mattplaysguitar
Max Output Level: -55.5 dBFS
- Total Posts : 1992
- Joined: 2006/01/02 00:27:42
- Location: Gold Coast, Australia
- Status: offline
A handy vocal reverb technique - non linear decay - AUDIO SAMPLES on post #10
EDIT: Refer to post #10 for audio samples I was just listening to the song Bird on a Wire by Sarah Blasko and noticed the non linear reverb on her vocals and how perfect it is. The song can be heard here: http://www.youtube.com/watch?v=EHz9qkH5xr8 I have not played around with this at all as have not yet had time, this is all thought out in my head! This is not gated reverb. That is a type of non linear reverb, but I'm talking about a time dependent reverb. I'm not sure how you would create this effect without having a reverb unit that can do this specifically. Waves Renaissance does it, I think. Basically it's an extremely long delay time, maybe three seconds or possibly even more, with a decent pre delay, and the tail is cut off in time with the music so when she starts singing a new part, the non linear sections kick in a cut the verb off much shorter than the 3 seconds it wants to go to. It's not going to be perfect for all songs, but it is great in this song as the predelay allows the vocal to still be really forward and have an amazing verb sound, but then the non linear cuts it off before everything becomes extremely muddy. I don't think a gated reverb would work well for this as vocal dynamics are complex, but gated works fine for a snare for obvious reasons - for the same effect. If anyone has any ideas how to recreate this outside the reverb plug (for those that don't have this option) I would love to hear them. I don't think it's possible, and needs to be done within the architecture of the reverb emulator, but I'm open to suggestions. To explain why this is different from gated reverb, I'll pass it over to the Waves manual: "Non-linear – most of the time, non-linear reverbs refer to gated sounds, but in this Type, we’ve provided for longer offsets and late reflections, allowing non-linear behavior which has louder reflections at the end of the maximum time, and very tight spacing when Size is at its minimum time." Maybe I'm wrong and gated would work. If so, that'd be great and I'll have to try it, but I don't think it would because: - tail end of a reverb hits the threshold where it would be gated normally - but just before it gets gated, another reverb starts on top of it, bringing the level up and stopping that tail from being gated. - the tail from the first sound plays out till it ends and the new reverb continues. - if it's a vocal, the reverb may never fall down low enough to trigger gating and the sound becomes very very muddy - if we process internally within the pluggin, cutting all tails off after a certain time, then that tail doesn't play into the rest of the reverb and make it so muddy. - A gate can't do this when the reverb sounds are all so close together, but doing this internally will make a huge reverb sound, but without the muddyness. Thoughts?
post edited by mattplaysguitar - 2010/05/14 22:56:37
|
The Maillard Reaction
Max Output Level: 0 dBFS
- Total Posts : 31918
- Joined: 2004/07/09 20:02:20
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 08:03:06
(permalink)
how about an envelope or two?
|
Philip
Max Output Level: -34.5 dBFS
- Total Posts : 4062
- Joined: 2007/03/21 13:09:13
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 11:04:59
(permalink)
Matt: I listened to your url intently (in sections). TBH, Whatever room reverb she has is too subtle for my ears to pick up. I think I do detect hypercompression on the lead vox ... and tube saturation (she may have driven the pre-amp gain into the tubes of her Avalon 737), and secondary volume maximization on the master buss, I don't know. Her ultra-rich level-vox effect is unnatural and disturbing to my crony ears ... inducing me to slam the volume down by 1/2 through (on 1st listen). You may disagree. To complicate matters, there is not enough cacaphony padding the 1st verse (just stormdrums): The ears get hypnotized by this naked compressed vox-trance and her words become arbitrary tongue-speaking, to my ears (I couldn't understand a sentence that she sang) IOWs, this vox seems an analogue hyper-compression phenomenon with a german condensor mic ... lower harmonics bleed as well.
|
8 ohm
Max Output Level: -89 dBFS
- Total Posts : 51
- Joined: 2010/03/31 21:09:14
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 13:09:13
(permalink)
You could run all your verb through a reverb Bus and automate the output with envelopes. Assign a fader to reverb level and ride it to the music. Jiggle the fader to the beat to give it rhythm Have to different reverbs on different bussed that come out at different times. A gentle, small one with lots of pre-delay for the intimate parts and a big swelling one for the high energy parts. Throw a little echo on a third buss and subtly ride it in on select parts. At the end of a couple choice phrases So subtle that you don't notice it unless you are looking for it. I love automaion on reverb levels. It gives a great dynamic sence of space.
