Chapter 4: Filter without coffee?

Now that we have got to know different sound generators, let’s look at how the “raw material” obtained in this way can be refined and processed – a classic sound processing is filtering. Certain frequency ranges are removed from the source material, while others are partially emphasized (and there are also sonic side effects – after all, we are not in the laboratory).

Some filters are reverences to the circuits of »historic« synthesizers. This gives interesting hints about their sound characteristics, but one should not expect a 1:1 copy of the original sound. In addition to the – certainly very important – filter, amplifiers, oscillators, mixers and envelope curves also play an important role in the creation of a “signature” sound.
A lowpass filter lets low frequencies pass (hence the name) and can filter out high frequencies above an adjustable “cutoff” frequency.
Multimode filters offer several other filter types in addition to the lowpass filter. A highpass filter lets high frequencies pass and can filter out low frequencies below the cut-off frequency, the bandpass filter only lets through a certain frequency band and filters above and below this frequency band, a notch or band-rejection filter cuts out a frequency band.
There are also special – somewhat more unusual – filters: lowpass gates, formant filters, filter banks that do not fit into the categories described above.
A vocoder is also a filter – or better: a series of filters with a special kind of filter control.
Finally, the phasers should also be mentioned: In principle, they also belong to the filters. They are so-called allpass filters, the effects of which we usually only clearly perceive in connection with the original signal.

In addition to the above-mentioned classification of filters according to the type of their intervention in the audio material, there is another important characteristic of filters – their edge steepness: how much of the filtered frequency range remains audible even just past the cutoff-frequency? This is usually specified in dB / octave: With a “24dB” lowpass filter, the frequency 1 octave above the cut-off frequency is already 24dB quieter than without a filter. That’s quite an effective filter, but musically it doesn’t necessarily make more sense than a 12dB or 6dB filter.

For polyphonic usage, I usually prefer 12dB or even 6dB filters for their smoother integration of each voice, but that’s just me.

Excursus: famous “role models”, how close are we to them?

Old synthesizers – if possible from the 70s or 80s – are wonderful: They sound great, warm, fat and have often made a noticeable contribution to the history of pop music. Unfortunately, these treasures are not only expensive, but often also require maintenance (and thus even more expensive). Alternatives are needed!

There are currently “historic manufacturers” who build new instruments, there are virtual digital emulations for the DAW or as hardware and there are always modules – especially filters – that are supposed to reproduce the charm of the old devices.

In all of these categories you can find fantastic instruments that, if you had actually made music in the 70’s or 80’s, you would have mortgaged your house and its inhabitants. But alas: These “late-borns” probably can’t please anyone. There always seems to be a smaller or larger piece of behavior, sound, smell, etc. missing. At least if you want to believe the knowledgeable synthesizer forums and their residents.

However, this discussion is rather fruitless and should not be conducted here. I would like to understand the similarities of the filter modules with their role models as a rough tonal direction, as a way of orientation. A modular synthesizer is not suitable for a 1:1 copy anyway, too many components (amplifier, VCOs, envelopes, etc.) differ from the original. Examples of famous “idea generators” of Doepfer modules are:

Filters:	Model / based on:
A-101-1	Steiner-Parker Filter
A-101-2	Buchla 292
A-102	EMS AKS
A-103	Roland TB303
A-104	Trautonium
A-105	Prophet 5 etc.
A-106-1	Korg MS-20
A-106-5	Oberheim SEM
A-106-6	Oberheim Xpander
A-107	ERWIK, Oberheim Xpander
A-108, A-120	Moog
A-122	Oberheim
A-124	EDP Wasp

Lowpass filters

A lowpass filter (LPF) is a filter that filters out higher frequency components and lets lower ones pass. They still define what might be called the “typical synthesizer sound”: most synthesizers of the 70’s only had a lowpass filter.

The filter in the A-109 VC Signal Processor is actually a “normal” lowpass filter, but it is discussed in the “Special Filters” section because it has a significantly expanded range of functions compared to other lowpass filters with VCA and panner.

The filter in the A-111-5 Mini Synthesizer Voice is similar. In addition to the 24 dB low-pass filter, an oscillator, an ADSR generator and two LFOs are also installed here. The description can be found in the chapter about the oscillators.

Posts on (pure) lowpass filters

Multimode filters and “special” filters

While a lowpass filter lets through low frequencies below the cutoff frequency (and filters out the high frequencies), multimode filters can also be switched to other operating modes or filter modes. The most common are:

Highpass (in which low frequencies are filtered, HPF),
Bandpass (where only an adjustable frequency band is allowed to pass, BPF) and
Band-rejection / notch filter (where an adjustable frequency band is filtered out).

In addition, there are other filter modes which, for example, have several “notches” in the frequency response that filter high and low frequency components with different steepness, etc. You will find either pure low-pass or multimode filters on the market today, pure high-pass filters such as the modules A-123 or A-123-2 are rather rare.

With some filters, these modes can be used simultaneously via several output sockets (e.g. A-121 or A-121-2, with restrictions A-106-6), some have faders between two filter modes (e.g. A-106-5 or A-124) – and then there’s the A-101-1, which doesn’t use different outputs for the different filter modes, but different inputs, so you can treat different signals differently in the same filter.

The filters discussed so far all roughly do what is also known from keyboard synthesizers: lowpass, highpass, maybe a bit of notch, but basically known types of sound shaping.

In addition, there are also filters that follow very unusual concepts, sound very “weird” or trace historical ideas for sound shaping, such as the Trautonium filter. These are basically also lowpass, bandpass, etc. filters, but with very surprising solutions in detail.

Or filters that are not all that unusual in their core, but can come up with interesting additional functions, such as the A-107 Morphing Filter.

Posts on multimode filters

Posts on (pure) highpass filters

Posts on (pure) bandpass filters

Vocoders

Vocoders consist of a series of fixed-frequency bandpass filters connected in parallel. That’s not particularly special, a graphic equalizer or a fixed filter bank is quite comparable.

What is interesting, however, is how the amplitudes of these filters are controlled: an audio signal (“modulator”) is broken down into frequency bands that are similar to those of the filter. The measured loudness of each of these frequency bands is then used to control the filter. If a broadband audio signal (“carrier” or “carrier”) is fed into this filter, the carrier signal is given the “acoustic fingerprint” of the modulator signal. And you can let the robots do the talking.

Unfortunately, Doepfer has discontinued the production of the entire Vocoder module series, so that the corresponding 5 modules can only be found on the used market (and are extremely rare!).

Posts on the vocoder

Phasers

A phaser shifts the phase position of its input signal and mixes the result back into the output signal. This results in characteristic frequency cancellations (comb filters).

Posts on (pure) phasers

Posts on multimode filters with phasers

Previous chapter:

Generating tones and noise

Next chapter:

Other sound shapers