Latest song...probably one of my best efforts

Register to hide this ad
Sounds great! The drumming section is nice, and the piano work is good too. I like the fade out too. :thmbsp:
 
That was really nice, like truly enjoyable to listen to. One goal of homebrew recordings, I think, is to make them not sound like homebrew recordings. The only part that really screams homebrew to me here is the piano part in the middle. It sounds small and canned. I'd find the biggest stereo grand piano sound you can find. It has to be huge since it's sort of carrying the song in that section.

That drum part is great. Is it a loop that you found somewhere or did you create it yourself? And what is your recording rig? I recently bought an older Mac (17" Powerbook) with GarageBand to start recording some stuff with. Monitors are Yamaha HS50M's. Mics are a pair of AT3035 condensers with a PreSonus Firebox interface. I've got the guitar end of things covered with several acoustics and an electric and a nice old Fender tube amp, but I'm kind of short on everything else. Next on the to-buy list are a MIDI controller and a bass guitar. Maybe also a drum machine or Drum Kat or something along those lines.
 
That was really nice, like truly enjoyable to listen to. One goal of homebrew recordings, I think, is to make them not sound like homebrew recordings. The only part that really screams homebrew to me here is the piano part in the middle. It sounds small and canned. I'd find the biggest stereo grand piano sound you can find. It has to be huge since it's sort of carrying the song in that section.

Thanks all for the comments. I definitely concur here that the goal realistically is to make a home recording sound like it had the kind of work and resources poured into that a major studio might have produced (if this is really possible).

The piano was just a freebie vsti that I used, I have many much better payware pianos but that was the most convenient and least cpu hit one that I had access to so in it went. I did use most all the tricks I've learned from past mixing so I can only hope to get better in the future. Goodness knows I don't have nearly as much equipment as I used to - but hopefully I can make up for that with careful tweaking and smart mixing.

Right now I'm getting away like a bandit with a cheap Dynex 16bit sound card hooked up to a pair of Creative HQ1400 headphones which actually destroys the sound quality I was getting from numerous other outboard "professional" cards and breakout boxes (had a couple of Digidesign mbox's, some MOTU stuff [shouldn't have sold that one[, and also ESI, EMU, outboard dacs, etc.)

Used to have a pair of Celestion 5's hooked up to a Yamaha P-2100 amp that I used to monitor on but I found I enjoyed mixing on headphones too much and never went back...
 
Rifftrax, would you care to share what recording tools or devices you're using? The SQ of the track you provided is quite excellent and to my ears, sounds better than many of the home or small "Pro-Tools" studio recordings I've heard recently.
 
Rifftrax, would you care to share what recording tools or devices you're using? The SQ of the track you provided is quite excellent and to my ears, sounds better than many of the home or small "Pro-Tools" studio recordings I've heard recently.

Yeah no prob.

Actually - it's kind of funny because (a little ways in the past) I used to have quite a bit of stuff that I used in my main recording rig. I had a couple racks full of hardware (50 + units) with a couple patch bays completely filled to the brim with hookups and multiple interfaces. At one point I was using the digital out on a ESI Juli@ to a Zero 24/192 dac hooked into an Antique Soundlabs MG Head OTL Mark III tube headphone amp running a pair of AKG K701s as my monitoring setup.

It was bloody lovely - really - but somehow, not quite "realistic" in a sense. Almost as if the transients to sounds had been lopped off and replaced with super-warm tube sustain.

These are past interfaces I used:

EMU 0202
ESI Juli@
Alesis Photon X25
MOTU 2408 MKII (w/PCI 424 card)
Mbox (2)
Denon DA500

As I mentioned - I don't really "record" currently (as I did often in the past, I had a 'mic locker' of about 2 dozen decent mics) but just do multitracking with synthesizers and sampled instruments and then mix everything track by track with one of the hundreds of vst/dx fx I have (freeware/payware, you name it). And it may seem crazy but I'm using a pair of headphones I got for $22 plus an $8 soundcard (Best Buy native brand: Dynex). Surprisingly, the SQ is really absurdly good, the headphone output from any interface with the exception of maybe the MOTU doesn't measure up to it which I think is pretty hilarious. So I just use that when I track (Creative HQ-1400 headphones + Dynex soundcard via internal headphone output).

