Phil O'Keefe Posted August 10, 2005 Share Posted August 10, 2005 This is a crosspost from a thread over on SSS, but I thought it deserved its own thread. Maybe someone who works on product development can do something with these suggestions. The topic was vocal doubling. Here's a link to the original thread. ********************************** Originally posted by acousticvoodoo: I really like the way double tracking was used in the early Beatles recordings. The qualities of Paul and John's voices really worked well together, and their ability to switch from harmony to "doubling" really gave them a full sound. It's one of the things the current electronics don't provide... a second, high-quality voice with an entirely different singer. michael You know, from time to time I think about this, and contemplate doing some coding and coming up with something new and different in the auto doubler world... but it would take a lot of work, and I don't think I could do everything I wanted to with a real time plug in. There's too many variables involved. Any singer, no matter how tightly they can double, when singing the same line twice, will have natural variations in several key areas: Pitch: Fluctuations can occur at any time, and do not vary in a periodic fashon, but rather, aperiodic ... which means pitch modulation, unless it's random, isn't going to work. Timing: This can also vary. You want it "tight" but not too tight or at set "delay" between the source / original voice and the "double". When sung twice, the "double" might actually preceed the original by a ms or three... (or a lot more ) or lag behind it. Again, in an aperiodic fashion that can and does change from moment to moment. Which again means that changing the delay times in a periodic manner isn't going to work... besides, have you ever changed a delay time setting on a delay that was working on a vocal? Not pretty... Amplitude and phrasing differences: Phrasing comes under the heading of "timing" (at least to my way of thinking insofar as this discussion), but amplitude can also vary from take to take, and I consider that a part of phrasing too. A little more emphasis on this word, a little less on the next pass... not to mention tonal / timbre differences, etc. etc. EVERY "doubler" hardware unit or plug in that I'm aware of uses a variation on the same theme: A delay and a pitch shifter. But they're static... you can offset the pitch by a certain amount of cents, and set the delay to such and such milliseconds, but once set, they're stuck there, and if you change them in real time while signal is beeing passed through them, it glitches. Sure, some of the units / plug ins might modulate things, but it's usually a sine wave modulation of pitch or delay time or whatever... IMO, too periodic and predictable to sound really authentic. I just don't have the time to mess with it, and frankly, as I said, I think it might be difficult to accomplish... but if a plug in company wanted to have a go at it, and could do something that took ALL of those factors into account and offered user control over the randomization of various parameters, IMO, that would be cool. Feel free to PM me if any of you reading this are a software developer and you want any clarification about what I'd like to see and / or my input / suggestions. ********************************* So, who's up to the challenge? Build a better / more realistic doubler, and the world will beat a path to your retail outlets. Link to comment Share on other sites More sharing options...
Members Kiwiburger Posted August 10, 2005 Members Share Posted August 10, 2005 I think Antares would say the've already done it with their AVP-1. In PC VST land, CloneEnsemble does very good doubling and more. Randomising pitch, formant, delay etc There are numerous VST modulation plugins - Kjaerhus GMO-1 is a favorite of mine. It can do pitch, delay and amplitude variations with a choice of waveforms, and can sync to host. So you could have things subtley changing over several bars if you like. Phase offset too. No need to burn your thumb on the flange anymore ... Link to comment Share on other sites More sharing options...
Members Kiwiburger Posted August 10, 2005 Members Share Posted August 10, 2005 You could also try automating your existing plugins by hand drawing in a wobbly curve with your mouse or controller - not too different from the random element of manually twisting the varispeed on a tape deck. I would love to see a very advanced delay plugin that had random lfo's. There are plenty of filters with random sample-and-hold type effects. My gripe about most plugin delays is that they often are strictly limited to sync-to-host-tempo. That's useful, but limiting - especially for doubling, where you want to specify the time in milliseconds, or even 10ths of a millisecond. My prefered interface for adjusting delay is a virtual 8 segment LED display, where you can scroll each digit up or down. So if the display reads 1234.8 - you could very easily change this to 1234.7, or 234.8 with a single mouse click. Sync to host should be available but optional. With phase offset plus/minus 180 degrees. For doubling, a random waveform would be cool so long as the delay time didn't stray too far from the sweet zone for doubling. Maybe a pseudo random waveform. Link to comment Share on other sites More sharing options...
Phil O'Keefe Posted August 10, 2005 Author Share Posted August 10, 2005 I think Antares would say the've already done it with their AVP-1. Nothing aganst Antares, but I don't think that does what I'm talking about. I have not spent any time with a AVP-1, so I could very well be incorrect, but in listening to the audio samples on their website, it sounds like AT pitch shifting / doubling, which is definitely not what I'm talking about. No delay time modulation / variation. The time difference between the attack of the source and the attack of the doubled voice doesn't vary from moment to moment. There is also no variable / random pitch fluctuation, independent of the "source" vocal, other than probably a simple LFO type setup. Not random enough for what I'm talking about. IOW, I want something that when the pitch of the source goes UP a few cents, the double is just as likely to intonate DOWN a few cents as it is to intonate upwards in sync with the source vocal. It's got to be random, but adjustable. Of course, the amount of deviation should be a user controllable parameter. Maybe a "Pitch: Tighter Looser" knob / control. Go tighter for less deviation from the source (lower probability of deviation) or looser for greater deviation / randomization. Or better yet, two controls: One for deviation range (for pitch, in cents) and a second to control the likelihood that a variation will occur. And a similar set of controls to cover the time deviations. PS Antares: You might want to use a different example for #2: Female vocals. I can hear the AT "working" on the word "are". But don't take this as a harsh criticism... AT in auto mode is far more likely to be audible than the pretty amazing results graphic mode makes possible in a skillfull pro's hands. And of course, there's no way to put "graphic mode" into a real time hardware unit. Link to comment Share on other sites More sharing options...
Phil O'Keefe Posted August 10, 2005 Author Share Posted August 10, 2005 In PC VST land, CloneEnsemble does very good doubling and more. Randomising pitch, formant, delay etc Just looking at a picture of the controls, this looks like a possibility. I'll have to give it a try. Thanks for the suggestion! Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.