|
tarsier
Max Output Level: -45 dBFS
- Total Posts : 3029
- Joined: 2003/11/07 11:51:35
- Location: 6 feet under
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 13:37:58
(permalink)
No, gated won't work, nor will envelopes on a reverb bus. Just off the top of my head, an impulse response will do the trick. You would have to edit the impulse to fit the tempo of the music, and play with the tail to cut it off how you want it to. Perfect space lets you do that by hand, but it won't follow the tempo of the song. The Lexicon PCM Native will also let you do this, I think.
|
The Maillard Reaction
Max Output Level: 0 dBFS
- Total Posts : 31918
- Joined: 2004/07/09 20:02:20
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 14:18:40
(permalink)
I noticed that the song in question has a very even phrasing pattern. Why do you think envelopes will not work? Just curious to understand your thought process. I make my own IRs with the Voxengo synthesizer.... I agree with the idea that a "room" can be suited to a particular tempo. best, mike
|
ShadDOH
Max Output Level: -70 dBFS
- Total Posts : 1027
- Joined: 2009/12/31 20:50:02
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 15:32:42
(permalink)
I wish I understood this stuff better :-( This is way over my head.
|
MemphisJo
Max Output Level: -79 dBFS
- Total Posts : 594
- Joined: 2009/02/08 20:24:22
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 20:26:38
(permalink)
when I clicked on that link Firefox crashed and I had to re start my computer to get it to work again! what's up with that?
|
mattplaysguitar
Max Output Level: -55.5 dBFS
- Total Posts : 1992
- Joined: 2006/01/02 00:27:42
- Location: Gold Coast, Australia
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 21:12:05
(permalink)
Philip - I'd agree that there is pretty strong compression on her vox and quite possibly some tube saturation too. I am supprised though that you seem to have gone as far as to say she was using an Avalon 737 (or are you just making an assumption?) with a German condensor mic and has lots of lower harmonic bleed, but you can't hear the hall verb... In my opinion the reverb is extremely wet and is very very clear. Listen to the sections where the vox ends and there is a gap, it is most obvious there. Then when the next line kicks in you can here the verb suddenly cut. Specifically, the times 21, 22, 23 seconds each have it very clear. My first impression was that a gate would work on these bits, but it won't because when the next line kicks in and the level is brought back up, the previous tail will still be playing with the new reverb which is going to really muddy the signal up. Also please elaborate on lower harmonic bleed as I have never heard of that one before and google gives me nothing! The vocal certainly sounds a lot more harsh on the youtube clip because of the lower quality, sounds much clearer on my cd. Maybe try listening to it in HD? Also are you on your monitors listening to this? Tarsier - I believe you are correct. Envelopes can't work because they don't cut the tail that bleeds into a new verb sound (I elaborated on this in the original post). But your idea on modifying a conv reverb would definately work. Excellent suggestion! Mike - I'll make an image to better describe it! 8 Ohm - I love automating reverb sends too, it works great but as I'll explain more clearly in another post, this ain't gonna work for this effect! MemphisJo - No idea what's up with that! It's just a youtube link! You can search for the video youtube if you prefer. The song is By Sarah Blasko and is called Bird on a Wire.
|
mattplaysguitar
Max Output Level: -55.5 dBFS
- Total Posts : 1992
- Joined: 2006/01/02 00:27:42
- Location: Gold Coast, Australia
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 22:23:22
(permalink)
Ok, I made audio examples instead. These were made with Waves RVerb. For clarity on what the reverb is doing, it's set to 100% wet. http://www.mattlyonsmusic.com/sonar_samples/ Right click and download if you wish as might be easier to watch and listen in a wave editor. File size is 4.68MB and 320 kbps mp3. Settings were: Predelay - 27.8 (even though it's 100% wet, this is important because of the early reflections) Time - 2.39 Size - 100 Diffusion - 100 Reverb Type - Non-Linear Early Reflections - 0dB Reverb - 0dB Mind you this is not tweaked to the tempo, just roughly there, but not really! Everything is in the one file. Playing order is: 1 - Original reverb sound with a linear decay/ 2 - Volume envelope. Could only do a fade at the end as all the other sounds were too close together. 3 - Gated reverb. Certainly not tweaked at all (or even set really!), but you get the idea. 4 - The effect I have been talking about. Non-linear decay set to the smallest value of 0.04. The following tracks 5, 6 and 7 are tracks 2, 3 and 4 played along with an inverted track 1. This shows you distinctly what is different about each one. For those of you who don't understand this technique, it's playing two nearly identical tracks together and flipping the phase of one so that every single thing that is the same about the two gets cancelled out, and everything that that is different still plays along together. My analysis: 1 - very muddy and wishy washy 2 - still verb muddy as nothing happens till the end where there is the envelope. The problem with this is that it sounds unnatural and no envelope is going to make it sound natural as the decay is much much more complicated than a simple lowering of volume 3 - gated verb on vox just sucks. surely it could be much better with tweaking, but save it for percussive sounds 4 - still very muddy but there is certainly a noticeable improvement in clarity, though it is small. This can't be mimicked with volume envelopes, but a convolution reverb would work. Most noticeable is the decay at the end. It is so much smoother than 2 and 3. I understand why but it's difficult to explain, hopefully you all understand why. If not, I can try to explain or someone else might like to. 5 - the first cancellation sample. Nothing is heard till the end where the tail was faded out. 6 - you can hear a few parts where the gate starts doing something, but it just can't get into the complexity of the reverb. 7 - you can hear all those tails getting cut off. These are what muddy the reverb more. All those tails are eliminated smoothly in the algorithm. Understand now why a volume envelope can't do this? So there you have it. I will admit that in many songs those tails would be masked anyway and simply using a well tweaked gate or volume envelopes or simply a linear decay would be fine, but in very sparse mixes, the subtleties could really make a big difference to your vocal sound. It's a technique I'm certainly going to remember for the right project.