I've owned/used almost every single major DAW in the past including: Cubase VST, Cubase LE, Cubase 2, FL Studio 7-9, Reaper, Reason 2, Sonar LE, Mackie's Tracktion, Cakewalk Project 5, Digital Performer, Pro Tools 6, Power Tracks Pro, Nuendo 1, and others I can't think of right now but I've settled on a combination of FL Studio for tracking/composing/prototyping, Reason for extra flavor, and Reaper for mixdown/mastering/post-processing which is what I currently use.

I think the real "magic" (YMMV) with the stuff I do may just be in the careful application and use of the instruments and fx that I own, which are very very numerous. I sold some of my payware but I've got a vst folder with hundreds upon hundreds of files and probably a couple hundred gigs of extraneous samplebanks that I use in tandem to everything else.

The thing I think makes one of the neatest tricks I've got though is the use of convolution reverb on each track. I've got a collection that I've compiled from numerous numerous sources of around 2,000 convolution samples that I use on each track during the mixdown process in Reaper. I took around a full entire days worth just analyzing the file qualities of these couple thousand files and separating them into qualitative banks and getting rid of the crappy ones. I open up an instance of SIR 1 and just cycle through the many folders I have to find the right one and just add (normally) a touch to give the instruments that 'life' and realism that can be the difference between 'nice sounding' and 'woah' for some instruments.

I have a couple other pretty in-depth tricks that I've discovered with certain vsts fx functionality and how I use them just in terms of eq/comp/reverb/stereo-expansion techniques that I use often. In most cases even good multi-samples and instruments start out raw and dull sounding but after some careful tweaking can suddenly sound like they are filled with character and life.
 
Well, I listened to it a couple more times and I'm really liking it. It's definitely got an X factor going on that gets its hooks into you and makes you want to listen again. You've got some really nice ambient synth pads going on back there that sweeten the whole thing and give it some nice swirl. I especially like the one that's prominent around 1:18.

What do you mean by "convolution reverb"? I've never heard those two words together before.

How did you create the drum track? Was it something you found and looped or did you build it from scratch?

Is the guitar part a live guitar or a synth part? If it's a synth it's really convincing. The symphonic voices have a nice realistic non-MIDI quality to them. Do you have any special tricks for accomplishing that?

What kind of spatial processing did you do? The chimes seem like they might have some magic dust sprinkled on them they way they inhabit the sounstage. When I move my head around they move in phasey ways (or maybe it's just because they're bright). The guitar seems to have a subtle panned tremolo effect, or am I just imagining that?

Where did you get the sound of the kids towards the end? Long shot but is it a sample from Pat Metheny and Lyle Mays' As Falls Wichita So Falls Wichita Falls? I love ambient real sounds like that -- kind of Pink Floydy.

Can you recommend a good inexpensive MIDI controller?

Your mix is really, really good -- the levels are right on the money. I'm listening on a set of fairly crappy Altec-Lansing computer speakers. Amazing that you were able to get it to translate from the headphones. That's the mark of a good mix (or one of them) -- when you can play it on a variety of systems and it works. I like how you let the drum part be loud and up front.
 
Well, I listened to it a couple more times and I'm really liking it. It's definitely got an X factor going on that gets its hooks into you and makes you want to listen again. You've got some really nice ambient synth pads going on back there that sweeten the whole thing and give it some nice swirl. I especially like the one that's prominent around 1:18.

Sorry, I realized I wasn't going to have time to answer this for a bit but now I've got some spare time...you're right about the 'x-factor' thing though - that's kind of been always what I've strived to make a part of my music.

What do you mean by "convolution reverb"? I've never heard those two words together before.

Convolution reverb works on actual full-bandwidth (audio-band) impulse recording of the properties of a real space. They are normally 'captured' by using a sound generator that has a extremely narrow ADSR (like a balloon pop, single firework explosion, loud clap, very short white-noise sample playback, etc.) that will excite the acoustic properties of the real space that is being recorded - then loading that "impulse response" (recording of that sound and the reverberation) into a convolution processor which "involves sending two audio samples through fast Fourier transform (FFT) algorithms, multiplying their spectra, running the product through an inverse FFT (IFFT), and playing back the results." The two audio samples would be the sample you want to apply the reverb too and the IR (impulse response) itself. The same type of idea can be executed to actually sample classic reverb units perfectly as long as they are non-modulating (involve some type of random start chorus/phasing, which many do). Other-wise you get a static but still very usable idea as a IR of what say a Eventide H3000 sounds like...

