On January 1, 2017, Chicago singer/songwriter Sarah Eide released her self-titled album. To help better educate audio students, I've decided to document the entire recording process of the album in three separate posts - including a Q&A from students:
I had recorded Sarah once before for a class when she was the pianist/keyboardist and a backup vocalist for a group called Wooden Rings. We shared a few words, found out we had gone to the same music college, and then we got back to work. So it was a surprise when she texted me out of the blue to record a piano/vocal album. Since I love recording acoustic instruments and working with singers, I enthusiastically agreed, and we quickly started planning.
My only advice so far is that pre-production is highly critical. For all the talk about 'vibe' and 'being one' with the artist - I still think that the best way to make sure that you and the artist are on the same page is to simply have a detailed conversation ahead of time and set reasonable expectations accordingly (Incidentally, this also helps save a lot of time and trouble in relationships too, but that's a different blog post). Through our several conversations, we established the following:
- There will be 8 songs
- The goal of the album is to have a 'live' feel. As if Sarah is performing these songs for an audience
- To point #2, Sarah wanted to record the piano and vocals at the same time. We wanted to avoid the record from sounding 'overproduced'
- With regards to style, Sarah associated herself with Tori Amos and (early) Joni Mitchell and playfully stated that she is "Tori on a budget" (nice!)
This setup has the following advantages:
- Sarah saves time and money by performing everything at once with no overdubs
- If we are going for a 'live feel', there really is no better option than just doing it all live
- It saves me time because I would be doing minimal editing.
And the following disadvantages:
- I've never recorded Sarah before. I didn't really know her yet. If she's not prepared, or just not very good - I'm severely limited in my ability to edit the performance.
- I've recorded piano before and I work with singers and voiceover artists all the time, but never at the same time in one room. I now have to figure out how to minimize sound bleed from the piano into the vocal microphone and from the vocal to the piano microphones.
My assistant engineer, Victoria, and I went to work.
The first thing we did was took a day to test out a few things in the room. There were a couple of reasons for this, the biggest being is this:
Getting a great recorded piano sound is difficult.
I cannot emphasize this enough. Audio engineers can sometimes be sensitive, insecure creatures. For reasons that I never understood, some audio folks do not like to admit that things are challenging. I'm not one of those people - so I repeat:
A piano is an extremely complex instrument and getting a great recorded piano sound is difficult.
A grand piano contains thousands of parts. Depending on the style of music - there are dozens of different ways to put microphones on it. A piano has a huge frequency range and a large dynamic range. Every piano is different and every pianist is different. The piano interacts with the room and the parts of a piano also interact with each other and the larger instrument.
That being said, Victoria, the students and I needed to figure out a plan to record this highly complex instrument that best fits the style of Sarah's music.
Luckily, a little bit of research goes a long way.
My audio engineer hero/idol is Al Schmitt. It turns out one of his regular artists, Diana Krall, also prefers to play piano and sing at the same time. Imitating what Al does is usually a pretty good recording strategy. So we did!
After following Al's setup and a bit of trial and error, we came up with this aesthetically pleasing number:
Behind the singer's microphone, there is a heavy blanket draped over the lid of the piano. Behind the blanket, there are extra layers of acoustic foam.
Additionally, there are large panels set up around the piano to minimize the reflections from the room. We decided early on that the room is not our friend for this recording (mainly because it's not a very good room) and if we really wanted this album to feel like a live performance - it would be more beneficial for us to artificially control a high-quality reverb unit rather than try to capture the sound of a mediocre sounding space.
Now the big questions of the day: What microphones do you use, and where do they go?
We had to achieve the following:
- Because the piano is the only instrument besides the vocalist, it had to cover a wide stereo space
- We needed to close mic it because we wanted to control the reverb ourselves
- If we use cardioid mics, depending on how we place them, we might have to pair them with an omni set to pick up the full frequency range of the piano (without going into too much detail - omni microphones pick up more bass than cardioids and will also, for better or worse, pick up everything else including reflections)
- We also had to pick a good microphone for the vocalist that would also pick up minimal bleed from the piano.
After a lot of thinking and a bit of trial and error. We decided on the following input list:
Piano Close: Gefell M930 stereo pair (condenser, large diaphragm, cardioid)
Piano Further Away: Sennheiser MKH8020 stereo pair (condenser, small diaphragm, omni)
Additional piano mic for the rear: Royer R-121 (ribbon)
Vocals: Sennheiser MD441 (dynamic)
Our initial idea was to use the omni mics as the "main pair" similar to how it would work for classical music. It makes sense as these mics cover a wider range and would make the piano sound more "real". Then we would fill in the attack on the piano using the close mics and fill in any missing mid-range with the Royer mic. (See Part 2 - mixing as to why this did not work as planned)
We chose the 441 for the vocal for a couple of reasons. We decided to use a dynamic mic to minimize bleed from the piano (although the 441 does pick up sound from the rear) but mainly because it sounds so damn good on female vocalists!
