Music Therapy - 2nd and 3rd Logic sessions

Session 2:

Pan & Volume Automation

Thinking of problems with panning from the last session, I realised that the whole group was moving around almost all of the time, so for some songs they may have been more to the right and in others - to the left. I used automation on panning to correct this.

As for volume, I searched for individual peaks for each track and then adjusted the volume so that it would peak right up to 0dB on loudest moments. To maintain continuity, I decided to keep volume constant during speeches in between songs, as those parts were highly compressed anyway.

Volume automation on Output 1-2

Later on...

I bounced the whole length of the event onto one stereo WAV file. I plan to split it into separate tracks and burn it as an audio CD with no pauses between tracks, so that it will be easy to navigate, but it will still be a good memory of the session for those who took part of it.

For the assignment though, I plan to make a separate CD with songs only.

Session 3:

Bouncing tracks for Songs Only CD

To get the WAV files for a CD, I copied the project onto a new file, and then used automation on Speech track to get rid of speeches between songs. I did keep the comments the course leader made during songs - that's why I couldn't just mute or delete the speech track.

Volume automation for when the speech track is not needed

I also adjusted lengths of song regions, as they were a bit inaccurate due to long cross fades between speech and song tracks. Then, having automation on Output volume already made on previous session, I bounced the files.

This is how the cross-fades looked like for the full session version...



... And this is how I adjusted them to fit to the track beginning.

Bouncing tracks for Songs and Comments CD

For this CD, I made a new project in Logic and imported just the one file I produced in previous session. I pasted the long sound file into 1 track and then cut it into separate regions which would reflect fairly natural separation of songs and comments related to them. When the comments of the course leader were long or not quite about the song, I cut them into totally separate region.

Quiet reflections on the module

During the session, there was a moment when all the course participants shared their thoughts on the module. They all sat down and spoke one by one. As they were speaking way quieter than singing, I decided to apply more compression for that moment:

Really hard compression for final thoughts: Threshold -34dB (meaning practically everything is being compressed), ratio 7.4:1, attack 10ms, release 48ms.

Additionally, I used limiter so that the levels wouldn't go above 0:

... and EQ to cut down noise and put stress on the speech frequencies:

I realised that with this setting, the overall loudness of this track would be much higher than the rest of the CD. To make the change smoother, I used automation:

Although it looks rather steep, it takes a couple of seconds before the volume reaches 0.

Then, having everything ready, I just selected used regions from Audio Bin and used option 'Save elected regions as...' - and it did the job for me! I finished with 16 separate audio files.

When I was burning CDs, I just didn't put any pauses between the tracks for the Session CD, so it would present the whole class without unnatural spaces, but still remain searchable.

And here is the result of my work: 2 sets of WAV files just waiting to be placed on the CD!

Music Therapy - Logic session

As usual, I started work with a quick listen to the record.
This time I heard it, I got a feeling that the whole stereo image is moved to the left. I confirmed my feelings by checking the left / right levels in the multimeter:



To correct it, I panned the stereo track by 11 points to the right. When I was trying to find the right pan value for the track, I found out that when I move my head sideways, the stereo image changes a lot, like it was missing 'middle'. To change this unpleasant feeling, I decided to send the whole thing to the mono bus. This was also a possibility to deepen the spatial feeling of the record by adding a bit of artificial reverb on the stereo track and EQing the mono track so that it would sound 'closer'.
This set-up resulted in nicely balanced sound, with good stereo spread and warm, pleasant centre.

The processors I used on the stereo track:

EQ:

• Bass cut @ 86 Hz
  
• Dip by 3db @ 290 Hz
  
• Top shelf from 5.4kHz up by 2dB
  
• top cut at 17kHz.
  

Space Designer – Preset 'Live Chamber', 0.444s. Dry: Max, Rev: -12dB (barely audible, adds space but not too much).



Limiter: Set up so that it doesn't activate unless they clap. I used it just to bring volume up to acceptable level.



The sound from stereo track was sent to Bus1 which I set to be mono. The only processor on the Bus1 was EQ: Low Cut @ 38Hz, 24dB/oct, Top Cut @ 1360 Hz, 6dB/oct.



