it possible you can post an example .wav?

I also record everything into the DAW rather than using the Juno to sequence and mixdown a whole project. That said, I still feel comfortable offering advice on preemphasis because I do use samples and have done enough tests to hear, see, and quantify the difference.

I just mixed down a song because it's been a while. I do notice a very slight change.

Personally, I like the mixdown version better. It's cleaner and clearer.

I guess before we go further, what kind of monitoring situation do you have going there?

You see the thing is, when I compare the mixdown file to the output, the difference is the high end cutoff. It's quite possible you like that sound better.

However, the 'punch' you are hearing may actually be the lack of definition in the highs. As you'll learn, everything has it's balance and if you take away one you inevitably end up raising another.

If you could post a clip I could be more specific, but I'm going to venture a guess that you have a hard hitting snare and hats in there stealing the show.

[KRE8A]

Hello Dusty,
When you have done a final mix-down on the Juno - G, the clarity of the mix will have sound character differences through any different system you play. This is because of the headphones you are using through the system and also the type of reference monitors you are using too. To get the music a-like on your computer, from the Juno - G; you will have to do your final mastering through a recorder sequencer program. Example ~ (LOGIC).
It's impossible to get any made music on the Juno - G to sound exactly or close sounding, on a different system, unless you are lucky to fluke it. Although the music will sound different anyway>, through the types of headphones or speaker systems.
You can get the song so close to how it was firstly originally recorded, only when you have done the final touch up through a sequencer program.

Just to quickly point out about pre-emphasizing and Normalizing. These two features are very dangerous when you execute to a mix and save, because the song gets re-boosted to a maximum level, below the maximum threshold; without peaking the LED indicators - on the mixer. Creating new frequency levels through your song, changing everything according to the levels and FX you have edited on. Believe me, I know through recording experience. I normally use these features when i am recording natural instruments without FX nor changing levels, ~ unless the sounds aren't to the medium of an EQ analyzer, or peaking through the roof.
If you are going to Normalize or pre-emphasize, save your original as a duplicate before doing so, otherwise you will have to go back to editing the entire mix again, unless you go through undo history, that is of course if you haven't erased history.

Keep up the good work

[Dusty Loops]

Hey dbijoux,
I'll post the .wav file shortly.

KRE8A,
Thanks. Most of my friends use LOGIC, but its going to take me some time and $ before I get to that stage. So far, I'm mixing and sequencing the whole beat in the Juno for now.

[Dusty Loops]

Here ya go.
I just hit mixdown and saved it as a sample onto the card and transferred on to the PC.

http://www.mediafire.com/?ygh3uiz2jwj

[dbijoux]

Thanks for sharing Dusty.

I don't think it sounds too bad here! There's a big 808 kick, snappy snare, some nice hats riding in the back, an off-beat sub going, and synth over the top.

The first thing I noticed was the snare sticking out +3db over anything else. Maybe obvious, but normalizing is only going to bring your levels up to that peak. A suggestion might be to bring that out to another track, lower it in the mix, limit/compress/finesse it.

KRE8A is right again about mixing everything in the Juno. It's going to be more difficult than with a DAW, but I don't think it's impossible. A DAW just gives you more tools, or you could even say more ways to f@*k everything up.

About the punch, I really don't hear it missing, although without being in your room listening to the original it's hard to tell. The kick/sub combination isn't really punchy in the first place; you might be able to tighten it up by dropping out the sub a little earlier before the kick.

Working on the Juno alone, you'll have to pay more attention to the instruments. The synth pad you are using takes up a lot of energy from lows to highs. Possibly EQ that, use a filter on the patch, or try a different sound to get it to fit better.

I liked the subtle hats in there adding some variety. The toms or block add a bit, but one might be hitting on the first double kick too.

All in all, you've got a few hurdles to get over - no monitors, no 'proper' mixing board, but keep the end result in mind you might be able to get through them.

Back to your original question of why it sounds different than the MIDI? Aside from the normalize/preemphasis changing the sound(for better or worse), any time you mix down or process audio, the Juno(any computer) needs to make digital calculations. It's not perfect, there will always be quantization errors. It's also only 16bit/44.1khz so there's room for improvement. That said, the more you learn, the more you can compensate to get the sound you want.

[Dusty Loops]

dbijoux ,

Thanks a lot for taking the time out to analyzing it. I haven't been able to get a single person from Roland to respond back about this, but this definitely helps and gives me some direction. If I get a chance, perhaps I'll try taking it to the studio and see how the it outputs via Logic. I'm still working my way around learning all the mixing functions on the Juno, so I'm sure it'll help me find the kind of result I'm looking for. I think I'm gonna do some research on mixing board and get a decent one. Just something to get by with for now.

Thanks again man. Really appreciate the feedback, everyone!

[dbijoux]

No problem mate. If it makes some sense maybe there is hope for us both.

Taking it out to another system, a pro studio even, definitely couldn't hurt. Mixing on your speakers you'll have to try things out everywhere to make sure it sounds ok.