|
spacey
Max Output Level: 0 dBFS
- Total Posts : 8769
- Joined: 2004/05/03 18:53:44
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 22:34:51
(permalink)
Matt I sure don't know how they did it but if were to try... I'd dup the vocal track and put my verb on 2nd as track effect and use a volume envelope to adjust as desired. That way I could adjust voice/effect levels too without messing with the main vocal track. And since I guess at things to try...this may be a very bad idea...
|
mattplaysguitar
Max Output Level: -55.5 dBFS
- Total Posts : 1992
- Joined: 2006/01/02 00:27:42
- Location: Gold Coast, Australia
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 22:58:23
(permalink)
spacey Matt I sure don't know how they did it but if were to try... I'd dup the vocal track and put my verb on 2nd as track effect and use a volume envelope to adjust as desired. That way I could adjust voice/effect levels too without messing with the main vocal track. And since I guess at things to try...this may be a very bad idea... Check out post #9 and #10 cause we have it worked out!
|
ShadDOH
Max Output Level: -70 dBFS
- Total Posts : 1027
- Joined: 2009/12/31 20:50:02
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 23:32:12
(permalink)
"And since I guess at things to try...this may be a very bad idea... "
LOL@Spacey!
|
Philip
Max Output Level: -34.5 dBFS
- Total Posts : 4062
- Joined: 2007/03/21 13:09:13
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 23:35:12
(permalink)
mattplaysguitar Philip - I'd agree that there is pretty strong compression on her vox and quite possibly some tube saturation too. I am supprised though that you seem to have gone as far as to say she was using an Avalon 737 (or are you just making an assumption?) with a German condensor mic and has lots of lower harmonic bleed, but you can't hear the hall verb... In my opinion the reverb is extremely wet and is very very clear. Listen to the sections where the vox ends and there is a gap, it is most obvious there. Then when the next line kicks in you can here the verb suddenly cut. Specifically, the times 21, 22, 23 seconds each have it very clear. My first impression was that a gate would work on these bits, but it won't because when the next line kicks in and the level is brought back up, the previous tail will still be playing with the new reverb which is going to really muddy the signal up. Also please elaborate on lower harmonic bleed as I have never heard of that one before and google gives me nothing! The vocal certainly sounds a lot more harsh on the youtube clip because of the lower quality, sounds much clearer on my cd. Maybe try listening to it in HD? Also are you on your monitors listening to this? Tarsier - I believe you are correct. Envelopes can't work because they don't cut the tail that bleeds into a new verb sound (I elaborated on this in the original post). But your idea on modifying a conv reverb would definately work. Excellent suggestion! Mike - I'll make an image to better describe it! 8 Ohm - I love automating reverb sends too, it works great but as I'll explain more clearly in another post, this ain't gonna work for this effect! MemphisJo - No idea what's up with that! It's just a youtube link! You can search for the video youtube if you prefer. The song is By Sarah Blasko and is called Bird on a Wire. ------------------------------------------------- Matt, "lower harmonics" *bleeding* (a Philip assumption and possible misnomer) ... when there is generalized tube saturation-excitation ... harmonics and 'colorful-distortion' may smear pleasantly into the lower EQ spectrum (which is what I meant by "bleeding"). As an audiophile, I detected her 'leveled' colorful tone and hyper-warmth, that is all. Indeed, her reverb is also lush and velvety (I just can't believe software reverb is giving that vox such warmth and color ... again, I'd 'probably' bet she is using a top-notch tube pre-amp mic pair to achieve this with outboard compression as well. I'll try to re-read this thread and listen again later as you suggest to better appreciate the rich reverb components you've validated. EDIT: Gated verb is a bit above my intelligence quotient, of late. 
post edited by Philip - 2010/05/14 23:39:26
|
bitflipper
01100010 01101001 01110100 01100110 01101100 01101
- Total Posts : 26036
- Joined: 2006/09/17 11:23:23
- Location: Everett, WA USA
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/14 23:43:32
(permalink)
When I listen to the Youtube video, all I hear is a short reverb with a long predelay. Try putting a delay before the reverb on a 100% wet reverb bus. Make the delay 100% wet, too. Lots of cool effects you can get that way, including getting away with a lot more reverb before the intelligibility of the lyrics suffer.