It's very cool stuff overall - indispensible if you're a musician/composer and will eat tons of your time when you realize what it can do.

http://emusician.com/tutorials/emusic_acting_impulse/

also:

http://en.wikipedia.org/wiki/Convolution_reverb

How did you create the drum track? Was it something you found and looped or did you build it from scratch?

It was a combination of things...I loaded a loop into a beat slicer and then played with it a bit to get it to taste - and then layered it with another beat I created to fatten the sound a bit.

Is the guitar part a live guitar or a synth part? If it's a synth it's really convincing. The symphonic voices have a nice realistic non-MIDI quality to them. Do you have any special tricks for accomplishing that?

Completely sampled of course. (I don't have a guitar that sounds that nice nor any interface/preamp to record with right now anyways...) TBH, I didn't even use and midi-type humanization techniques here (they are all the same velocity level for every note if I remember correctly, and occur right on the beats/divisions of a beat) It may simply have been in the processing I used in the mix...?

What kind of spatial processing did you do? The chimes seem like they might have some magic dust sprinkled on them they way they inhabit the sounstage. When I move my head around they move in phasey ways (or maybe it's just because they're bright). The guitar seems to have a subtle panned tremolo effect, or am I just imagining that?

There's a couple of amazing fx (vst) that you can find if you look up "bootsie vst". Those are a couple of those well kept secrets... :D

Some others I love (which seem to get no love) are Crysonic's stuff, especially "Sindo" which has a special "shuffle" and "bass trim" knob which work some serious subtle magic for spatial enhancement without totally skewing a stereo image. There is also an incredible Antress vst comp. plug which is essentially a Manley dual-comp copy that has some odd but pretty amazing effect on the stereo field gained just by placing it on a track (I use it on the master bus each time I mixdown without fail).

Other I love the heck out of are Voxengo's Crunchessor and Warmifier, Refined Audiometrics CLAS, and of course SIR 1 for convolution processing, though I think I just found another free convolution processor to 1-up it...

Where did you get the sound of the kids towards the end? Long shot but is it a sample from Pat Metheny and Lyle Mays' As Falls Wichita So Falls Wichita Falls? I love ambient real sounds like that -- kind of Pink Floydy.

Nah, not anything sampled like that. It was just a sampled I gleaned from on of my sample packs I've purchased or downloaded from royalty-free sites I think. I've got a big collection of those.

Can you recommend a good inexpensive MIDI controller?

Definitely. Depends on what you want though.

Small controller with midi-assignable knobs/faders: Alesis Photon X25 (has it's own decent audio interface too! nice touch)

Large weighted stage keyboard: Casio Privia PX-100 (get a used one)

Medium non-weighted for uber-cheap: Yamaha PSR-225 or thereabouts (get off of CL and you're good to go, perfect for touch response, dirt cheap and synth action which can often be better than weighted honestly for most tracking purposes)

Your mix is really, really good -- the levels are right on the money. I'm listening on a set of fairly crappy Altec-Lansing computer speakers. Amazing that you were able to get it to translate from the headphones. That's the mark of a good mix (or one of them) -- when you can play it on a variety of systems and it works. I like how you let the drum part be loud and up front.

It's taken a lot of practice to get good at mixing on phones (hundreds and hundreds of songs)...but I think I'm getting to where I want to be with it.

You might like this one too: http://www.newgrounds.com/audio/listen/284585
 
Last edited:
Oh yeah, that's a nice one too. It's got a similar catchy X factor as the other one. In fact I might even like it better. Its arrangement makes it a little more instantly gratifying.

