My ongoing relationship with the delightful Shedd family is continuing. After recording their recital, they approached me with this request. One of the daughters participates in pageants. Part of the pageant competition is a talent. The idea is that her talent would be to dance ballet to a song that she recorded, essentially being a double talent. Pretty cool idea, really.
The father did an arrangement of Amazing Grace for the family to play. The trick is that the piece couldn’t be more than 90 seconds. They practiced a lot and we scheduled a Saturday afternoon for the recording.
Since I don’t currently have a useful recording space (especially for a 9 piece ensemble) we did the recording on location at their church. I packed up the exact same kit I used for the recital recording and showed up two hours before the downbeat to set up. The church sanctuary where we were recording was rectangular with a peaked ceiling. There was moderate absorption mounted on the walls, but I wouldn’t want to record highly percussive material in there – there was a still a lot of flutter echo when I clapped my hands. For a string ensemble, though, it would do just fine.
The instrumentation was going to be three violins, three violas, two cellos and harp. My plan was to use the large diaphragm condenser (LDC) on the harp, the ribbon on the violins, the tube mic on the cellos and the small diaphragm condenser (SDC) on the violas. The instruments would be set up in a semi-circle in the large open area at the front of the sanctuary. I set up all the boom mic stands, created for myself a cozy little spot for my laptop and interface, plugged everything in and waited for the talent to arrive.
When they all came I began adjusting the mics and getting sounds. The funky one was the harp – I had always mic’ed a harp from the front, but the harpist mentioned she had seen it mic’ed from the back near the soundboard since that is where the body of the sound emanates. I moved the microphone around and decided that I did, indeed, like the sound from the back better. See mixing, below, for some of the consequences of that decision.
The violins and violas all played standing. I set the lead violin (they were playing in three part harmony) in the middle and put the ribbon mic about as high as I could get it, moving the other two players around until I had a nice balance. The violas were arranged similarly, although there was a significant skill difference between them. The older son was a very accomplished viola player – his two little sisters maybe not quite as much. I placed him in the center flanked by his sisters. The height difference also meant a big volume difference between the players, which worked to my advantage.
The cellos where a little bit of a challenge. I repositioned them several times before I found something that worked, with the microphone between and slightly below the music stands with the players angled toward each other in a shallow “V”. Monitoring for the performers was not yet necessary as everyone would be playing together until we got a good take, followed by overdubs as needed.
The fifth take was a keeper, with the dad giving it a listen for approval. The way the song arrangement works it builds up from a solo violin at the very beginning to the entire ensemble in the middle with a dramatic ending. In all it’s a very dynamic piece of music. From a recording perspective I now had five beginnings to choose from. Only the second and third parts where we were going to overdub individual instruments would be tied to the fifth take.
We went back and recorded the violas and cellos individually again, then called it a day. This individual recording would give me a little more flexibility in editing and mixing (see below for how important this was). For the violas I went ahead and had the brother play the part twice, each time on a different instrument. The mother and daughter each played their own cello parts. Total recording time was less than two hours (which for a well-rehearsed 90-second piece of music is not too bad at all).
The screenshot to the left shows the results of the recording. The top four tracks are the instrument sections, and the fifth take on the right has the individual instrument overdubs below.
When I got it home I edited together a single take. I ended up using the beginning from the third take and the middle and ending from the fifth take. Everything that was recorded with the ribbon and tube mics needed noise reduction. I’m loving the Izotope RX plugins. They work absolute magic on a track, reducing the noise without affecting the music at all. Lesson learned: Record a few seconds of silence to aid in noise reduction after the fact. I didn’t do this and had to sneak my noise reduction samples from the very short periods after I pressed record and before the music started playing. I also learn that it’s a processor hog, though. By the time I was done with the mix I was maxing out my CPU and had to be careful. If I adjusted any control on any plug-in while the music was playing back I would often get a dropout and the playback would glitch and halt.
I also took the individually recorded parts and tuned them as needed. There an old joke – how do you get two string players (or winds, for that matter) to play a unison in tune? Shoot one of them! Most of the work needed to be done on the cello-playing daughter. The mother also played cello, and her tuning was almost dead on. Fortunately, Sonar comes with a stripped down copy of Melodyne which does the trick nicely.
Once everything had been edited and cleaned, I turned to mixing. While I haven’t had much opportunity to mix classical music, when well recorded it is relatively easy, but also easy to overprocess. I bused the background string choir (violas and cellos) to a single bus so I could use that to easily control the volume of them together once I had them balanced. Everything was run through a console emulator and the tape emulator was added to the master bus. Also I ran the mix through the Sonar Breverb set to my favorite Warm Orchestra patch (foreground instruments with less reverb, background instruments with more).
Once I had the levels set I started with EQ, using a high-pass filter to cut the rumbly lows out of everything. I do this by listening to the track in solo and raising the cutoff until I can just hear a difference. Then I back it off until I can no longer hear it. In the end the tracks are less cluttered, and any applied compression won’t be triggered by excessive lows that don’t need to be there anyway. Also, the difference between the high-passed mix and the un-EQ’ed mix was astounding. It was so much cleaner and clearer with the EQs engaged.
At this point I started experimenting. I bounced the mix down to a stereo file and dropped it into a new project. As I continued experimenting I would take each bounced down track, import it into this new project, and use that to compare the changes to make sure I was making an improvement. I mostly experimented with compression, trying different options and amounts, learning as I went. Again it is very easy to overcompress classical music. In the end I had an LA-2A clone on all of the string channels and my favorite Density Mark III on the master bus set to mid-side processing.
