Digression: synthesis tips and tricks

Synthesizers and effects are nice things to have a round, but by themselves, they are not enough. Most music machines are difficult to use properly and often we have no direct way of knowing how to achieve some particular sound. In these situations, experienced sound engineers rely on their know‐how and intuition while the casual music maker is left out in the cold. This little digression is intended to function as a bag of tricks on how to achieve some of the most wanted effects. It will be a repository type of thing and grow as needed.


One of the hottest trends in the 90s has been retro/lo‐fi sound. These terms refer to the scruffy sonic outline which emanated from the early recording studios of the rock industry. The task is to duplicate some or all of the imperfections of that day’s audio equipment and distribution media. One may ask if there is point in all this, after all we have gone digital precisely to get rid of some of the imperfections which plagued the analog studio. The answer is the same as with fuzz boxes and tube amplifiers: they are generally liked, some people swear by their name and a great deal of expressive musical tradition depend on their presence. Similarly, lo‐fi sound has come to be associated with certain emancipatory aspects of music, like the inherent back to the roots, everybody‐can‐do‐it attitude of punk.

Lo‐fi as a term is the rigorous opposite of hi‐fi. It means low fidelity. We can see there are three ways to achieve that goal: we can run the music through precisely the kind of old equipment hi‐fi rose to supplant, we can model the effects of such a treatment by means of DSP or we can just shred the sound into pieces and hope nobody knows the difference. Old equipment is the way to go, if you have it. It is also the way we will not touch here. Modelling on the other hand is good if we want to retain control over the sound, but is generally quite CPU intensive if done right. The last way is easiest and the most common, but has to be used in moderation since the results are not very consistent. To get a clear picture of what we’re trying to accomplish, we divide analog distortion into four classes: skewed frequency and phase response (filtering), equipment instability, nonlinearity and noise.

Frequency domain artifacts are usually the most benign of these—they result from the limited frequency response of old equipment, such as microphones, tape recorders, cheap (guitar) amplifiers and imperfections in the studio environment. (Many best selling vinyls of the past have been recorded in places like storage rooms, staircases and so on. Some of the reverbs on early Elvis tracks have been captured in an empty basement with conrete walls.) If anything like equalization has been used, the hardware of the time has usually not been designed with perfectly linear phase shift in mind. The first key to lo‐fi is limited, nonlinear frequency response. try cutting everything outside a 100‐3000Hz region by a gentle sloping filter.

By equipment unstability we refer to the fact that analog devices have a tendency towards thermal drifting (especially synthesizers, which need constant tuning), sync problems (e.g. early open reel tape did not keep time very well), low tolerance to miscalibration (for example, analog noise reduction reacts poorly to misaligned tape heads) and highly unlikely interactions between different devices (like feedback and power supply coupling). The most obvious of these problems are the ones associated with tape, such as wow (slow variation of tape speed which makes pitches crawl), flutter (a more rapid variant of the former) and fade‐in/out (tape moving sideways over the head—results in periodic high frequency loss and blurring of the sonic image). These can be attacked by time variant filtering, delay processing and judicious use of feedback.

Nonlinearities are also an inherent part of the analog picture. These arise in throughout the signal chain, from microphones to loudspeakers. Overdriven amplifiers produce harmonic and intermodulation distortion (harmonics). Poor gain circuits exhibit zero crossing distortion (the signal does not go through zero smoothly, but lies in zero even if a small voltage is already present in the input). Microphones and loudspeakers can exhibit serious couplings between nonlinear distortion and feedback (like some of the oscillation produced when the bass cone of a speaker hits the element core). Coupling transformers display hysteresis (a mode of nonlinear behavior where the output of the element does not return to rest when input is removed—the output always tries to stay closer to the previous peak value). Funky things also happen when an analog tape is saturated—the result is a mixture of bad frequency response, hysteresis, soft clipping and so on. All in all, filtering, feedback and table lookup usually suffice when effects of this sort are needed.

Finally we have noise. The most usual source is thermal noise in electronic components. This is the constant hiss present in all electronic equipment as a result of moving electrons. Other noise sources include dust, spot noise (from the master platter metallization procedure) and scratches on both vinyl and tape. A turntable can insert very low frequency (VLF) hum from the actuator into the outgoing signal, as well. Power line coupling leads to the omnipresent 50/60Hz hum. All these are easily inserted into the sound and when compounded by the distortions described in the previous sections, sound really funky.

In the digital domain, we have a unique sort of distortion which can be used to advantage when putting the grit back in: used sparingly, aliasing can add significantly to a track.