Editing and improving chamber choir recordings

Hello,
I am looking for recommendations on how to edit choir recordings. The goal is (of course) not professional mastering, nor the transformation of a lay recording into CD quality … What interests me is what meaningful possibilities I can try out. I’ve googled a lot about this, but didn’t find much on this particular topic.

Initial situation: Recording of a semi-professional chamber choir (24-28 singers in two rows) with a cappella program in a church room. Depending on acoustics and room XY or AB with a Tascam DR 40, no other microphones (I would have them, but I almost never use them at the moment … and I am not concerned here with the question of miking either). Recorded with 44.1/24bit wav. I usually place the device about 7-10m away from the choir in the middle behind the conductor and control it moderately (better a bit lower, at least never so that it overdrives). This usually means that I have less room acoustics, but the choir is very present, which is what I want.

The result of the recording is basically always quite good. At the moment I use Audacity (under Linux) only for editing. As a rule, I can hear the recordings “quite well” on all playback sources (good headphones, good stereo system, moderate headphones, car radio …) - by this I mean that, as a rule, the dynamics are such that I don’t have to keep on playing around because it’s too loud or too quiet …

What I would like to know is which editing possibilities I could try out that positively and audibly influence the choir sound or the ambient sound or other things (dynamics, bass, treble, etc.) - without destroying or overdrawing anything. First of all I ask for general recommendations. What about compression? Does that make sense? Or reverb? And if so, then which one, how much, and what can you concretely recommend?

I would be very happy about constructive tips. Here is a link to a sound example: https://we.tl/t-tzTtyZxQcV
Thank you & greetings,
Eriol

I’m good with that.

Any compression or other processing you do is going to make the room sound funny. That leaves you with minor tone controls and even that could be overkill.

There is a minor haystack tonal bump between 1900 and 3500 and I think that would be my first correction. That seems to be popular with some microphone systems to artificially increase the crispness or presence.

Both recordings have “stuff” below 20Hz and could use a high pass filter, but you shouldn’t be able to hear that.

Koz

There is a minor haystack tonal bump between 1900 and 3500 and I think that would be my first correction.

Hi Koz,

thank you for your answer. Could you tell me what I could do in this respect - I mean, which functions or tools from Audacity I could use to play with in order to influence this issue (minor haystack tonal bump between 1900 and 3500)?

Another question: I have seen that some people recommend also to “de-noise” the recording. However, my feeling is that there is not too much background noise and if I completely remove it it may sound too static/artificial/dead …

Is there anything else you could recommend to try out?

Thanks again!
Eriol

Is there anything else you could recommend to try out?

No. When you started your post I imagined the performances in my head and that’s exactly what the samples sounded like. It’s not unusual for people to assume they need to apply tons of effects, filters and corrections. If you accurately captured the performance, just leave it alone.

It’s Effect > Equalization. A pre-baked curve will not be available for this. I’ll see about designing a correction based on your samples.

You can certainly capture a pure, sterile performance in a studio, and partially simulate it with aggressive noise reduction, but we know going in that you are in a church during a live performance and that’s exactly what it sounds like.

As we go.

Koz

Thank you so much, you are a great help! And do not get me wrong, I am not seeking for tons of effects; as I wrote, I like what I got from the Tascam recorder … I just want to better understand what I could try, and to understand if there is any filter or so that could be useful at all. And you exactly pointed out for something that I have never tried before:

That would be great. I will try to read myself in the use of this effect as well in order to better understand what it means.

Again, thank you for your great support!

Eriol

Open one of the performances.

Effect > High Pass Filter: Frequency 20Hz, Roll-Off 12dB > OK.



That will get rid of the earthquake tones. Recorder error??? It’s pretty common.



Download ChoiraCapella.xml (attached).


ChoiraCapella.XML (452 Bytes)
Effect > Equalization > Save/Manage Curves… > Import. Point to ChoiraCapella.xml > Open > OK.