Oddly enough, I totally get the whole convolution reverb thing after you've explained it a little. They made us do FFT's and convolutions by hand in engineering school so I was familiar with the basic concept...just wasn't sure how it applied to reverb. But now that I think about it -- yeah, pretty clever. I guess you could model an acoustical space by trying to dial in all the various the delay parameters -- early and late reflections and all that -- but how much more elegant to just find its impulse response and then stimulate it with the input signal in the frequency domain. The FFTs would probably be pretty demanding on a processor but these days with all the processor power available that's probably not such a big deal.

Do they actually go out and find a nice sounding space and ping it and measure the response? If so I could imagine there being some seriously realistic reverb models out there. My theory has always been that the big studios were safe since they had great sounding rooms that couldn't be replicated by a digital box, but with this approach maybe a box really could come close to capturing the subtle magic of a great room.

I also wonder if guys like Korg and Line 6 are using a convolution approach for modeling guitars and amps and stuff.

Thanks for all the other infos. Very, very interesting.
 
Do they actually go out and find a nice sounding space and ping it and measure the response? If so I could imagine there being some seriously realistic reverb models out there. My theory has always been that the big studios were safe since they had great sounding rooms that couldn't be replicated by a digital box, but with this approach maybe a box really could come close to capturing the subtle magic of a great room.

I also wonder if guys like Korg and Line 6 are using a convolution approach for modeling guitars and amps and stuff.

You hit it right on the money...there's a really popular one that banks off just this approach, and has the sound quality part in spades (for acoustic space IR recording and usage for a dedicated plugin - see the equipment they use for sampling):

http://www.audioease.com/Pages/Altiverb/AltiverbMain.html

This other one in a sense is even more spectacular for what it offers:

http://www.audioease.com/Pages/Speakerphone/speakerphone.html

Also right about modeling through FFT convolution for microphones, guitar amp simulation, etc. - normally always how it is done short of creating an algo that simulates individual circuit arrangements.

An even more extended take on the convolution idea is this guy - it's a different spin and gets more complicated (not sure exactly how it functions but it's supposed to be amazing and eat cpu for breakfast) based on volterra kernel processing:

http://www.acusticaudio.net/modules.php?name=Products&file=nebula3
 
This is great stuff!

Now something I was scratching my over the other day starts to make sense:

http://www.youtube.com/watch?v=0gqCtPcN5NE

It's Dweezil Zappa playing Eruption on an SG through some kind of modelling rig that totally transmogrifies his sound into Van Halen's guitar tone and on the record including the room/verb ambiance. It's pretty amazing (starts at 3:10).

I can see what the guy in the third link is doing with his dynamic convolution, and it addresses what I saw as a weakness when I was thinking about the static IR approach that some of the other guys apparently are using. The problem is that the impulse response isn't always going to be the same for all signal types, especially when you've got nonlinear stuff like distortion going on. So you need to grab a different IR depending on how some parameter of the input signal changes. Like I could imagine there being a lookup table with a range of IR's based on input amplitude or something.

From reading his FAQ, it sounds like they're already doing that with dynamic convolution but he's going a step further by not abruptly switching between IR's but "morphing" between them, which I'm sure sounds more natural. It's probably only perceptible on some some subtle psychoacoustic level since I'm guessing the standard dynamic convolution and even the static convolution would sound very convincing (and I'm guessing that reverb sounds would be the most amenable to being convincingly modeled with static convolution, since I'll bet a room's basic response doesn't change much with varying input signals -- unlike say a guitar pickup where when you hit it hard it might compress or distort or do something else nonlinear ----- which gets me back to wondering how those COSM guys are doing what they do). But it's getting that much closer to how a real acoustical system works and so there would be some benefit in realism.

He also talks about standard IR approaches being "memoryless" which I take to mean that what he's doing has memory in some sense. I'll bet he's got it where older decaying responses continue to tail off according to the IR that was in effect when they first entered the pipeline rather than switching over to the most recent IR, if that makes sense. Assuming I'm understanding it right, I can see how this would chew up processor power since you could have layers and layers of different IR's going on at the same time.

He mentions that the volterra kernal guy published a paper on his method in the AES journal (which I used to subscribe to back in the day) along with the results of doubleblind tests. I'll bet that would be an interesting read.

I just watched this:

http://www.audioease.com/Pages/Speakerphone/speakerphone_whatsnewmovie.html

Too cool...
 
Back
Top Bottom