Posts by Per

Hey nightlight, I'm using two cheap and terrible mic's (but they're what I've got). An SM58 and an e609 silver.

Initially I was using no pre-amp at all. So I don't think the preamps are a problem, especially given that the pro's run the mic through a desk with expensive pre's as part of their signal chain while profiling.

Perhaps if I used the preamp to add compression or distortion it could be an issue as DonPetersen said but I've avoided that, most of the time without even using software preamps, just the vanilla line-in's on the interface (so it's just going through the audio cards pre's). I found it improved things when I started routing this way because I could reduce the volume to a level the Kemper would accept after pushing the amp a bit louder. But please do listen for yourself to the latest set of tracks in my last post, I could be hearing things in which case I need to find some other way to attenuate mic volume.

From my experience in the other thread (which I think still isn't over) :

Things I have learned thus far.

Background noise is the enemy of high end. Crank the amp to solve this, the difference between amp and background will be greater and the result will be that you don't lose as much treble. I think this is because the Kemper see's background noise as frequencies to ignore/notch out, I think only a ML algorithm would be able to eliminate it and extract audio without eliminating the frequencies. Fingers crossed things improve in the future because that initial chime and fret noise really suffers due to this. Anyhow when you make the amp louder the relative volume of the background noise decreases and the problem is (almost) solved.

Mic level is more important than you think. The previous solution leads to the next problem. Your mic will send a signal that's way too hot to the Kemper when the amp is at appropriate volume. The Kemper will literally tell you that the signal is causing it to clip. The solution is to put something in the line to control the mic volume and bring it back to "Mic line" levels (which apparently is a lot quieter than normal line levels). For me I plugged the mic into one of the spare inputs on my audio interface, then using it's mixer routed it back out to one of the spare outputs and then plugged that into the Kemper. That way I was able to use the mixer built into my interface to reduce the volume. You could also just use any mic pre-amp as they all seem to come with a volume control. You should try to match on the first page of the Profiling mode, as you can watch the output LED while switching between Kemper and Mic signals and also listen to both as well as then doing minor fine tuning adjustment using the level control (right most dial under the screen) on the Kemper.

Reverb radically alters the frequency response. The Kemper doesn't capture your room sadly as an IR with a little convolution reverb module (I wish it did as it would solve this last one), but I've found that listening very closely and building a reverb that matches the sound of your room as closely as possible will make a *huge* difference to the profiled sound. I cannot emphasize enough about this one because you might think that a close mic'd amp has no reverb... you're wrong. I found raw profiles sound really dead and lackluster in an almost unidentifiable way, but get a good reverb on there and you can find complexity and missing frequencies pop back out at you. It's crazy what a difference this can make to a profile and it's almost universally overlooked.

Yeah I bet that's a tough one to sing. One option is to split the line into two parts, that way you can have all the content but give her the space to catch a breath and even do a little harmony.

Hey Christoph - I found the levels to be quite different here after profiling. Like I said I had to adjust them in logic. The session wasn't DI for the gearslutz profiles. Before profiling during setup I only adjusted mostly by ear and eye (on the master output LED) while switching the A/B on the first page of the Profiling Mode, but once the profile was done and after I'd recorded both re-amped signals (Kemper and real amp) by sending via SPDIF to the Kemper then switching A/B I found that that I had to adjust the relative volumes in the DAW's mixer (logic X) to get them to match.

Wouldn't it be normal for the two volumes to be radically different though? One is line level the other mic?

Anyhow I did some more profiling. I think one thing I'm learning is that there's a big improvement if you use a preamp... but not the way you'd expect. It's in order to make it quieter! If you make the amp too loud then the Kemper can't handle it, so up till now i've only had the amp low enough that the Kemper doesn't complain. However if you use a preamp in the Apollo and use it to turn the volume down then you can crank the amp and the result seems to be much better... so far of course I may just have cloth ears. At any rate louder amp means lower background noise floor.

I've just done another two profiles - a merged and a studio one. I'm getting pretty fatigued at this point so I don't know how far or close these are (what do you guys think?), but I did notice that with the DI profile the source has a little more "hair" to it than the Kemper result, however this time the merged was I think slightly better than the studio version.

Here's the audio files, apologies for having them in a zip rather than separate, but hopefully you'll give them a try out. I couldn't capture the profiling session for the DI though. so it's just re-amped signals.

http://peranders.com/music/dem…/KemperStudioVsMerged.zip

And the two profiles (one merged, the other the studio, both with the same cab) are attached. With the recordings I took time to try and get a closer verb in the Kemper, I wish I could capture a short IR and use a convolution verb with the Kemper on this too because I feel sure that this accounts for the larger portion of the remaining difference, it's surprising how much difference it makes.

permesacrunchmands.zip

Anyhow guys, please let me know if I'm getting closer, as I said my ears are tired.

ckemper - I know but honestly that's what it was, I made those on Gearslutz specifically that way right after profiling before saving the profile because I wanted to keep the background noise loop as I know how important it is when trying to do that comparison especially with cleaner tones. I have no idea what happened to the noise, but as you can hear the Kemper actually does have more amp "hum" noise, just the background noise went AWOL.