The amount of the mono bus in the mix was reduced by 4.3dB.



For the parts where the course leader was speaking between songs, I had to use completely different set of processors. Keeping the dynamics of the sound is not that important for those moments, as they mostly consist of speech. Therefore, I was free to use quite heavy compression: Threshold of -22dB, ratio 6.1:1, attack 2ms, release 34ms.



I also used EQ:

• Low cut at 134Hz
  
• Dip @ 540Hz by 9dB
  
• High cut @ 6800Hz, 6dB/oct.
  

Onto such prepared track, I pasted regions of speech cut out from the stereo track. I cross-faded them to make transitions smoother. This resulted in great improvement of audibility of speech, and let me keep the dynamic changes intact for when they were singing.


Close-up of the fades


Wide view of the cut track

Now, having everything cut and mixed, I only have to adjust levels of each song and find a smart way of splitting the project into separate tracks.

Music Therapy recording

Last Wednesday, I recorded the last meeting of one of the groups from Music Therapy course.

The recording was quite a challenge itself, as they all stood in a circle and were supposed to move around across it as well. Getting the levels right was also a problem as they sometimes sang altogether and sometimes it was just the leader saying something in between.

The room

The recording took place in the Recital Hall in Helmore building at ARU. The room has nice acoustics, not too dry and not too reverberant. There is a wooden floor, plastered walls with a lot of thick curtains on them and a balcony. Just by the looks of the room, I was expecting much more reverb, but actually it was just OK.

Microphone set-up

When I first heard 'circle', I immediately thought of putting some sort of stereo array right in the middle. Then they told me that they will be walking around and across, so I was forced to put my microphones outside the circle. As I realised later, it was a good choice, because this way I got nice, clear stereo image. If the mics were in the middle, the effect might get really weird and hard to mix in stereo.

I decided to put the mics as high as I could extend the stand, about 3 meters away from the edge of the circle. Putting them in some distance was crucial because otherwise the difference between people standing close and on the other side of the circle would be too big.

The NT5 were spread apart at wide angle (about 135 deg.)

Top view of the set-up


Another view of the set-up.

Equipment used

I first wanted to record using Rhode NT4 stereo microphone, but I was told in the studio that someone else had some problems with the left part of it, so I decided to take a pair of NT5s. I took the NT4 anyway, and after I set NT5s up and started recording, I set NT4 up as well.

As I was quite in a rush while setting up, I decided to use the recorded that I know - Marantz. I hooked it up to NT5s and recorded as stereo WAV files, 16 bit resolution, 48 kHz sampling rate.

Marantz control panel.

I was also offered a use of the brand new H4 digital recorder. I used it to capture sound from NT4, in 24bit resolution and 48 kHz sampling rate as well. The new recorder performed very well and I guess I will start using it now, especially that it has a stereo pair of microphones built-in.

First listen

I am not very happy with the stereo spread that I've got from NT5s. Although they were spread rather wide apart, I guess they just were too far away from the circle to get the full stereo image. NT4, although it gave me very nice signal in good quality thanks to H4, is not good enough to be used simply because I started recording it way later than on Marantz.

Anyway, there is a presence of spatial placement of different voices, and quite nice effect when they sing altogether. When listening on studio monitors, I had a feeling that the recording is lacking top frequencies. I will work on that in the following sessions. 

The levels were set right. There was obviously some clipping, but as I checked it only occurred when they were clapping, which for me is acceptable. The background noise (especially people walking and talking outside the Recital Hall), so well filtered out by our excellent human hearing, got onto the recording with rather high volume, but still way quieter than the primary content.

Video comparison

Here is the video that lets you have a rough idea on valve processing I've done.
Unfortunately, the audio quality through YouTube is terrible. I will keep looking for a better way to present audio files on this blog.

UPDATE:

Apparently, Vimeo offers much better image and sound quality. I think it should be OK to use it here:


Valve processors from Karol Pilch on Vimeo.

Fitz Choir - finall Logic session

Yesterday I finished working on the choral material from Fitzwilliam Chapel Choir.