The challenge was the harp recording. With the microphone placed where I put it the frequency response up and down the scale was a bit uneven. It was also having serious trouble cutting through the mix. I tried several things, starting with transient shaping to try and sharpen the attacks. This was the first time I used the TS-64 from Cakewalk. It works really great, but by the time I got it modified enough to cut though, it was beginning to sound unnatural. So I pulled the TS-64 and tried EQ. I boosted the highs to try and get it to cut through, cut a little of the high-mids to take away some of the harshness and cut the sub-bass to get rid of the rumble. It still wasn’t enough, so I added compression.
I turned to Bluetube’s Analog Compressor, which is included with Sonar. I found that by setting the input, ratio and attack just right I could get it to punch through the mix nicely without being overbearing. Now, however, the darn thing was too bright. I pulled back the high shelf on the EQ until it fit right. Ironically, I ended up pulling the high shelf into a high cut before it sounded right, but it did fit now.
One other trick that I did involved the reverb. While I love the Warm Orchestra preset on Breverb, I found that either it was too much or too little in this mix. I think the issue is that with as many instruments as this arrangement uses, the reverb was just adding mud to an already full mix. I tried rolling off the lows, increasing the predelay, and shortening the length but none of the results were satisfying. In the end I added a compressor to the reverb, but sidechained it to the lead violins. The result of this is the signal from the violins causes the reverb to compress, or duck, out of the way of the main instruments. This keeps the reverb from obscuring the main instruments but allows it to bloom in the quiet spots at the end of phrases or the end of the piece.
After a couple of rounds of bouncing the mix down, listening to it in my car and coming back for minor tweaks, I sent the mix off for review. A few days later I received a reply that they were unhappy with some of the performances and were wondering if I could do a few re-records. I said sure and we set up a date.
We set a weekday evening for the overdubs as the timeline to have the mix submitted to the pageant judges was looming. We met back at the same church with just the violin players. The father took a more active role producing this time (since he wasn’t playing along with the kids) and that helped tighten up the performances he was worried about.
I changed the recording up slightly this time since I was only recording solos. I brought my SDC and ribbon mics and arranged them mid/side. I really like doing this for the lead instrument in my recordings as I can get it nice and wide without losing definition and presence. I left both mics up for all three performers but only turned on the side mic for the lead.
We did the recording in two separate sections. The beginning of the piece is just the three violins so we recorded that on it’s own, doing several takes of each performance. The second and third violin played along with the best take of the lead to make sure everything was in time. Those are the clips on the left in the screenshot above.
Speaking of being in time, the middle part of the piece has the entire ensemble playing, so I brought a two-track mix of that part and had the violins play along with it, again one at a time. Those are the clips in the screenshot more to the right of the timeline.
Unfortunately at this point we had run out of time so we left the ending the piece as-is. The ending was the best part of it, however, so no harm done.
I brought the new tracks into the old project and edited everything together. Now that the lead violin was by itself on two tracks (one for mid, one for side), I was able to to a little more with it. In the screenshot you can see that I sent the two mono channels to a stereo bus panned hard left and right.
I didn’t have room on the screenshot for the two processors on that bus, but MSControl is what converts the two mono signals into one stereo signal (and the CA-2A is the compressor again). The EQ is the next most important after M/S conversion. Since the violin plays fairly high I rolled off all the way up to 500 hz before I started hearing a difference. I little dip in the upper 1700’s cured any scratchiness and a little lift with a high shelf helped the instrument to sing.
In the end they decided that they liked the middle part, but prefered the original beginning. This is because the overdub was played a bit faster, and upon listening they thought maybe it sounded a little frantic. When all was said and done the mix looked like the composite screenshot to the left. The overdubs are at the top of the screen, with the original recording in the middle (with each Section track above the solo overdubs). At the bottom are the buses.
You can see that the string bus has two automation lanes – one for volume and one controlling the lo-mid channel on the EQ. I had a problem with a single note near the end of the piece ringing out quite loudly. I tried several things to fix it, including a regular EQ cut, but that left that note out when it didn’t ring earlier in the piece. The answer was to create the automation cutting that note only when it was too loud. The gentle rise in volume at the end of the piece helped the low note being played by the string choir sound a little more strongly.
When I was satisfied I bounced the mix down to a stereo master and imported that into a brand new project.
Adjust to Time and Call It A Day
As I mentioned at the beginning of this article, there is a strict 90 second time limit for this piece. When I had put everything together it clocked in at about 92 seconds. (With the overdubbed beginning it was 89 seconds, which gives you an idea how much faster that third of the song had been recorded). The reason I created the new project was that I needed a clean file to perform the time stretching algorithm on.
It is really amazing was can be done with an audio file these days. Once upon a time it would have been necessary to run the tape at a slightly higher speed to adjust the length, which would also adjust the pitch. Sonar includes a Fit To Time utility that I used to shave 2 seconds off the length.
In the end the family is thrilled with the results and I have learned a lot about recording a classical ensemble. In the future I will try and get a little better separation between the instruments when recording them all together, which will increase my flexibility later when editing and mixing. Also, the violas bled all over the cello track and the violins were well into the harp channel.
Also, two mics on the harp, I think – one in front and one behind. That will save me the work in mixing to make sure everything can be heard. In the end I really hope I get to work with this family again. The harp player leaves for college in the fall, however, so we’ll have to see.
For now, please enjoy Amazing Grace, by the Shedd Family Players.