question for Selig/audio RE developers

Need some fresh sounds? Want to show off your sound design skills? Here's the place!
User avatar
guitfnky
Posts: 3291
Joined: 19 Jan 2015

Post 25 Apr 2021

this is a question for Selig or other audio developers, but not really specific to RE development per-se (so, mods, feel free to move this wherever appropriate, if I'm doing it wrong)...

the reason I'm hoping for Selig's input specifically, is because of his de-esser RE--but of course, anyone's insight would be super helpful!

the basic question is, where did you source your test vocals? is there some kind of resource devs use for that sort of thing? (or did you just have a variety of stuff available from past projects?) I've tried googling 'male/female sibilant vocal examples' and the like, and I get a lot of youtube results, and explanatory blogs, but I was hoping to find examples I could use for testing, instead of tutorials on how to de-ess.

a year or two ago, some may remember I created a de-essing combinator, and there was a fair amount of discussion around it--one of the things that Selig said that really stuck with me is that it's a lot easier to create something like that for a single voice/instrument than it is for the general public. basically, everyone's voice is different, and my de-esser may sound awesome on my voice, but terrible on someone else's.

fast forward to today, and I've been using my combi on my own voice, and it continues to work really well in Reason, but I've recently started using Ableton Live Suite, and wanted to try my hand at programming in Max For Live--I figured, what better way to learn than to try to recreate my de-esser as an M4L plugin. I've got it working, and again, sounds pretty good on my own voice, but I want to see how it handles a variety of other voices, so I can hone in on the different setting ranges, etc.

PhillipOrdonez
Posts: 1939
Joined: 20 Oct 2017
Location: Colombia

Post 25 Apr 2021

There must be a plethora of badly produced podcasts out there with plenty of material to practice with.
Perpetual license holder

User avatar
guitfnky
Posts: 3291
Joined: 19 Jan 2015

Post 25 Apr 2021

PhillipOrdonez wrote:
25 Apr 2021
There must be a plethora of badly produced podcasts out there with plenty of material to practice with.
haha, that's a great point, and might end up being my backup plan. :lol: hoping to find something I won't have to manually search around for and extract audio files from though.

User avatar
selig
RE Developer
Posts: 9156
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

Post 25 Apr 2021

I have decades of projects to choose from, across a wide range of voices/styles/experience. I just selected around 30-40 (can’t remember exactly) and selected a line or two that had sibilance from each. Since I did not take a “threshold based” approach to my de-Essex I didn’t have to worry about levels, so I just left the level of the vocals alone. Some were more compressed (during recording) than others, which also turned out not to be a problem for my approach but should be a concern if using de-easing techniques that are threshold based.
Check out sites like this for additional material if needed:
https://cambridge-mt.com/ms/mtk/
Selig Audio, LLC

User avatar
guitfnky
Posts: 3291
Joined: 19 Jan 2015

Post 25 Apr 2021

selig wrote:
25 Apr 2021
I have decades of projects to choose from, across a wide range of voices/styles/experience. I just selected around 30-40 (can’t remember exactly) and selected a line or two that had sibilance from each. Since I did not take a “threshold based” approach to my de-Essex I didn’t have to worry about levels, so I just left the level of the vocals alone. Some were more compressed (during recording) than others, which also turned out not to be a problem for my approach but should be a concern if using de-easing techniques that are threshold based.
Check out sites like this for additional material if needed:
https://cambridge-mt.com/ms/mtk/
awesome, it looks like this should be a huge help, once I figure out the filtering options. thanks, Selig!

mine is definitely a threshold-based setup, and that's partly why I want to go through this exercise. I'll need to see how things behave with different levels, sibilant problem ranges, attack, decay times, and the like. now that you mention it though, I might consider a pre-gain knob ahead of the actual de-essing processor so it's easy to get things into a good working range, so I can minimize the necessary threshold range.

I'm thinking about tackling it by getting my track playing at a good level, then dialing in settings that sound good for that voice, and noting the different values I had to change. then doing that for as many other voices as I can, without going too crazy. then, I'll dump all that into a spreadsheet, find my min and max values, and build those into the plugin (of course adding a little buffer on either side).

User avatar
mcatalao
Posts: 1456
Joined: 17 Jan 2015

Post 28 Apr 2021

Mike Senior's book is one of the "mixing manuals" that I've carried over the years, it's quite practical, touches A LOT of stuff from the whole production process and the material you linked has been a great reference for mixing and keep learning stuff!

  • Information
  • Who is online

    Users browsing this forum: CommonCrawl [Bot] and 0 guests