LYRA-4 is based on four generators, which we will call voices. Its design does not follow the traditional approach of a subtractive synthesis VCO. Instead of relying linearly or logarithmically on control voltage, they resemble the tone generators of old electric organs. Hence the use of the term "voices" instead of "VCOs". Lyra makes extensive use of non-linearity, and its voices are built in such a way that allow this characteristic to be fully expressed.

The voices are organized in two pairs (12, 34), forming a tree-like structure.

The voices can operate in electric organ mode or in FM synthesis mode, where each voice and its envelope act as independent FM operators. The influence of each voice in the FM synthesis decreases as its envelope decays.

LYRA-4 also features the following modules:
HYPER LFO: a complex low-frequency generator whose waveform is synthesized from two simple LFOs, by summing or multiplying their frequencies. It also includes a sync mode. It can modulate the selected voice pairs and the MOD DELAY.
MOD DELAY: a delay module with feedback. Its distinctive feature is the ability to self-modulate, where the output signal modulates the sampling frequency of the delay, generating unique effects.
DISTORTION: it is located at the end of the chain, after the delay. This allows the delay to also influence the distortion, generating greater sonic interaction.
Despite its experimental nature, LYRA-4 is a professional instrument. Its output dynamics and frequency response are balanced and ready for live performances or studio recordings without the need for additional processing.