Since you're only on the Juno, where there are no EQ, compression strips, you'll have to be more creative how you do things. Many people like how Roland sounds fit together from the start, but you still might have to use MFX or tweak the patches. The best thing about that is, you'll be forced to make your instruments fit now rather than throwing a bunch of EQ or compression at it later.

[TheMvuser]

This doesn't make sense.

Rolands usually have de-emphasis active on the output stage. like on the MV8XXX and the MC909. This basically Rolls off highs, thats why you need to emphasize the samples...because they are de-emphasized on the outputs.

if the sound is transferring and it is duller on the PC then either the Synth is artificially UPPING the high Frequencies or the PC is LOWERING them.

you need to know how the emphasis schemes are implemented on the Synth.
are the outs adding extra highs. is it de-emphasizing on bounce down.. etc

[dbijoux]

What doesn't make sense mate? Did you follow the whole thread?

It's pretty well common knowledge now that Roland employs an emphasis scheme in their DSP. You are correct, there is essentially a -6db filter @12khz in the output stage. There's a post floating around here highlighting it on the Juno, the same on the Fantom-X and it's experienced on the G as well.

In regards to your question about the bounce down; I can say if you're "resampling", the deemphasis filter is employed as the signal is going through that output stage. On the contrary, if you "mixdown" internally and copy the .wav files directly to your computer, you bypass the output and the highs will remain in-tact.

Hope that helps sort it.

[TheMvuser]

sorry dbijoux.. I was referring to the original post here.

Quote.
When I convert my performance into an audio track and save it on to my card, the audio doesn't have the same punch and brightness to it as it did previously. The performance will sound a lot live through the speakers, but soon as its been converted and on my computer, it doesn't sound the same through the speakers. Any possible ideas to what might be the cause? Thanks.

dbijoux I meant this doesn't make sense.. above.

I didn't really (mean) it doesn't make sense.. what I mean is

It looks like Roland are implementing different pre- emphasis / de-emphasis schemes all over the place.

Like for example... you could have de-emphasis on ALL outs and for ALL conditions...Digital / Mono / Bounce down to Wav internally.

in THIS condition, everything would be the same because the WAV would have the same signal frequency response of the de-emphasized audio you are listening to on the Synths OUTS and under ALL conditions.
Soooo it would sound the same because it IS the same in frequency terms when you are listening to it after you have completed doing, whatever it is you were doing..

This is the case on My MV8000, even if I burn a CD through its mastering scheme the audio gets de-emphasized.

However, one condition where this fails is if I save an internal sample as a wav inside the MV to a file and then Load that over USB to a computer it is then, (brighter). I would have to save the file, reload it .. add pre-emphasis and then save it again.... madness.

but the crazy thing is, it looks like the designers of Roland gear are putting in different schemes in different machines so the process is not even consistent.

they seriously need to get a couple of new algorithms in these machines to compensate for these oversights.

Dusty Loops looks like he has to re load it in his synth and then de-emphasize to me.

sorry if I am just repeating advice already given.

I can't work out if Dusty Loops is getting less hi frequency or more hi frequency.
I think Dusty is encountering a de-emphasis curve personally.
Lets say for example, Dusty's synth outputs to the speaker don't have de-emphasis applied, or they have a compensation curve active BUT
the Digital or other paths of the architecture do NOT have that compensation and therefore the de-emphasis is effectively active..
Record into sound card aaaand it is DULL because you have just hit a path which is being de-emphasized.

really Dusty needs to do one (with) pre emphasis and one with de-emphasis too.
but dusty has to make sure that the file is at -12db first because the pre-emphasis may clip.
This is getting crazier and crazier... :-I

Roland need to be made aware of these issues and add OS upgrades to sort them out.
unless of course you think it is ok for a machine to go way outside frequency response spec and even clip your signals..

Roland manuals omit to give any procedural advice regarding these (vital) issues.

[dbijoux]

NP mate, I replied to this before your other posts. I think you are exactly right when you say there may be a variety of deemphasis processing involved. I have not specifically applied any testing on Mixdown comparisons, since it's not a part of my workflow, but I do recall there being a noteable difference in sound the few times I tried. There is also a difference by transferring audio via USB - as the audio will not go through the ouput filters a second, third, or even more times.

On the other hand, I have tested the output stages in a similar fashion to the method you've described here and elsewhere. Here is a graph and hopefully sufficient explanation:

http://forums.rolandclan.com/index.php?action=show_thre...

I also agree, this is a very confusing subject and seems to cross over many of the products @ Roland. If it is a varying degree of processing(algorithms) or it can be chalked up to user experience, I cannot be completely certain. I first learned of the pre-emphasized samples from Artemiy, since he had spoken with some technical team on this.

Still, there is no mention in any manual(of-course not) and no official explanation on the methods used, or best-practices for artists working with the machines. To a point, I can understand this could be proprietary information. However, as you've pointed out the effects can be less than favorable, especially in the case adding pre-emphasis would clip the waveforms. :\