Wheresthedug - Thanks. Yeah the hotplate definitely does affect the tone. But my THD Hotplate was attached all the way through that session so I'd still expect the merged profile to sound in the same ballpark as the "studio" profile, or even the original amp, it just doesn't. I'll make a new profile later on today and see what happens, I'll do a studio profile at the same time so there's a direct A/B with the Hotplate in the signal chain the whole time.

Hey Christoph, thank you so much for checking it out. That’s fascinating about the noise effect.

Actually the samples were recorded during A/B compare immediately after profiling. I know the Kemper records a loop of background noise for then but those examples are what it came up with and the treble difference is clearly noticeable.

Speaking of it being clean (it’s not really they’re both at edge of break-up), when it tried a merged profile the Kemper really thought the source was much cleaner and gave it a completely different distortion sound to the actual amp sound (much less crunchy), I wasn’t able to match at all (more or less sounded nothing like the source amp). The DI profile was taken from a THD load box I plugged in for the purpose which I know isn’t the best but the line out signal shouldn’t be dramatically different to what it sends to the speaker, just much attenuated. I should find and share that data too. But it’s only for the Mesa, the Marshall doesn’t have an accessible speaker cable (closed back little bugger).

Fantastic job and great take on the outro solo! You nailed it on this one

Nice song! Great vocals.

Two tiny nitpicks, for me I'd have punched the backing track up a notch when it gets going, but I have bad taste in commercial country style music with this sort of song so don't listen to me there. The only other thing for me is there are one or two points at the highest notes where your daughter is running out of breath after such a long line and it's affecting the pitch slightly, I know it's verboten but personally I think it's ok if you just slightly melodyne that one note ever so gently so it can be pure and breathy, either that or perhaps introduce a gap in the lyrics so she has space to take a breath there.

Otherwise awesome job!

Quote from bshaw92

Is there a possibility you got an unlucky bad cable?

I don't think so as the same cables and mics are used to make the re-amp tracks (I use the Kemper in profiling mode to send the signal to the amp), and you get all the treble in those.

I did a new quick session today while waiting for a compile, see the post just before this one, you can hear what the profiler is outputting and how the amps are responding, maybe the Kemper team is able to better identify either what I'm doing wrong or if there's the slight possibility of a bug.

So while I was sitting waiting for a compile to finish I thought to myself - what if I were an engineer at Kemper, looking at this, I'd probably think "that's a bloody useless bug report, I need data", so I decided to do a loopback of the mic through my apollo to capture profiling session data itself.

I know this isn't yesterdays stuff and I've been futzing around with the settings a little on the Mesa one so it won't exactly match up with yesterdays settings, but regardless I figured it might be useful info.

So here you go ckemper & team :

http://peranders.com/music/dem…per_profilingsessions.zip

The resulting two profiles from the session are attached. You will want to disable the effects section on the Mesa one as before I saved it I enabled a studio EQ and touch of verb.

Hi, the Mesa (first profile) is a 5:25 Express using an e609 silver mic, the second profile is some crappy Marshall practice amp using an SM58.

No to any switcher or splitter.

The signal path is as in the manual, Kemper send to amp input via jack, mic to return via XLR, guitar to Kemper front plate during profiling and for original take with Kemper delivering over SPDIF left dry, right monitor mono. Then reamp signal to Kemper via SPDIF, return all monitor (default setting). Space off, pure cab off, everything else in default state on the Kemper (had to factory reset it recently in order to be able to install the last OS update).

It’s possible however, if there were a gate it would be audible in the re-amp through the amp itself (it isn’t), and also that doesn’t apply to the other amp which isn’t a Mesa and shows an even more stark loss of treble.

Sitting thinking about this later, the frequency difference is so great I’m left wondering... it can’t be possible that it’s a foldover artifact can it? Just the way other frequencies go up as the treble goes down.

I think my money is still on some sort of lo-pass, or noise gate, or the resolution of the cab section though.

Interesting, maybe. But I just tried with my other little amp I have in America, which is just a horrible little practice amp and while the Kemper did an admirable job and in many ways improved it's tone what's noticeably missing is really just the treble/definition. The Kemper adds in quite a bit of mids too, it's a surprisingly wide deviation in frequency from the actual source :

http://peranders.com/music/demostuff/Kemper_per_reamp2.wav

vs the actual amp

http://peranders.com/music/demostuff/SM58Amp_per_reamp2.wav

Based on the fact that we know my inputs are good, and the output is also good, and from what I can tell my Kemper sounds much like everyone else's Kemper out there, then I think either I have the worst two cases for it to try and profile that just flummox the algorithm or perhaps this is an problem in the current OS (I'm using the latest 6.0.0.14574), because I don't think I have a buggy unit (though you never know).