In General

My work was just to cut the whole material into separate tracks, make them fade in and out nicely and set the levels so that the full range of bit space is used.

Gate

As I noticed in few pieces, there is a space between the song and the applause. I tried removing this silence and putting clapping right after the song is finished, but that sounds very unnatural. Instead I used a noise gate set to decrease levels by 16dB when nothing is going on. The noise from EQ and the record itself gets then decreased, but it is still possible to hear people coughing and making noises before they start clapping. This sounds much better.
To be sure that it doesn't cut anything during the songs, I used automation to keep it off except for the end of the song.


EQ and gate used in final bounce-down.


Automation on gate's reduction. For when I need it, it goes down to -16dB, otherwise it stays at 0 - no change.

EQ

Additionally, I used EQ once more (but this time I relied on Logic's own EQ) to reduce wide band around 750Hz by -6dB in the deepest place. After a while of rest after the previous session, I found those frequencies still a bit too loud.

Levels

I found that even during the loudest parts of songs, I can still rise the levels by 2.1dB. I couldn't keep this loudness though, as this would make applause heavily clipped. To have both songs and clapping with proper level, I used automation again - just to reduce loudness from 2.1dB to -3.1 during applause.


Automation on Output Volume during applause.


Wider view of automation on the output.


This is how the finished track looked like.

After all that, I only had to bounce the material down, track by track. I set the output sampling rate to 44100Hz as this is the rate that is going to be used when burning a CD.

Now I only have to burn a CD and it's all done!

Valve kit pics - Fitz Choir

Here we go:


Compression settings for channel A. There is a MASSSIVE input gain, because I was a bit scared to bring the levels up while recording. The threshold is set so that it enables the compression only on really loud sounds - so it pretty much works like a limitter (ratio of 1:30). Very fast attack - not to let any loud sounds through.



Compression channel B + meters. Settings are the exact mirror of channel A (somehow I couldn't get the Stereo button to work). Gain makeup a little tweaked to keep the balance between left and right even.


EQ: Gain, Low-cut, High-cut, Low freq band and Low-medium freq band. The shelf/band switch was at 'shelf' position (LF).



EQ: High medium freq band, High freq band and master gain. Note that only Channel A (bottom one, apparently) controls the EQ as it is the master when they are linked using Stereo button.



Screenshot of Logic session while bouncing down from valves. On the right: Logic EQs. top one cuts down bottom of the signal that went to valves, to reduce humming noise I wrote about earlier. It was much easier to filter it out using Logic EQ than valves. The bottom EQ shows the reduction I done to the bounced down signal to eliminate hiss made by valve EQ.



Screenshot comparing original and bounced down tracks - note rather dramatic inrease of levels!

Fitz Choir - post-production

OK, I finally opened the Fitz Choir recording in Logic, in ARU Mastering Studio. After a couple of minutes of figuring out how the signal goes around the studio (congratulations to Nick for clear and easy to use layout!), I started listening.

First listen

As I suspected just from listening how the concert sounded live, I discovered quite high frequency content in middle range, between 500Hz and 1.5kHz. I don't quite like records that have too much in there, so I put the recording through TLAudio valve kit.

Compression

I used the compressor just for increasing the levels to drive the EQ. I left the compression ratio at 1:30 and set the threshold so it would work as a limiter for the loudest sounds only. Attack was set to minimum, and release not much longer.

EQ

In the EQ unit, I did some basic adjustments - I took down 600Hz + surroundings, took up what was below 400Hz using shelf EQ, cut off what was below 250Hz using low-cut. Also, I decreased 1kHz with surroundings because I didn't like what was in there.
That all, combined with incredibly warm sound of valves, created beautiful sounding effect. EQ leaves quite a lot of hiss though, which I will have to remove in Logic later.
I am not quite happy with the stereo spread - but as I said before, I didn't want to set the microphones too wide apart in case the audience wasn't quiet enough.

At this very moment, the whole thing is bouncing down to new tracks. I now plan to tweak the result in Logic (possibly use the unprocessed recording with added reverb for applause), split the songs and the whole processing should be over!