In the equalization main window, Select Curve: ChoiraCapella. It should look like the illustration > OK.



Once you install a custom curve, it should stick and you only have to install it once, but use it as many times as you wish.

If you like the loudnesses of the songs, then you’re done.

Others may post. Processing a music performance can be a career move as you have already noticed.

Koz

You can experiment with (narrow/widen) the stereo-image,
e.g. making it more mono (narrower) can improve intelligibility.
https ://wiki.audacityteam.org/wiki/Nyquist_Effect_Plug-ins#Channel_Mixer

Could split into mid & side channels and equalize those separately to attenuate the loud sibilance.

Thank you! I have applied this. I believe adapting the equalization curve indeed improves the recording. Is there any “manual” on how to modify/equalize recordings in general (and maybe even specifically choir recordings)? I am still a bit struggling whether ChoiraCapella is a bit too much, it affects especially the higher voices (mostly the tenors, which I must admit is something where I am highly prejudiced because I am singing tenor in this choir … But this is indeed something to play with; I need to better understand this, though, in order to become more efficient than just trial and error …

Definitely! Koz, would you recommend to modify loudness? As I wrote, I have been a bit conservative (the Duruflé is low level) but the overall balance of the recording is correct to my opinion. This being said, one could maybe increase the overall loudness - what exactly would you recommend “to play with” in this respect?

This is indeed interesting and I will try this out. My feeling is that stereo in this recording is already narrow (I recorded AB here) but I need to apply this.

Could you give me some guidance how to apply this exactly?

Thanks to both of you for your great tipps! I will have a lot to try and understand better now.

Eriol

In Audacity on Windows there are free plugins where you can adjust the stereo width in real-time, e.g.

In Audacity on Windows this can be done in one step using a free plugin where the equalization of mid & side can be adjusted separately in real-time … Graphic Equalizer Plugin, 7-Band, Harmonic [VST, AU, AAX] - Overtone GEQ - Voxengo
[ I would cut the treble on the side channel to reduce the sibilance ]

It is very highly recommended that you save WAV copies of all the work in a safe place. You don’t want to be the person who made an audio correction and the computer went into the mud…and took the show with it.

This being said, one could maybe increase the overall loudness

I did exactly that to make the tests more obvious when I created the correction—at least on Notre Pere.

Select everything (button on the left) > Effect > Amplify: New Peak Amplitude -3dB > OK.

You should see the blue waves jump up louder.

You can modify ChoiraCapella correction. Attach 1 is the correction as I wrote it.



Attach 2 is one of the control points dragged to a new place. Just click and drag.



The goal is to move each control point straight up half-way. That will reduce the overall effect by about half. The first and last points are at zero and don’t move. Half of zero is zero. The next last point is going to be rough to tell. Guess at it, or just move it to zero. This point isn’t that important.

Each of those points is a slightly different musical tone, but you’re right, they center about the Tenor range.


If you’re not far enough into the weeds yet, I used Analyze > Plot Spectrum to tell me what to do.

Spectrum 1 is the original clip. Musical tones are along the bottom. Bass tones to the left, louder is up. It measures the whole performance or whatever you have selected. Try to make the four bottom window settings match to get the same view.

The tones to the left of about 30 are thunder and earthquakes and usually cause problems. That’s what Effect > High Pass Filter fixed.

Look at the waves just up and to the left of 4000.

See the little haystack there that doesn’t seem to follow the general shape or slope of the rest of the wave?

That’s what ChoiraCapella fixes.

If the show sounds better with the haystack, then either leave out the correction or change it slightly.

The peak way on the right is SS sounds, lip ticking or crispness. You can design an equalization to work on that, too.


Clicking the little control points works, but you can switch Equalization to “Graphic” mode with sliders. That’s more how a music store equalizer would work, but note the actual correction isn’t quite the same. Experiment.

Koz

You might consider preparing MP3s for the performers and audience before you start these corrections as you may be doing them for a long time.

Koz