Why adjusting the song’s tonality to the kick drum sample may be a good idea
--
In music production, one known practice is to tune kick drums to the song key. In a 2007 interview, music producer Scott Storch recommends doing the opposite: transposing the song to make the tonality fit a TR-808 drum sample. According to Storch, the practice makes the 808 samples more audible. This post demonstrates that Storch’s recommendation can indeed make the 808 sample gain ca. 10dB in the mix once played on loudspeakers.
Tuning the TR bass drum to the song’s tonality
The Roland TR-808 drum machine, manufactured between 1980 and 1983, is one of the hallmarks of recent music, including rap/hip hop, dance, electronic music, and derivative genres. Its small but distinctive set of percussive sounds — particularly its kick (or bass) drum — are still widely used today. The TR-808 kick drum, with its characteristic pitched long-decay ‘’boom’’ sound, has become a synonym for a cross-breed of a kick and a bass sound.
The waveform confirms that the sample is tonal. The tonal aspect derives from the TR-808 generation technique, during which an oscillator produces a sawtooth wave that’s filtered to make it close to a sine wave.
The STFT also confirms the sample is tonal. Harmonics are present near the start of the sample and then fade out. The sample’s pitch is briefly higher near the start, then decreases to a stable value.
The tonal aspect of the 808 kick drum sample justifies tuning it to the song’s tonality. The practice also applies to kick drum samples other than the 808’s. In Scott Storch’s words:
“I mean, sometimes we think we’re hearing it right, and then sometimes you gotta go and fine-tune it […] I know there’s a lot of producers [who] will, I guess put an 808 in the song, and there will be chords and stuff clashing with it, and […] if […] your ears are really in tune with that stuff, you realize it’s just like [“not so convincing”kind of gesture]… Sometimes it actually does something cool to the track, but… […] I like to […] get into that, and tune the kick to match […] the bass line or whatever the chords are doing, you know sometimes I leave it, and then, change it for the hook or whatever […], I just try different stuff… then…[even when there is] not an incredible amount of tune, that carries over regular kicks , like short kicks, and I find myself sometimes at least even trying to tune, like a regular you know kick drum sound, and get it close to where most of the chords are in the song… it’s almost impossible to do but, you know…I try”.
The horizontal solid blue line follows the TR’s fundamental. The horizontal dotted blue lines follow the TR’s harmonics. The vertical blue lines denote the kick drum’s onsets. The TR sample that plays the role of both bass and kick drum is tuned to the song’s tonality.
Tuning the song’s tonality to the TR bass drum
Later in the same interview, Storch suggests that instead of tuning the TR bass drum sample to the tonality, one should do the opposite and adjust the tonality to the bass drum sample:
“Another thing that’s really interesting, is, sometimes, producers will program a song in a certain key, and they’ll try to program a 808 under it, and it’s like the key of the song is almost too low to really let speakers do what they need to do with the bass so, I recommend […] modulating the song up,transposing it up a couple of keys, and you’ll be surprised how much more level you can get out of the song. ‘Cos, you know, anything really below […] a low E, when they go down further, it’s like the speakers are gonna not, let you turn it up, you don’t feel the bass response.”
Scott Storch’s assertion can be interpreted in the two following ways: (1) if one transposes up a TR-808 bass drum sample, its acoustic level once played by the loudspeakers will increase, and (2) conversely, if one transposes down a TR-808 bass drum sample, its acoustic level once played by the loudspeakers will decrease. Let’s put Storch’s assertion to the test. Let’s evaluate the median fundamental frequency for the TR-808 bass drum samples, gather the frequency responses for several loudspeakers, and see where the TR-808 bass drum samples’ frequency lies in the loudspeaker responses.
In the diagram below, the boxplot represents the response of 36 loudspeakers, with percentiles 5, 25, 50, 75 and 95. The vertical grey line at the right of the grey rectangle shows the median fundamental frequency of the TR-808 “long” kick drum samples (the ones for which the fundamental frequency is most important) as provided by the Trisample library (49.5Hz). The grey area represents the frequency of the TR-808 median frequency when tuned down to one fifth (seven semi-tones).
Scott Storch’s position is coherent with these measures. In accordance with assertion (1) above, transposing down TR-808 kick drum samples can make them lose ca. 10dB in the mix once played on loudspeakers. In accordance with assertion (2) above, transposing up TR-808 kick drum samples can make them gain ca. 5dB in the mix once played on loudspeakers. In music genres like rap / R&B, a genre familiar to Storch in which a lot of 808 kick samples are used, a strong kick is obligatory. Therefore it may indeed be a better idea to tune the song to the 808 rather than tuning the 808 to the song.
Conclusion
At first glance, Scott Storch’s suggestion of transposing a song to make it fit with the kick drum appears to be counter-intuitive. A kick drum may not be considered tonal, and if it is, then the song’s tonality may appear to be more important than the kick drum’s fundamental frequency. The analysis above shows that if the kick drum’s level is important for the music genre and if the kick drum features a tonal component (which is the case in particular for the widespread Roland TR drum machines such as the 808 and 909), then transposing the song to make its tonality fit with the kick drum may indeed be a better idea than transposing the kick drum to make it fit with the song’s tonality.
In the context of A.I. music creation, the above principles (tuning of the kick drum, tuning of the song to the kick drum, loudspeaker responses) are useful to know in the cases where a model generates a kick drum from context, or conversely if a model generates tonal content from the drum parts.
Our team’s page: Sony CSL Music — expanding creativity with A.I.
