Ableton Live experts: how do you "sharpen" voices
January 17, 2014 4:23 PM   Subscribe

How can I make slightly muffled voices on audio conference calls sound sharper or crisper using Ableton? Is their an audio equivalent of "unsharpen mask"? Bonus points if I can do it over a specific frequency range.

This post describes 1-3 mornings a week of my life. In order to improve the quality of my conference calls I've been running the audio through Ableton Live. The compressor helps balance the whisperers with the people who appear to be shouting into a $1 mic while the equalizer helps me manage levels on specific voices.

I haven't found a solution for improving slightly muffled voices though. Any ideas? Any other tips for improving vocal audio quality over sketchy international lines?
posted by stp123 to Computers & Internet (6 answers total) 4 users marked this as a favorite
Best answer: To answer this question, consider what "slightly muffled" actually means: it means that the high frequencies are attenuated. High-frequencies, which include sibilance (sssssss), are vital to the understandability of speech. You can't just boost them using the EQ because they might not be there in the first place, plus there'll be a lot of noise there you'll be boosting as well, though that might be a place to start.

What you really need is an exciter (named for the Aphex Aural Exciter, a piece of actual audio hardware). Basically the exciter synthesizes high frequencies by pitch-shifting the available mid-range frequencies up an octave or two (to create harmonics) and also adding some high-frequency noise (shaped with an envelope follower on the mid-range frequencies). Then some distortion and phase shifting are added and some dynamic range compression applied.

Unfortunately Live doesn't seem to have a true exciter, though I did Google up a recipe for creating one using the saturator. This doesn't sound to me like it actually synthesizes the high frequencies, but it might help if you already have some high-frequency content to work with. You can get VST plug-ins that implement more true exciter functionality; you should be able to use these with Live, I think.
posted by kindall at 4:58 PM on January 17, 2014 [5 favorites]

You could use a gate on each track to mute extraneous noises when the person wasn't speaking. You'd have to set the threshold based on your audio levels.

You could set an EQ to high-pass everything at around 150-200Hz -- that should clear up some of the muffle. Adding a couple of dBs around 3k should help with speech intelligibility, and you could do a slight high shelf starting at around 6k if you want more high end still.

The ThrillseekerXTC is a free VST exciter that could be good for adding some high end too.
posted by ludwig_van at 4:59 PM on January 17, 2014

It's going to be hard to create enunciation where the data isn't there. However, the ear is a good machine for filling in the blanks, especially with voices. Start by cutting the bass, boosting midrange, and playing with treble. Compress LAST.

That said, it sounds like this is kind of a people problem. They should be speaking up, using better equipment, and as an awesome extra would be recording locally so that they could send it to you so you can mix in clear voices to the master recording.
posted by rhizome at 5:07 PM on January 17, 2014

This is a stupid blogspam post, but it DOES list a bunch of exciter VSTs.

That's you want, a plugin like one of those. I would honestly just google "free exciter VST" and try every single one.

You wouldn't believe how many times i've solved an audio processing/effects/synthesis problem that way. I've also created some very entertaining/interesting/unexpected results.

I will always maintain that "wait, how did i just do that?" is the best way to make audio related things work.
posted by emptythought at 7:01 PM on January 17, 2014

What kindall said, but in a pinch you can try backing off the low and mid end a bit (as opposed to pumping the high end) with the EQ, then raising the overall volume.

Also, you're allowed to speak up and say "before we begin, can everyone make sure they're near a microphone? I often cannot hear the conversation otherwise." Then follow it up with a polite "excuse me, I'm having trouble hearing you" when they forget. I do this regularly (doesn't help that I'm hard of hearing) and nobody's ever given me a hard time over fact they seem to respond well, I assume out of courtesy and because it's me basically telling them I am interested in what they have to say.
posted by davejay at 10:30 PM on January 17, 2014

Would they be offended if you ordered a $20 USB microphone on Amazon and had it delivered to them?

Garbage in, garbage out. It's very difficult to get crisp audio from sloppy audio. Unsharp Mask will make a good photo pop, but it won't fix an out-of-focus photo.
posted by scose at 9:01 AM on January 18, 2014

« Older Horror movies from where the only surviving victim...   |   Table top brainteaser puzzles Newer »
This thread is closed to new comments.