Our recording set up:
The good news and most important thing: Sarah was great!
We were able to get all 8 songs recorded in one day. Most songs only took one or two takes. Each recording truly was a performance - which made the most important aspect of the process - the "performance feel" be as real as it can get.
Below there are some more detailed answers that our students like to know. If there is something I missed, please feel free to contact me and ask me!
Q: Why did you put the microphones in those specific locations?
A: A lot of it was educated trial and error. We walked around the piano while someone played and tried to find the most interesting places to pick up. Once the microphones were set we adjusted them to spots that we found most aesthetically pleasing.
Given the close proximity of those Gefells, we added the ribbon mic to the back of the piano to pick up any mid-range stuff we may have lost.
Q: What preamplifiers did you use?
A: The close piano mics went to Wunder preamps. The Sennheiser omnis went to Great River preamps and the ribbon mic and the vocal mic went to Pacifica preamps. I don't know the model numbers. To me the mic selection, mic placement and the performance is WAY more important than the specific preamp (assuming that it is a high quality preamp).
That being said, the reason I picked the Wunders is because the piano we had was "dark" sounding and the Wunder preamps have a very nice, pleasing high shelf EQ to help brighten it up a bit. The reason I picked the Pacificas is because I like them. The reason we picked the Great Rivers is because Victoria likes them. There was no method behind it.
Q: Did you use any EQs in the recording process?
A: None, other than the high shelf on the piano. I follow the Al Schmitt philosophy - and we let the microphone location be our EQ. If something didn't sound the way we wanted it, we'd move the mic.
Unfortunately the "dark" sound of the piano is an issue with the instrument itself - so it was necessary to cheat a little with the high shelf. The piano actually had several deficiencies (more on that in Part 2: Mixing). I wish we had Tori Amos's Bosendorfer, but we didn't :-(.
Q: You didn't mention compressors at all. I hear that engineers like to compress the vocal going in. Did you?
A: We did not use any compression in the recording process. Here's the rationale:
A compressor, by definition, is a very simple device that turns things down automatically when they get too loud. What's too loud? Up to you! The fun of a compressor is in the intricacies as to how it does it (you could easily have an entire class dedicated to this). By definition, a compressor reduces the dynamic range (the range from the quietest sound to loudest sound) of a performance by turning down the loud stuff and keeping it closer to the quiet stuff.
For dynamic performances (meaning that things go from very quiet to very loud), it's logical that engineers will put a compressor on the vocal as a precautionary measure in case things get too loud for the system to handle. Similarly, a lot of engineers will put a compressor on the singer not just to avoid problems but because they actually prefer the sound of it. Cool!
In the case of this recording, my goal was to keep as much of the dynamics of the performance intact as possible. It means we had to record the vocal a little quieter than normal to make sure that when Sarah gets really loud that she doesn't distort. In fact, we had to do a second take of a song precisely because she got loud, distorted, and there wasn't a compressor to protect us. We apologized to Sarah and turned the preamps down. The tradeoff is what we call a "higher noise floor" - meaning that the level of the signal (the sound we want) is closer to the noise (HVAC, other sounds we don't want).
The other reason (more detail in Part 2) is that compressors are reactive, meaning that they turn things down after they happen. Granted, they can do this extremely quickly (and some can "look ahead") - the result is that using a compressor alters the transients (very quick initial bursts of sound) going in. Now that could be a good thing (i.e. "taming the transients of a snare drum") but my ears are extremely sensitive to changes in the transients of a piano or vocal (much to the annoyance of my colleague Tim Rusin) - so I didn't want to risk losing them in the recording process.
TL, DR: By not using compressors, I sacrificed a little extra noise to keep the real dynamic range of the performance intact. I also chose to keep the transients as they were happening, for better or worse.
Q: How much editing to the performance did you have to do afterwards?
A: Go Sarah! There were only two small performance edits on the entire album, and both were on the piano while she was not singing. There were no edits to any of the singing, we just kept the best take. What you're hearing sung on each song is a real performance, start to finish.
I was pretty much unable to edit the vocals anyway since there was a decent amount of piano that was picked up by the vocal mic. Grabbing vocals from other takes would also grab the bled-in piano performance from the other take and it would sound awful. There was also no tuning and no click track.
Q: Were there any mistakes that you made? Anything that you would do differently?
A: Audio engineers don't make mistakes, we make creative decisions!
For real though, we could have spent more time on the omni microphone pair (more on that in Part 2). Hint: piano lids are reflective!
Also there was a technical issue with one of the preamps that we didn't notice until we sat down to mix. Luckily we were able to fix it.
On to Part 2: Mixing!