Hold Me

This is my cover of Hold Me, by Logan Smith. Featuring me on vocals and alto sax.

Notes about making this:

Hold Me is a song I've been listening to for a while now; being in a vaguely angsty period of my life lately, I had been wanting to cover some angsty music, and this fits the bill (along with a couple other candidates that I might get to eventually).

But it took me a while to get to the point where I actually started recording, because I wasn't sure where to go with it. I usually don't do straight covers because I feel like they wouldn't add anything new except my participation, and that's often not an improvement on the original. Then I listened to Olivia Rodrigo's new album Sour recently (unironically pop album of the year, by the way), and I felt something about the second track, "traitor" (Spotify, Youtube) resonate with Hold Me. I had an idea and recorded a quick demo and liked the vibes, and did the rest over the next week or so.

If you have a metronome or ears, you might notice that the time is all over the place. I tried recording with a click track, but didn't like the way it felt - I don't think I'm comfortable enough singing to really be expressive while staying in time. So I decided to sacrifice tempo in order to focus on the vocals, and recorded the entire part a cappella.

Then, I overdubbed all of the saxophone parts. I often write out every single note I play, but in both the demo and the final I took a page from my friend Andy's book and just ad-libbed all the parts from vague ideas of chords; this went better than I originally expected, and I might continue with it.

The recording process itself took a good amount of time, because I kept picking up stray noise from the street, like really noisy motorcycles and helicopters and whatever. I ended up recording phrase-by-phrase, which helped me get cleaner takes, and also helped line up entrances better. I also recorded 45 minutes of room tone to find a continuous segment that was as long as the entire recording; this really helped with covering the edges between each snippet, as well as getting some nice thematic (but not disruptive) car-passing noises.

I also fixed a configuration issue I had in Audacity, which made it annoying to overdub parts. Previously when I tried to overdub recordings (by hitting record new track while an existing track plays), sometimes the recording would show a couple samples (literally less than a millisecond) and then suddenly stop. My old workaround was to take the click track or recording and play it through some other device, and just sync and combine the recordings later. This was really annoying for this because it's pretty tempo-loose voice, and so I ended up running Audacity from the command line. I quickly spotted the error ALSA lib pcm.c:8526:(snd_pcm_recover) underrun occurred whenever this early stop behavior happened, but wasn't sure how to fix it. Eventually flailing around, I hit on increasing the recording buffer size (and corresponding latency compensation). Overdub recording worked every single time after that. Honestly, I'm not sure why doing this should help for an underrun, but in practice it fixed my issue entirely.

Related to Audacity - I was talking to my friend Michael L. recently about making music, and he asked me why I use Audacity and not something like Ableton or Reaper. I've tried out more "serious" digital audio workstations (DAWs) before, and was always overwhelmed by the sheer quantity of knobs to fiddle. I definitely think of music in the acoustic terms of various instruments I play, and so the things that I come up with tend to be things I could perform live (of course, with several clones of myself). So, Audacity is more than enough for the task of taking various tracks, stitching them and overlaying them, and balancing their sound. But I wonder if this is some kind of catch-22 that's limiting me: I record in Audacity because the recordings I put together are straightforward, and the ideas I have are simple and easy-to-record because on some level I know I'll need to use Audacity to record them. And so I stay in this very analog mode and never learn the tools that I would need to make more electronic stuff. It's not like I'm ever going to be performing this live, so why not just use an electronic pad instead of painstakingly recording all these sax parts? I don't know. Maybe next time.

I think there's a lot of room to improve here. For example - my vocal range and the saxophones were too similar in both range and timbre. This ends up feeling muddy, and it was hard to balance them in a way that the vocals were clear but the saxophones were not entirely quiet background parts. Maybe recording the sax parts in stereo (instead of mono) could bring more depth without being louder. There's also just a lack of bass, so it might be even better to just bite the bullet and octavize some parts into the lower range (or add piano, accordion, or synth in that range).

But I won't fix this, and a bunch of other things too. I did the demo in a day, and then the recordings completely alone over two days the next week, and then mixing over two more days soon after. In hindsight, I think doing the recordings alone was a mistake - I got a lot of good feedback after the mixing stage, but I really didn't want to go back and re-record the vocals, and then the backing sax parts after that, and then maybe mess with the later sections of the song to make them fit better, and so on. I'll keep the ideas in mind for future work, and make sure I have someone else in the room for a tighter feedback loop during the recording stage. But for now, this one is done.

Thanks to Yee Aun, Andy, and Joanne for feedback on this cover. Thanks to Michael L. for ideas on recording music in general.