you're reading...
gear/software, Mixing

Mastering Beach House

Man, has it been busy.

After 2 years, Beach House returned to WFUV for a Studio A visit with Rita Houston. We weren’t able to videotape the session, which was just as well because the absence of cameras allowed all of us to focus solely on capturing an impressive musical performance. I hope to write a separate post about the techniques employed during tracking, so for this post I’ll gloss over it.

The live 2-track mixes were captured to 1/4″ tape and mastered in Adobe Audition at 88.2kHz. Here’s the final result of “Myth”:

Here’s the raw mixdown:

The First Pass

As is evolving into a somewhat standard process for beginning, the first steps of the first pass involved some cleanup. As an example, I heard strange resonance at 1:48.7 (1:22 in the clip above), which I hunted using standard EQ-hunting technique (eg. high-Q boost and sweep). Switching to the spectral display allowed for precise extraction.

Resonance circa 280Hz is highlighted

Resonance, looped:

Resonance removed

I searched for harmonics, but wasn’t able to isolate them reliably. After that, I set about addressing the midrange, solidifying imaging, tightening up the low end, and a slew of other things.

There isn’t room to show every plugin, but it shows many of them (click to enlarge). Pic taken at the loudest point in the song.

A few decisions:

  • Used Variety of Sound’s BaxterEQ in mid-side mode to gently brighten up the side content while attenuating some LF mud.
  • I took a 2-stage approach to limiting, using the TC limiter shown above, followed by Sir Elliot’s at the end of the chain. The attack is a bit slow at 4.3ms, as I wanted Sir Elliot’s Mastering Limiter to do a lot of the work on transients.
  • I used band 3 of the multiband compressor to try to tame harsh midrange content in the vocal which began to evidence itself. Low threshold and a very low 1.1:1 ratio to preserve transparency.
  • After auditioning a preliminary pass on a few repro systems I decided to tame the kick’s sub frequencies a bit. Despite the NS-10 subkick having a free air resonance around 44 Hz, I found the offensive fundamental to be 55 Hz. The EQ on the right shows attenuation of fundamental and harmonics.
Here’s the result of the first pass:

Second Pass

While at first I was satisfied with the first pass, certain aspects began to bother me over the following few days:

  • The mix sounded like it was pumping with the kick and snare ever so slightly.
  • The vocal midrange still seemed uncontrolled and fatiguing, especially when she was really singing out.
  • Sibilance which didn’t bother me before, probably due to ear fatigue, seemed very harsh on certain systems.
  • The song sounded good to my ears, but was still missing some dynamic movement.

Focusing on the midrange once again, I dusted off Cakewalk’s linear phase multiband compressor.

Extracted waveforms of the area of interest; LP-64 bypassed on the left, in-circuit on the right

Don’t listen too loudly – this is basically all 2kHz.

While you can see some amplitude differences between the 2 waveforms, note that the LP-64 has makeup gain applied. As I hear it, the worst offender comes with the lyric “to GREET you.” While the issue is very apparent there, the example is indicative of the issue at large. The second half of the clip is the processed version, which sounds a bit more under control to my ears.

The track was already at a point of delicate balance, and my typical attempts with a DeEsser sound like I was beating the track with a hammer. Inspired by 2-stage limiting, I decided to try a 2-stage approach to de-essing, which resulted in a more transparent correction.

First stage on the left, 2nd stage on the right. Note the differences in the frequency display around the area of interest

Volume Envelope
Finally, I felt the post-bridge break down and final verse and chorus could be enhanced by creating more dynamic contrast between the two sections. While the band’s arrangement was growing more sparse during the breakdown, creating a natural decrescendo, the limiters were undoing a lot of that. So I put it back in with a volume envelope.

This depicts a written decrescendo leading into the breakdown, and a rapid crescendo leading back into the final chorus

Here’s a loudness graph for the whole song, which kind of surprised me but seemed to look as it should at the same time. You can probably guess what sections the red markers indicate.

That’s it, post done, not writing anymore.


No comments yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Follow @DanHodd for updates, or subscribe with your RSS reader.
%d bloggers like this: