Complete Classical Music workflow

[bradhurley (Bhurley)]

For any decent ensemble that rehearsal around about the same levels as they perform, I’ve found -12 dB peaks work for me.

Nice, thanks, I’ll give this a try next time. These days I’m mostly accompanying a singer and she does have a tendency (which is common, of course) to sing much more loudly when recording or performing than when doing a level-check, but I’ve been perhaps too conservative.

[Aleph]

I asked a colleague yesterday about a decent (but not super expensive) microphone pair to record flute and she seconded my idea of the sE8 pair (essentially a KM184 clone). Given that the first thing you should probably do is have matched or same model microphones, these would be relatively inexpensive and produce great results. She also recommended whatever microphones you use to experiment with ORTF plus an extra spot microphone aimed at the end of the instrument either in front or directly in line with the body of the flute. In a good room you might be able to get away with just the ORTF pair.

Thank you! I think I can afford the price of the sE8, 370€ here. However a third mic is impossible in my Audient. The Audient iD14 only has inputs for two mics, expandable to 10 through ADAT or S/PDIF.

Years ago in the only classical-style recording setup I’ve played in, I played with a harpist and the mics were set up quite a large distance away (we were recorded by some classical recording engineers from WGHB radio in Boston) but only slightly above my eye level.

Here is an interesting comparison about flute mic position with examples. My space, however, is much smaller.:

http://soundmedia.jp/nuaudk/05Fl/index.html

Mi position is more or less the A, but a bit wider.

[anon60445789]

The Audient iD14 only has inputs for two mics,

Oh yes, I forgot you only had two inputs. Enjoy with just ORTF or other near-coincident array!

Here is an interesting comparison about flute mic position with examples.

This great. Shame there isn’t a showing of cardioid arrays (ORTF, NOS, DIN, XY etc) in addition to the omni pairs. For showing how the sounds changes based on microphone location, however, this is fantastic (with excellent product placement for DPA )

[anon60445789]

@Aleph another quick thought. I measured your Density 21.5 in more depth and found LRA = 23.09 LUFS

While there is no set limit for classical, I’d suggest your LRA of 23.0 LUFS is probably on the wider end of acceptable dynamic playback in a typical listening environment. Certainly there are points where I struggle to hear the quietest moments where I listen to most of my music. A little manual gain-riding or transparent compression may be desirable but I’d wait for now and there are certainly classical albums that maintain this kind of LRA.

Also note that at -13.86 LUFS integrated, your track would be turned down on Apple Music streaming but almost perfect for YouTube and others. Not really a problem though given you haven’t compressed the heck out of the track. It might mess with the balance of your album though if Apple goes on a track-by-track basis.

If this is indeed your loudest track (integrated) I’d certainly consider turning it down between 4 and 5 dB while reducing the LRA a bit. If you take both Syrinx and Density 21.5 together, you could reduce everything by about 3dB and be in the -18 LUFS integrated ballpark. I take the loudness stuff seriously but I admit that I may be a little conservative. For certain, wait until you have finished recording the whole album before making decisions about dynamic range and loudness levels.

As a rough comparison, I just ripped Bach violin sonatas disc 1 from Faust/Bezuidenhout (Harmonia Mundi) and found integrated at -19.13 LUFS with an LRA of 8.89 LUFS. Peak was -0.21 TPFS. For Mahler 1 conducted by Boulez (DG) I found integrated at -16.94 LUFS, LRA 23.52 LUFS and, surprisingly, a peak of +1.63 TPFS!! If you bring that down, we get a corrected integrated value below -18 LUFS. I also note that the 3rd mvt has integrated of -33.19 LUFS. Gergiev conducting Mahler 1 (LSO live) gives -20.94 LUFS, LRA of 23.11 LUFS and peak of -0.07 TPFS.

As a side note, my wife often complains when we listen to Mahler symphonies in the car. There seem to be long stretches where the traffic and engine noises completely drown out the music and we are left reconstructing/guessing in our heads (prize for the passenger who correctly gives the downbeat on the next audible passage).

[Aleph]

For certain, wait until you have finished recording the whole album before making decisions about dynamic range and loudness levels.

Thank you @anon60445789 for your analisys! Then I’m going to wait to buy the sE8 pair and start everything from scratch. I think it’s going to be the best.

What tools do we have in linux to do that kind of analysis? I use Ardour’s analysis when I export the piece, but I don’t see any files being saved. I’ve also used Klangfreund’s LUFS Meter plugin, but it’s in real time, and online at https://youlean.co/online-loudness-meter/, but it’s not very complete.

Here’s another good video with a comparison of stereo techniques. For me the ones I like the most are the spaced pair and the ORTF.

prize for the passenger who correctly gives the downbeat on the next audible passage

Very nice. Yes, listening to classical music in the car is impossible…

[anon60445789]

What tools do we have in linux to do that kind of analysis?

