Getting A Prayer To Owen Meany Working On Your System
I spent about three weeks trying to get the vocal track from A Prayer To Owen Meany separated cleanly from the instrumental back in 2023, and I learned enough through trial and error that I figured I should document the actual workflow before I forgot the specifics. This isn't some polished tutorial from a blog that sells courses, it's just what worked for me. The source material for A Prayer To Owen Meany exists in a few different stems depending on which platform you pull from. The original album release has the full mix baked at roughly 142 BPM in common time, but if you're doing anything with the individual vocals or the string arrangement, you need the multi-track files, not the stereo master. I tried working off the iTunes version first and kept running into phase issues around 2:14 where the piano and strings overlap in a way that makes any kind of EQ surgical work nearly impossible. The lossless WAV from the Bandcamp download is worth the eight dollars if you're serious about this track. I'm running Reaper with the free version of izotope Nectar for vocal isolation, and for the instrumental stems I use a combination of RX Advanced and my own custom EQ curves. The problem with A Prayer To Owen Meany specifically is that the production is fairly dense in the midrange, which is exactly where most vocal separation algorithms choke. I tried several consumer-grade stem splitters and they all produced garbage artifacts around the chorus entry at 3:02, which is when everything swells simultaneously.
The workaround I ended up using was surprisingly simple. Instead of relying purely on AI separation, I ran the track through RX's dialogue de-hum first at a very conservative setting, then applied a dynamic EQ carve specifically at 2.4 kHz and 4.1 kHz, frequencies that the vocal sits in heavily during the verses. Then I fed that into Nectar with the isolate vocal setting at about 60 percent, not higher, because anything above that introduces that watery chorusing effect that ruins the intimacy of the original performance.
A Specific Problem I Hit and How I Fixed It
About four hours in, I noticed that every time I pulled the vocal stem out, the cello part in the second verse disappeared entirely. It turns out the cello and the lower vocal register occupy almost identical frequency space in that section, which makes sense compositionally but makes it a nightmare for source separation. I solved this by doing a manual gain ride on the original track, boosting the cello frequencies at 180 Hz by about 3 dB in that specific passage, running the separation again, and then blending the two results together. It takes longer than a one-click solution, but the output is actually usable. I've since learned to do this as a two-pass process for any track with dense orchestration, but A Prayer To Owen Meany was where I learned that the hard way. This method isn't perfect and I want to be honest about that. If you're trying to do this on a laptop with less than 16 GB of RAM, the RX processing on a full-length track like this will eat your CPU and may cause buffer underruns. I had to render each stem to its own file rather than processing in real-time. The final vocal isolation from A Prayer To Owen Meany will never be as clean as the original multitrack session files, and anyone telling you otherwise is selling something. The best I ever got was about 85 percent clean, and even that required manual edits in the DAW to remove the occasional artifact between phrases. If you're only looking to download the finished track for personal listening, just stream it or buy the album. This whole workflow only matters if you're a producer, a remixer, or someone working on a project that requires isolated elements from this specific recording. The effort is probably four to six hours total for a first attempt, and maybe two hours once you've done the process twice.
Alternative Approaches
I should mention that Moises.ai and Lalal.ai both have options for this kind of work now, and they're faster, but they tend to over-compress the stems and lose some of the ambient room sound that gives A Prayer To Owen Meany its emotional impact. If you care about fidelity over speed, the manual RX plus Nectar pipeline I described is still better. If you just need something done quickly for a sync placement or a rough demo, the automated tools are acceptable and will save you the afternoon I wasted on this. The track runs approximately 4 minutes and 38 seconds on the standard release, and the full orchestral arrangement by David Wingo is worth listening to with headphones before you start any editing work. You'll hear layers in the final minute that you probably didn't notice on first listen, and those layers will determine how you set your EQ curves for whatever you're trying to pull out.
