PreSonus Blog

Friday Tip: Better Vocals with Phrase-by-Phrase Normalization

Unless you have exceptional vocal control, some vocal or narration phrases will likely be softer than others—not intentionally due to natural dynamics, but as a result of sketchy mic technique, running out of breath, or not being able to hit a note as strongly as other notes. Using compression or limiting to even out a vocal’s peaks has its place, but the low-level sections might not be brought up enough, whereas the high-level ones may sound “squashed.”

A more natural-sounding solution is to edit the vocal to a consistent level first, before applying any compression or limiting, by using phrase-by-phrase gain changes that even out variations. The advantage of adjusting each phrase’s level for consistency is that you haven’t added any of the artifacts associated with compression, or interfered with a phrase’s inherent dynamics. Furthermore if you do add compression or limiting while mixing, you won’t need to use as much as you normally would to obtain the same perceived volume and intimacy. A side benefit of phrase-by-phase normalization is that you can define an event that starts just after an inhale, so the inhale isn’t brought up with the rest of the phrase.

Ready to tweak that vocal to perfection? Let’s go.

  1. Open the vocal event in the Edit view, and open the Audio Bend view.

  1. Click on the Event, and choose Action > Detect Transients. Then click on Remove Bend Markers to start with a clean slate. Your event will look like the above screen shot. (Note: If the vocals have phrases that are separated by spaces, you can choose Transient Detection, Standard Mode, and then click on Analyze. Lower the threshold so that the Bend Markers fall only at the beginning of phrases. However, you’ll may need to move, delete, or add some markers with complex parts, which is why I find it easier just to place Bend Markers where needed.)

  1. You can now close the Audio Bend view if you want more room for the waveform height. Choose the Bend tool, and click at the beginning of each phrase to add a Bend Marker. If a section that needs to be adjusted starts in the middle of a phrase, you can add a Bend Marker before the section that needs tweaking anyway, even if there isn’t silence (we’ll explain why later).
  2. Once you’ve separated the phrases with Bend Markers, select the event in the Edit view by clicking on it with the Arrow tool. Then, choose Action > Split at Bend Markers. Now each phrase is its own event.

  1. Click on an event, and then adjust the gain so the event reaches the desired level. Do this with each event that needs tweaking—done!

 

Note that if audio continues before and after the Bend Marker so the Bend Marker can’t land on silence, Studio One generally handles this well if you place the Bend Marker on a zero-crossing. But if an abrupt level change causes a click at a transition, simply crossfade over it by dragging the end of one event and the beginning of the next event over the transition, and type X to create a crossfade. Adjust the curve for the most natural sound. In extreme cases, fading out just before the click and fading in just after the click can solve any issues.

So why not just do this kind of operation in the Arrange View? Several reasons. First of all, the Edit view is a more comfortable editing environment. But also, sometimes detecting transients will place the Bend Markers accurately enough that all you need to do is split and change levels—it’s much easier than doing a series of splits in the Arrange view. And if you count keystrokes, clicking to drop Bend Markers that define where to split and doing all the splits at once is easier than clicking and splitting at each split. Finally, while in Edit view, you can take advantage of the Bend Markers to adjust phrasing.

While this is a highly effective technique (especially for narration), be careful not to get so involved in this process that you start normalizing, say, individual words. Within any given phrase there will be some dynamics that you’ll want to retain—never lose the human element.

  • I’ve recently come across from Protools to Studio One. I’ve always done the same technique with the clip gain line. It takes longer than Vocal Rider or heavy compression but the reward is a much more professional vocal performance. You can put every word and every breath exactly where you want and accentuate certain words if the vocalist didn’t already. Therefore using much less compression later on but still getting a more intimate sound.

    Thanks, this article helped me move this technique over to Studio One 😁👍

  • Skyline UK

    I use the Arrange view, split off into separate events where I need to adjust and then use step 5. What am I missing?

  • Eunan James

    Craig,

    Thanks for that. I’ll keep an eye out for it. I found that using vocal rider on ‘smooth rider’ , then assessing the track, making any necessary tweaks to the the volume automation that VR might have missed, and adding a small amount of compression, seemed to work better than anything I’d managed before. I try to use as little compression as possible.

    But I’m just running a home studio for my own projects; I’m not big league or anything like that, so I’m grateful for any left-over crusts! Do you mind my asking, what settting you were using in VR?

    Many thanks,
    Eunan

  • Anderton

    I’d really like to thank all the people who are contributing their expertise, I think that’s what makes these tips valuable.