Ardour’s built-in analysis is excellent. In the Ardour preferences you can have it save the analysis graphics if you want to keep a copy without doing a screenshot. For realtime, Klangfreund LUFS meter is indeed excellent (and the even more complete Multimeter which also allows for offline analysis) but don’t forget x42’s meter collection which contains every possible meter you could ever need.

For command line I can recommend installing the ebumeter package via apt or your GUI package manager. It includes both a realtime meter and, more importantly for our purpose, ebur128 which will give you all the cool analysis. I recommend running it as follows:

ebur128 --lufs --full <input file>

It seems it can only do one file at once but, of course, a little bash scripting would work wonders. In any case, it seems a lovely way to check a wav+cue export. If you don’t use the --full option, it will just give you Integrated, LRA and true peak level which in many instances is all you need.

EDIT: I think I can rattle off a script to return the values of the whole album given a set of individual tracks. Basically I would write integrated, LRA and peak for each file into a three-column text file via loop and then use awk output to run mean on integrated and highest value on LRA and peak. Is there a more efficient way?

[Aleph]

Is there a more efficient way?

I don’t know but this sound great! If you can do it could be fantastic!

[anon60445789]

Here we go with the album loudness script (requires the ebumeter package):

https://pastebin.com/bvbPNkq4

It’s probably really inefficient but it seems to work fine
Save the script to somewhere like /local/usr/bin and check out the generated html table file (loudness_analysis.html) and screen output with individual data on each track plus whole album analysis. HTML file is generated in whichever directory you launch the script. Here’s an example PDF from one of my own albums:

loudness_analysis.pdf

Usage: album-ebur128 [files]

e.g. album-ebur128 *.flac

EDIT: Independent of the efficiency, I’ll probably tweak the output slightly over the coming days. If there is anything else that might be useful, let me know. I believe I need to seriously re-think the calculation of the integrated and LRA as I don’t think running the mean of the individual tracks (or max of the LRA) is the same as running the whole album wav in one go (probably because of the relative loudness gates?). Hmm… @x42 any ideas about how I can achieve this only with ebur128?

[Aleph]

Save the script to somewhere like /local/usr/bin

Hello @anon60445789 your script looks great! However, it’s not working for me.
I think the path is an error, it should be usr/local/bin not local/usr.
I added the file to: usr/local/bin/album-ebur128.sh
I also made the file executable unsuccessfully.

The terminal output is:

~/Music/CD Wav
album-ebur128 *.wav
bash: album-ebur128: order not found

Of course, ebumeter is installed.
I also checked all the file names, because Ebumeter doesn’t support accents or spaces.

[anon60445789]

I think the path is an error, it should be usr/local/bin not local/usr.

Of course, definitely usr/local/bin. By the time I finished the script and typed the response here it was almost 1am.

I also made the file executable unsuccessfully.

Did you do the following? (need to use sudo or root if you are trying to do this while in usr/local/bin`):

chmod u+x album-ebur128.sh

Alternatively, You can use the GUI file manager before you move the script to /usr/local/bin. Right-click on the file, go to properties and ensure that “Allow this program to run as a program” is ticked.

FYI, I never bother naming a script with .sh at the end. Bash doesn’t need it there, as long as the script has been made executable.

So, in summary, I would rename to album-ebur128, use chmod or file manager to make it executable, and finally move it to usr/local/bin. You should then be able to run the command from any directory you like.

[Aleph]

No way, sorry.

Edit:
As a clue, could it be any incompatibility with Spanish characters?

[anon60445789]

No way, sorry.

Are you saying it is still not working?

When I run ls -l on the file I get this:

-rwxr-xr-x 1 root root 1764 Nov 3 07:21 /usr/local/bin/album-ebur128

[Aleph]

still not working

When I run ls -l on the file I get this:

Sorry, how to rum ls -l on the file?

[anon60445789]

Open a terminal and type:

ls -l /usr/local/bin/album-ebur128

[Aleph]

ls -l /usr/local/bin/album-ebur128

$ ls -l /usr/local/bin/album-ebur128
-rwxr-xr-x 1 root root 1824 nov 3 09:40 /usr/local/bin/album-ebur128

[anon60445789]

And when you first open up the terminal you cannot start typing alb… and TAB completion to get the full command?

[Aleph]

Yes I can but the output is one more time

album-ebur128 *wav
bash: /usr/local/bin/album-ebur128: /bin/bash^M: intérprete erróneo: No existe el fichero o el directorio

[anon60445789]

Can you type which bash in the terminal and tell me what it says?

[Aleph]

luis@antix1:~
$ which bash
/bin/bash
luis@antix1:~

Could it help?

bin/bash: bad interpreter: No such file or directory

[anon60445789]

/bin/bash^M: intérprete erróneo: No existe el fichero o el directorio

OK, I think I know the problem. the /bin/bash^M is saying there are Windows/DOS style line breaks. This is extremely odd because I used Featherpad on MXLinux to create the script and I don’t have problems here.

Apparently, you can install dos2unix and run sudo dos2unix /usr/local/bin/album-ebur128 and see if it fixes it? How did you create the file from the pastebin link I gave you?