 All else is in doubt, so this is the truth I cling to. My Stuff
|
The Maillard Reaction
Max Output Level: 0 dBFS
- Total Posts : 31918
- Joined: 2004/07/09 20:02:20
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/15 08:01:31
(permalink)
"But your idea on modifying a conv reverb would definately work. Excellent suggestion! Mike - I'll make an image to better describe it!" For the last two years I've pretty much only used convo reverb and I've been using my own home made true stereo IRs exclusively... I guess selecting or identifying a IR by length just seems contextual to the process. Is the picture going to be of an IR? What triggers the change in duration on the Waves plugin? I'm trying to wrap my head around understanding how the Wave plugin knows when to back off. It seems like you could automate some parameter envelope within a reverb plugin to go in this direction. best regards, mike
post edited by mike_mccue - 2010/05/15 08:02:36
|
mattplaysguitar
Max Output Level: -55.5 dBFS
- Total Posts : 1992
- Joined: 2006/01/02 00:27:42
- Location: Gold Coast, Australia
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/15 09:27:43
(permalink)
If you have a convolution reverb with an impulse with a 3 second decay time, you can imagine what that impulse looks like, right? Now, you know how it has that option to put a volume envelope on the impulse itself? You can't change the decay time without it being an unnatural decay because different frequencies decay at different rates (dampening setting) so simply putting an envelope on the impulse to cut the decay down to 1second will effectively change your dampening setting and make the reverb sound different because the high frequencies won't be attenuated the same way. These are the limitations of impulses. But it may sound fine anyway. Regardless, this is still happening. Now, when you did this, you probably changed the volume slop of the impulse so that it is zero at 1 second and 100% at the start of the impulse. This won't give you a linear decay because there was already a slope of decay in the original impulse, so it's going to be more of a slow curve. Is this getting too wordy? Pictures would help, but I'll keep trying for now! So that's our attempt to change the delay time of an impulse and keep it as close to linear as possible (the longer the delay time, the easier it is to keep it linear, but the sound is all going to change a lot as you are only using the start of the reverb sound. Time squashing with a perfect algorithm would be more accurate. In my example, you say, ok, for it to breathe with the music we need 600ms of reverb. So you leave the envelope at 100% all the way to 600ms then cut it right at 600ms down to 0. Or a bit of a slop if that sounds too sudden. This will create the non linear decay which will be different to a gated reverb, which is also a non linear decay. Why is it different? Think of the sound in sections. Lets split a 100ms second vocal into 1ms parts. The first 1ms has this impulse applied to it which lasts for 600ms. The next 2ms part has a 600ms bit added to it and so on. What this means, is when the last 100ms of vocal is played, that first 1ms still has 500ms of reverb left in it and the 100ms section has 600ms left to go. Now lets fastforward to 500ms from there. The first 1ms reverb component has reached the end of it's 600ms and it cuts, but the final 100ms of the vocal still has 100ms to play out. The result? a gradual cut of reverb sections over a 100ms period. This is something a gate or volume envelope cannot mimic. The gate gets to that 700ms point, where everything ends in the orignal example. At this point, all reverb parts are playing. It then cuts them all at once. Not smooth. The reverb part of the 1ms vocal part plays for a total of 700ms and the reverb of the 100ms part only plays for 600ms. It's just not as smooth. You can hear this in my audio samples. Hope that makes some sense?!?!
|
tarsier
Max Output Level: -45 dBFS
- Total Posts : 3029
- Joined: 2003/11/07 11:51:35
- Location: 6 feet under
- Status: offline
Re:A handy vocal reverb technique - non linear decay
2010/05/17 11:49:29
(permalink)
Envelopes can't work because they don't cut the tail that bleeds into a new verb sound Exactly. Over on the "EQ of reverb" thread I posted this: everyone should read Spaces Speak. Are You Listening? by Dr. Barry Blesser of Lexicon and EMT fame. And read his whole website. And listen to spaces. Neat ones show up in the oddest places. Just thought it might be somewhat applicable here. Or not.
|