I'm wondering if professional profilers ever encounter this, if they do maybe that's what the EQ they all seem to slap into the X slot is for? When I tried the EQ it couldn't bring back the complex treble sound though, just boost what's there. Maybe it's only something that crops up under a really specific set of circumstances. I can't imagine no-one else has encountered this at all.

Haven't tried it with another amp, however the profile is accurate in others ways including the amount of compression. It's the loss of treble that doesn't make sense.

No you're right that couldn't be the case and I just realized why. Originally when I made those tracks I was using the A/B switching during profiling to record both re-amps, so the mic'd up amp version is going through that return circuit and yet clearly it has all of the treble content, it's just the profile that lacks it.

I was wondering whether it's down to background noise, or perhaps overzealous noise gating in the profiling process? On the other hand you can hear how much BG noise there is in the mic'ed up reamp and it's not really getting in the way at all. The Kemper seems to have a more extended bass than the original when looking at the analysis tool in Logic (but it's not great because it only has 64 frequency plots and it's linear rather than log so it's very hard to see in the critical areas what's going on).

I wonder if it's down to the cab resolution?

I feel both sides on this and know where Christoph is coming from.

It's really hard to get good beta testers in software development, must be even harder with hardware like the Kemper. It's even more difficult to debug something here because there are so many variables. What sort of mic are you using, what preamp, what cables, is your room treated or not, how much background noise is going on as you're profiling, how did you measure and match your levels, are you using a power soak or anything else in your signal chain, how do you get your DI signal? etc before you even get to the Kemper itself. Most people will get worn out eliminating all the variables, and at some point they're going to be "but now it doesn't sound anything like the sound I wanted to actually capture". People will give up.

There's so much potential user error, so much potential for people who just don't get the difference between amp in room and amp in studio.

At the same time I also feel that there's a slight insinuation that only feedback from pro's making commercial profiles matters. And that doesn't sit well. Obviously it matters if those guys aren't happy, they should be listened to, they know the process inside out. But I feel that my feedback in the thread further down about treble woes is still valid and while it may well end up being user error I still want it to be looked into. I still want to play my part in helping this product be the best it can be because I own a Kemper, I have skin in the game.

Hey guys, thanks for the thoughts.

Monkey_Man - Yes, it adds some clarity but makes it sound quite different.

sambrox - I thought about this too, but it didn’t improve things.

The profile is in the rig exchange if anyone has a decoder to check out what’s going on with the cab frequency graph.

I tried making a merged profile too, but interestingly the Kemper decided the signal was way cleaner than it in fact was when I did that and the merged result was dramatically different when it came to distortion, I’d have thought this would preserve more treble however the top end was still missing. I was getting much closer with a classic profile.

I thought about what if it’s because of power amp distortion too, so I pulled the master all the way down to below 1, and still the same problem.

I’m wondering if it’s the conversion on the Kemper? Is there some way to tell the Kemper to use SPDIF for the return during profiling? Then I could ensure that it uses my interfaces converters for the job.

Locked to off.

I'm just wondering, is there any sort of lo-pass on the return input? Just because I've been trying to profile my Mesa again and I keep running up against a problem of missing detail at the top end. I'm assuming there's no way to take SPDIF as the Return source?

https://www.gearslutz.com/boar…?p=14082324&postcount=125

The problem is manifesting here. The other people in the thread are in fact right upon re-listen a few hours later and with less ear fatigue, I'm definitely missing that detail in the upper end. But just jamming up definition or using a studio EQ doesn't do it. If I try and do a merged profile the result seems to completely lack the distortion character too. It's like I'm missing that tiny bit of transparent upper mids area that makes it sound glassy, like you can hear every nuance of the guitar from the amp, but with the Kemper it sounds like distortion is being overlayed evenly.

The trouble is it's driving me up the wall. I've tried profiling a dozen different ways, re-routing the mic through a UAD preamp and back out to the Kemper, adding verb to the Kemper source in cause it's a room thing (it really doesn't seem to be, although I couldn't quite nail the room sound with the inbuilt verbs). Changing definition and clarity settings, adding EQ's, these gave a very glossy high end but didn't bring back the detail there (if you know what I mean).

Of course the most frustrating thing is that I have no technical language to explain exactly what I'm hearing there, only diffuse emotive terms such as clarity, glassiness, transparency, grit, complexity of distortion, hearing the "woodiness" of the guitar push through... gah.

If anyone has any suggestions, or have I hit the limit of the algorithm/cab resolution here with the Kemper when it comes to distortion and detail?