Working Through Gaye Whats Going On Analysis
Most people come at this from the wrong angle. You open a DAW, throw a spectral analyzer on the bus, and start chasing problems. That is not how you do it. The process starts before you touch a single plugin. You listen to the track repeatedly, not as a consumer, but as someone trying to figure out why certain frequencies are fighting and where the arranger actually wanted the focus to sit. For Marvin Gaye's "What's Going On," that listening exercise alone usually takes two or three days of full attention across different playback systems. I know because I burned through a week on a college project that ended up being basically useless after I finally admitted I had been misinterpreting the bass line from the start. The foundational step is getting a clean reference track, which sounds simple but immediately becomes complicated. The original 1971 recording was pressed at a slower vinyl speed and mastered differently than the CDs that came out in the '90s. If you are pulling from a streaming source, you are almost certainly getting the 2001 remaster or the later SACD transfer, neither of which matches the original tape bounce the engineers were working from. I had a client who sent me stems from what they thought was the original mix and we spent six hours trying to reverse-engineer the low end before I realized they were using a digital file sourced from a vinyl rip through a cheap USB turntable. We switched to the Motown official box set transfer and the entire bottom spectrum changed character. It matters more than most people expect. Once you have the right source, the next layer is stem isolation. The track has multiple vocal overdubs, a saxophone line, percussion swells, and a bass line that operates at roughly 60 to 80 Hz with harmonics stretching well into the midrange. You cannot simply slap a EQ notch on the master and call it done. I ran into a situation last year where a producer was trying to clean up the vocal bus on a tribute mix and kept getting phased artifacts every time he pulled 3 kHz. The problem was not the vocals themselves, it was the string arrangement panned hard left bleeding into the same frequency band. Splitting the stem analysis by frequency range instead of by instrument gave me a much cleaner picture.
The midrange is where this track gets deceptive. The congas, the tambourine, the vocal harmonies, and the guitar all occupy roughly 800 Hz to 2.5 kHz. A beginner will reach for a wide cut around 1.2 kHz to reduce muddiness. That destroys the vocal presence. The fix is narrower. A Q of about 2.5 at 1.1 kHz with a 2 dB reduction preserves the body while clearing space. I learned this the hard way during a live session when a engineer friend of mine tried to fix a muddy master bus by sweeping broadly and ended up making the entire mix sound hollow. He called me in five minutes later. We reloaded the project, narrowed the bandwidth, and it took about twenty seconds to fix. Spectral analysis tools like iZotope RX, Voxengo SPAN, or even the free SPAN build are useful here, but they are only as good as the reference you feed them. Running these against a commercially pressed version and an acetate transfer will give you two entirely different pictures. I always recommend comparing at least three versions before drawing any conclusions about what needs correction. The original tape bounce shows more high-frequency content above 10 kHz than most modern sources. If your analysis makes the track sound too harsh by current standards, that is likely because you are comparing against a compressed digital master rather than the actual source material. Another practical issue that nobody talks about enough is phase alignment between the stereo components. The original mix used panning tricks that put certain elements hard left or hard right while keeping others centered. When you collapse this to mono for club play or broadcast, some of those elements cancel out completely. I encountered this when a radio station asked me to prepare a clean edit for airplay and the saxophone solo disappeared entirely around the 2:15 mark. The fix was not a mix adjustment, it was identifying the out-of-phase material and flipping the polarity on one channel. Ten seconds of work that saved the whole pass.
If you are doing this analysis for restoration purposes, the order of operations matters. Start with noise removal on the low end, then address DC offset if present, then move to spectral editing, and only then touch EQ or compression. Doing it in reverse will just push problems around without solving them. I have seen engineers spend hours de-noising a track that had a phase issue at its core, which made the de-noiser amplify the cancellation artifacts instead of removing them. The whole process went from a four-hour job to about forty minutes once we fixed the phase first. The final thing to keep in mind is that no amount of analysis will make a bad source sound great. If the original mix has a muddied low end because of the limitations of the studio equipment at the time, that is part of the recording. You can clarify it, but you cannot rewrite it. I tend to tell people working on this kind of project to define their end goal before starting. Are you preserving, restoring, or remixing? The answer changes everything about how deep you go and which tools you trust.
Get the Full Details
