[LAD] GuitarSynth

Fons Adriaensen fons at linuxaudio.org
Sun Apr 26 09:10:30 UTC 2015

On Sat, Apr 25, 2015 at 08:35:56PM +0200, Albert Graef wrote:
> > Question: I tried a demo product which did polyphony, with similar
> > latency as my app, which claimed to have a full version with
> > near-zero latency.
> >
> > Is this actually possible?
> Sounds like snake oil to me, but don't take my word for it.

Sound very unlikely. One problem with polyphonic pitch detection
is two or more notes that share a number of harmonic frequencies
and start at the same time. This will occur with many chords that
contain simple intervals like octave and fifth. 

It is possible to detect which harmonics are shared (they will
have a different amplitude / phase profile) but this requires
tracking them for some time and hence additional latency.

I suspect the same is even true for human perception. If we hear
a perfect fifth chord we may have the impression to have detected
that it consist of two notes immediately. But there are many
examples of our brain playing tricks and 'backdating' the result
of an observation which has actually taken more time than we

Real-time polyphonic pitch detection is still a research topic,
just look at the publication dates of some of papers already
mentioned. For a guitar it may be easier than the general case
due to the restriced frequency range of each string and in
general a clear attack of each note.



A world of exhaustive, reliable metadata would be an utopia.
It's also a pipe-dream, founded on self-delusion, nerd hubris
and hysterically inflated market opportunities. (Cory Doctorow)

More information about the Linux-audio-dev mailing list