There doesn't seem to be much related material published on the
internet - perhaps because SERAFs are a new idea. At present, the aim
of the SERAF project is to promote internet publication of relevant
material, in an effort to prevent the existence of patents that could
make new music illegal in the future. To this end, I hope you will
forward the 'story so far' to anyone you think might want to
cooperate, such as responsible composers or programmers with an
amateur interest in digital audio and computer music, so that they can
contribute with their own ideas.
Google's archive of the relevant threads.
> firstname.lastname@example.org (Stephen Norris) wrote
>> email@example.com wrote:
>> >Your interest in the actual neuron stimulation in relation to the
>> >sound, is the intriging aspect of your intentions.
>> >My square waves were only of many posible air compression functions.
>> >Experiment with sine wave also. Wavelets also.
>> >To obtain neuron unique sounds requires the musician to look to the
>> >inventor sometimes. Maybe try using the wavelets of digital signal
>> >processors as the note? A beautiful harmonic scale could be
>> >Maybe a fancy synthesizer has wavelets?
>> Sine wave and 'square' wave analysis are very generalised mathematical
>> methods of analysing and constructing audio waveforms. Neither bear much
>> relation to what is actually happening in the ears and brain when a listener
>> hearing, for example, the spoken word 'three'.
>> In 'wavelet' analysis, sound consists of context-dependent wavelets;
>> can be part of larger wavelets and can have subsidiary wavelets. In the
>> (Synaptic Energy Redistribution Audio Filter) that I outlined, the
>> wavelets would be the individual amplitude/duration pairs corresponding to
>> individual synapse firings (or their functional equivalent) which occur in
>> Larger wavelets would be constructed appropriately by the composer,
>> depending on what effect they are trying to achieve.
> I think what you're talking about has already been realized:
> Most relevant:
The Neuron is clearly a superb synthesizer/mixer and features many powerful
innovations, notably in its impressive user-interface. The website shows
Stevie Wonder with the instrument, and there can scarcely be a better
The designers have trained a neural network to recognise significant tonal
characteristics of sounds played to it, and users can change the sound quality
by manipulating these characteristics. ("Neuron has a basic concept of the
sonic qualities of a sound and how they interact with the human auditory
perception. [...] Years of research [...] have been invested to deduct a set
of parameters from any given sound based on pattern recognition and proprietary
A SERAF (synaptic energy redistribution audio filter) is NOT a neural network,
but uses digital signal processing techniques similar to waveshaping in an
attempt to change the tone-colour of the input sound by triggering specific
conceptual (that is, hypothetical but possibly real) neural synapses within the
brain of the listener.
I'd be interested to have an opinion from Neuron's designers on SERAFs. It is
possible, but rather unlikely, that Neuron's neural network already
incorporates all the potential power of SERAF methods, but in any case it would
certainly be an interesting exercise to try to develop unusual SERAF sounds
that Neuron does not understand - I'm sure Neuron's designers would want to be
involved in such experiments. I wonder how horribly complicated the legal
situation would become if the 'propriety' of Neuron's methods were ever
questioned and the precise workings of a neural network became evidence in
Reply by Stephan M. Bernsee●May 3, 20042004-05-03
Sounds interesting, although I see no real connection to the <Neuron>
synthesizer from what I've read here. To be honest, I don't quite
understand how this SERAF is supposed to work anyway - do you have
more information? The bits I was able to dig out in this thread don't
make much sense to me.