  • Anderton

    Interesting you should mention that, because although I found that Vocal Rider wasn’t as effective for my needs as this technique, I did come up with a process that involves the Vocal Rider and limiting that works really well for narration. It’s more complicated than using only the Vocal Rider, and there’s a specific Vocal Rider preset that’s needed for it to work properly. When I told the folks at Waves about it and included an audio example, they were sufficiently impressed that they’ve asked me to do a video about this process, which will end up on their site at some point. It sort of splits the difference between the compressor approach suggested by Dan Bires, and the more labor-intensive Friday Tip.

  • Anderton

    Yes, the Distressor is definitely cool. However, I find that working with vocals is different compared to working with narration, where I’ve found this technique to be most useful. And of course there’s no compression, so there are zero artifacts. But the other cool aspect is you can do things like not include breath inhales when you define the region to change gain, or pick up specific consonants when intelligibility is crucial. Granted, this kind of “manual labor” takes *much* more time than inserting a compressor or limiter, so it’s important to decide how detailed you need to be to get the results you want.

  • Eunan James

    Hi there and thanks for all those wonderful tips.

    If I may just add that I’ve used Waves Vocal Rider recently on voiceovers and vocal tracks in music, and was surprised at how good it was. It took the pain out of editing sections, and worked up a level, uncompressed vocal in jig-time. I had read some mixed reviews about it, but Paul White in SOS gave it a big thumbs up, and that was good enough for me to try it out. It’s fantastic if used sensibly.

    It might be the solution to manual editing, which can be a real time-thief.

    Many thanks for your wonderful tips Craig; they always elucidate new ways forward.

  • Dan Bires

    put a distressor plugin on your vocal track. set it to opto 10.1. then dial the attack to 10 and release on 1. You will barely hear compression and lower vocals will come up naturally in the mix. time saver. The distressor is an amazing compressor for vocals.

  • Gregor Beyerle

    don’t forget to summarize all this into a macro for easy hotkey access!

  • Jeff Evans

    Craig brought up something important as well regarding using the VU for narration and for vocals. Even when you get everything looking great on the VU for vocals eg singing, there may be some areas that could sound a little unnatural. eg sudden changes in level through a phrase. So use your ears as the final judge. Sometimes you have to make some fine adjustments to get things sounding natural again.

  • Anderton

    The VU meter is a good tip, especially for narration. I find that with vocals, if you’re dealing with a lot of sections, you need to adjust mostly by ear to preserve the intended dynamics. I’ve found that if you don’t need to modify a lot of clips, then you don’t need to use transient detection. Transient detection becomes more relevant as you increase the number of splits, or need to vary phrasing while you’re at it.

  • Brent (Doc)

    Great tip,! i do this all the time, great tip for increasing speed of workflow ! thanks !
    Increase/reduce threshold for number of markers, hit bend tool to add/subtrack where needed, and split, adjust done, thanks bro !

  • Jeff Evans

    A good thing to do is insert a VU meter over the vocal track and set it for a ref level. The VU is excellent and was originally designed for voice so it responds well. Watch the VU as you edit the individual sections of the event. (aiming for all phrases to average around 0 dB VU) I do this but don’t use transient detection. I find you can make less cuts in fact. Often several phrases will be at the right level anyway (check with the VU meter) and you can often get away with less cuts in fact. Just the very loud or quite bits. You will get better at judging the height of the waveforms too in time so they match up better. You can only do this too BTW because the waveform actually changes when making gain handle adjustments.

  • Anderton

    I find the Edit view is a more comfortable editing environment. But also, sometimes detecting transients will place the Bend Markers accurately enough that all you need to do is auto-split and change levels—it’s much easier than doing a series of splits in the Arrange view. And if you count keystrokes, clicking to drop Bend Markers that define where to split and doing all the splits at once is easier than clicking and splitting at each split. Finally, while in Edit view, you can take advantage of the Bend Markers to adjust phrasing.

  • Charles Dickens

    Why not just use split event or split event with crossfades and then raise or lower the volume.? Is there a difference in the sound? Does using the bend marker find the trnasients better?

  • Anderton

    Correct; technically speaking, normalization is about applying a *constant* amount of gain to reach a target level. In this case, because each phrase is isolated, they require different amounts of gain but the goal is still to reach a target level (average rather than peak). I do find Edit view to be a more comfortable environment for doing this than Arrange view, although it works in either one.

  • Not what I’d usually call ‘normalization’, but a very useful approach. I’ve tried doing it in Arrange view, with mixed results.