28 Nov
2007
28 Nov
'07
1:39 p.m.
Hi
I'm trying to make a guitar tuner ladspa plugin.
I already wrote a first draft, but listplugins can't find my plugin in
the library :
$ listplugins
...
/usr/lib/ladspa/TunerUnit.so: <---
/usr/lib/ladspa/tap_doubler.so:
...
it could be because I've chosen a uniqueID already in use.
is there a list or way to register a uniqueID ?
here is a part of my _init() function:
g_psTUDescriptor->UniqueID
= 4053;
g_psTUDescriptor->Label
= strdup("tner");
g_psTUDescriptor->Properties
= LADSPA_PROPERTY_HARD_RT_CAPABLE;
g_psTUDescriptor->Name
= strdup("Universal Digital Tuner Unit");
g_psTUDescriptor->Maker
= strdup("Rémi Thébault");
g_psTUDescriptor->Copyright
= strdup("None");
By the way, for the moment I use FFTW for the frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
For a 0.7 Hz precision with fft, I need 65536 samples, what will take a
lot CPU load every 1.5 sec.
It is obvious that so much CPU Load is no acceptable for a small plugin
in a realtime application.
Rémi
28 Nov
28 Nov
3:11 p.m.
Rémi Thébault wrote:
Hi
By the way, for the moment I use FFTW for the frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
depending on your signal you could just count zero-crossings (with some
filtering to get rid of any DC and higher harmonics)
mfga.dsr
IOhannes
4:03 p.m.
On Wed, 2007-11-28 at 18:39 +0100, Rémi Thébault wrote:
By the way, for the moment I use FFTW for the
frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
For a 0.7 Hz precision with fft, I need 65536 samples, what will take a
lot CPU load every 1.5 sec.
It is obvious that so much CPU Load is no acceptable for a small plugin
in a realtime application.
There are a couple of algorithms that can get decent results with just a
couple of periods of data, but I think they all use some form of FFT or
convolution. A web search for autocorrelation and "fundamental
frequency" should find something.
--ll
4:09 p.m.
On Wednesday 28 November 2007, Rémi Thébault wrote:
By the way, for the moment I use FFTW for the
frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
For a 0.7 Hz precision with fft, I need 65536 samples, what will take a
lot CPU load every 1.5 sec.
Mario Lang has written a commandline tuner using FFTW, too.. It works fine in
realtime it seems. Maybe it's worth to take a look..
http://delysid.org/tuneit.html
Flo
--
Palimm Palimm!
http://tapas.affenbande.org
29 Nov
29 Nov
4:52 a.m.
On Wed, Nov 28, 2007 at 06:39:00PM +0100, Rémi Thébault wrote:
Is there a way to identify a frequency with quite high
precision without
computing a big spectrum analysis ?
- Do a series of short FFTs with a raised cosine window and L/2 overlap,
then use the evolution of the phase to find the exact frequency.
- One short FFT to find approximate frequency, then start a PLL
to track the signal.
Both can be very precise if implemented correctly.
--
FA
Lascia la spina, cogli la rosa.
5:32 a.m.
On Wed, 2007-11-28 at 18:39 +0100, Rémi Thébault wrote:
By the way, for the moment I use FFTW for the
frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
For a 0.7 Hz precision with fft, I need 65536 samples, what will take a
lot CPU load every 1.5 sec.
If you use the peak spectrum (by calculating the peak locations through
parabolic interpolation) and picking the lowest peak, you can get quite
precise results with a much smaller window size (~1024-2048 samples
should be fine).
This method has some flaws though, particularly if your sound has a weak
or missing fundamental. In this case you might want to look at
Autocorrelation/AMDF/ASDF, which don't require the fundamental to be the
strongest component, or even present at all.
You might be able to find some useful code in libxtract
(http://libxtract.sourceforge.net/): xtract_peak_spectrum(),
xtract_f0(), xtract_failsafe_f0(), xtract_asdf() etc.
One approach might be to do a pre-filtering stage to determine whether
or not most of the energy in the spectrum falls above or below a certain
frequency threshold, and then pick a PDA based on the result. In my
experience, it tends to be low instrumental sounds (e.g. low piano
notes) that tend to have a 'missing fundamental'.
Jamie
--
www.postlude.co.uk
30 Nov
30 Nov
5:21 p.m.
Rémi Thébault wrote:
By the way, for the moment I use FFTW for the
frequency identification.
Is there a way to identify a frequency with quite high precision without
computing a big spectrum analysis ?
You might want to look at the Goertzel algorithm (see a DSP book or
http://en.wikipedia.org/wiki/Goertzel_algorithm).
Note, this doesn't effect how big of a window you need to resolve a
particular frequency. But, if you want to compute a big FFT, but will
only be looking at a relatively small subset of the freq bins it
outputs, then this can be more efficient.
Ken
5991
days inactive
5993
days old
linux-audio-dev@lists.linuxaudio.org
6 comments
7 participants
participants (7)
-
Florian Schmidt -
Fons Adriaensen -
IOhannes m zmoelnig -
Jamie Bullock -
Ken Schutte -
Lars Luthman -
Rémi Thébault