denemo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Denemo-devel] Audio input for Educational


From: Richard Shann
Subject: Re: [Denemo-devel] Audio input for Educational
Date: Sat, 11 Sep 2010 09:37:46 +0100

I have thought a bit about this, and if your experiments with the audio
detection of notes from different instruments convince you that it is
worthwhile to go ahead the next step would be to clone the code used to
do d-GetMidi. Like d-Command and others, this works by running the gtk
main loop, allowing the gui to stay refreshed while the scheme script is
waiting a note in the input.
I can help with this step - the important thing would be to look and
think at how the note recognition is responding to instruments (voice
even harder) that might be used. What happens if you restrict the range
so that *only* the note desired would be detected? IIRC the parameter
for the number of repeats required before a tuning note is accepted is
hardwired in the C code, this might need to be extracted. 

When not in tuning mode *every* note detected will be accepted as a
note, since it is for playing in music into a score - quite the opposite
of tuning mode. In fact, IIRC there is a control to ignore duplicate
notes in the input, which reflects the fact that when playing in
acoustically it sometimes did a double detection of a single note. With
that on, if you were playing in a piece with a repeated note you had to
play it four times:)

Richard



On Fri, 2010-09-10 at 19:31 +0100, Richard Shann wrote:
> Jeremiah
> 
> Before doing a lot of implementation work
> I think it might be a good idea for you to try out the note detection
> using the tuning option of the popup. Try with different instruments,
> and (potentially) different settings.
> 
> When tuning is selected it does some repeated checking of the note
> detected (IIRC) and the parameters set are tuned to detect my
> harpsichord. Other instruments may need other settings.
> Also note that there is onset detection going on - a steady note may not
> be detected, while repeating it may.
> 
> Without the tuning option it detects notes much more readily (IIRC) but
> may easily get spurious ones. Also, note that there is a frequency
> range, set very high 4500 Hz I think, which should be lowered.
> 
> Richard
> 
> 
> 
> 
> _______________________________________________
> Denemo-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/denemo-devel




reply via email to

[Prev in Thread] Current Thread [Next in Thread]