Solo violin recording session

Solo Violin Recording - Bach Loure (BWV 1006:II)

#

Recording Session Information

#

The recording session was 1 hour long, including setup/down, with the recording running for 38.24. Noa David, the performer played Bach’s Violin Partita no. 3 in E major BWV 1006:II. Loure. The session was recorded and edited in REAPER.

Since it was a fairly simple setup and I had extra equipment and channels available I decided to setup two pairs of mics (which was overkill). I setup an XY (1) and an AB (2) pair at 1-1.5m away from the performer, but in the end I decided to not use the AB pair, as the XY pair provided the sound quality I required, and the AB pair didn’t add anything further to the mix. I had them positioned above the player, facing down towards the bridge of the violin. I also instructed the performer to avoid moving too much, as this can change the sound of a take in a narrow pickup pattern, making comping harder.

Microphones

#

The microphones used were setup as follows:

The Neuman KM184 is ideal for XY or ORTF setups because of its small footprint and the ease of aligning its capsule in a pair. They are however, quite pricey.

The AKG C414 XLS II is also very pricey, but somewhat more versatile because of its alterable pickup pattern, allowing it to be used in many different setups. The capsules are quite hard to align due to the shape of the protective mesh, and its footprint is larger than the Neuman.

The finished comp edits were in the following locations:

Here is a copy of the score, highlighted with the edit points for the comp:

Further editing information

#

I worked with two REAPER sessions, one for my source recording, which included all my microphone tracks, and all my mixing edits. I then imported this into a second, Edit session as a sub-project, and worked on my comping edits, and reverb in this session. This method allowed me to edit my source recording mix independently from my comping edits, leaving a single stereo track to work with in the Edit session.

The mix session

#

An EQ has been applied to the ORTF. I applied a High pass filter to remove the unnecessary low frequencies around 200HZ, and a Low pass to roll off the high end slightly around 20kHZ, to reduce the harshness of the source recording.

I decided the levels were good with the one pair of microphones, so didn’t change the faders, leaving them at default. With the A/B I found too much of Noa’s breath sound was picked up, alongside too much room sound, which was undesirable as I added atmosphere in post via an artificial reverb, rendering the room sound irrelevant.

The edit session

#

In my editing process I liberally used markers to decide where to put edit points and mark out issues that needed resolving. In my edit session I applied a unique colour to all the different takes, so that it was easy to tell where edits were and which takes I used.

This session is also where I did the majority of my post processing work, applying an artificial reverb to the source sound, as the recording studio had a naturally dry room sound, and my goal was to make the performance sound like it took place in a concert hall.

I used 4 main tracks, a REVERB track, which contained the artificial reverb, an EDIT track, which contained the final edit of the source recording, a TEMP track where I would temporarily drag my takes for editing, and then a TAKES folder, which contained all of my various takes for different parts of the piece, lined up with where they happen.

On the REVERB track I used an FX chain of two EQs, and an impulse response based artificial reverb.

During the recording process I inserted a marker at every new take, which made splitting the source recording into the various takes easier in the Edit session. I began with two full takes of the piece, which formed takes 1 and 2. I then split the piece up into various sections, and worked through them chronologically with Noa, aiming for at least two takes of each section (some problematic sections had considerably more takes). I then lined up all the takes under their respective sections in the full take, which made auditioning different takes much easier.

I named each take track according to the bar number it represented, and a tag (e.g. T1) indicating which take number it was. I also made sure, as mentioned earlier, that all my takes had a different colour to make identifying them easier.

I worked with a full take, splicing in alternative takes as necessary to correct mistakes in the performance. A big challenge I faced was the amount of breath noise that was picked up in the recording, which made a lot of normally good edit points such as rests difficult to splice into without cutting the breath short.

In my editing process I would extend the audio past the point at which I wanted to cut and overlap my takes. I would then adjust the crossfade region to happen exactly at the desired cut point, just before or after a note as sounded to mask the transition. The volume of some of takes was increased so as to be uniform, as Noa moved slightly over the course of the session, rendering some takes slightly louder than others.

I made sure to begin and end the track in silence, fading the beginning and end takes slightly as well, to avoid any pops/clicks from background noise. I rendered the file slightly longer than the audio region, to avoid cutting the reverb tail short.

Rendering

#

I rendered the final piece via the master mix, as a WAV at 48000 sample rate with 24 bit bit depth. I made sure to mute the TEMP track, and all of the take tracks, only rendering my EDIT and REVERB tracks. I set my bounds as Time selection, making sure to leave a tail at the end of my selection to allow the reverb to decay without being cut off.

Reply by Email