Forums > MaxMSP

piano pitch detection ?

October 8, 2007 | 4:15 pm

I find it very hard to detect the pitch of piano sound because the result is very unstable. Even if the lowest note on a piano can show a very high pitch detection result. I am thinking that maybe I can capture the correct pitch when there is a new attack. But I still cannot succeed.

Have anybody used piano pitch detection result to determin whether to trigger some events or not ? (I am not planning to buy a piano sensor.) Any better ideas ?

I actually asked the same question on the forum before and did not get a good solution to it. But I think that the fundamental frequency of a piano pitch should have the highest level. If so, there should be some way to differentiate, eg., a higher pitch D and a lower pitch D by capturing data only at the beginning of an attack. Am I right ?

Please give me some suggestions. Thanks.


October 8, 2007 | 5:25 pm

Do you know about overtones?

Your best bet, from what I can understand of your post, is to route your audio input through a variable low pass filter and try and get just the fundamental tone(s) through to your frequency interpreter. It’s probably getting confused by the overtones.


October 8, 2007 | 5:44 pm

For monodic intruments, have a look at fiddle~ or yin~.
For polyphonic (!!!) instruments (especially piano with template),
have a look at the Arshia Cont’s work with Sparse Non-negative Matrix
Decomposition.

Pure Max external:

http://cosmal.ucsd.edu/arshia/index.php?n=Main.Transcribe

FTM external:

http://cosmal.ucsd.edu/arshia/index.php?n=Main.Multipitch

best,

O./


October 8, 2007 | 6:40 pm

If you wanna have something else than transcribe~, you should use a
mechanical midi detector. There is a fairly good and reactive one
from Moog ($1495) : PianoBar

http://www.moogmusic.com/detail.php?main_product_id=71

Be careful, this is fragile. There is a very annoying and useless
sensor on the left side you have to tape to avoid program changes (or
something like that) !!!
The sounds from this device are crap.

O./


October 8, 2007 | 7:05 pm

Or you could use the Disklavier in the MEIT. (for you Univ. of North
Texas alumni)

The good news about piano notes is that you don’t have to worry about
reattacks…

Peter McCulloch


October 9, 2007 | 2:46 pm

transcribe~ looks great,

i have a g4 so it runs really crunchy on mine, so i cant test it too well

for those that have worked with it on faster machines, hows the
tracking/latency?
howd you mic the piano?

On 10/8/07, olivier pasquet wrote:
> For monodic intruments, have a look at fiddle~ or yin~.
> For polyphonic (!!!) instruments (especially piano with template),
> have a look at the Arshia Cont’s work with Sparse Non-negative Matrix
> Decomposition.
>
> Pure Max external:
> http://cosmal.ucsd.edu/arshia/index.php?n=Main.Transcribe
>
> FTM external:
> http://cosmal.ucsd.edu/arshia/index.php?n=Main.Multipitch
>
> best,
>
> O./
>


Viewing 6 posts - 1 through 6 (of 6 total)