< Back to IRCAM Forum

Pipo.yin

MuBu
#1

Dear list,

I already submitted a question about the differences between yin~ gbr.yin and pipo.yin in the MAX Sound Box group about the differences between the different yin implementations. I experimented with pipo.yin and it seems that the frequency values I get are not half as good as the real time values from the yin~ stand alone. It does not seem like I use very different parameters (I am looking at the frequencies using MIDI values). The quality (periodicity) values strike me as odd as well :

The mubu.process values:

  
----------begin_max5_patcher----------  
643.3ocsTssbaCBD8Y4uBFd1QCRVWrxS4+noSGDZsMYj.M.x1oYx+dWPlZ2T  
GOdRm9f.1iVXuc18sEIzV8QvRIOR9FII4sEIIAHOPxI4D5.+nnmaCpQE5gAP  
4nKm+mCN5B34oD7WiSNf7pTQ5.qvHGcZiMppZZPO45AW3gXmPG4NwNoZ6OLf  
vM6Hk0o4kEqZJWRxXUMoLbecXKOOkQ99oaJ6BFV29xC40QiXcu1CA7HxFsxs  
gKBfqt.yJ+Y.KqHkc1CkpnCl4wdewB+xx6L4zN4bZ00C3rSnyPtWGg4nk1xU  
ao+NptR9.i55OjOxKhqWKcrh84oi+4XTAGPa7W0+go1ozQiV.VKYGzO9fG4A  
LJH7oNolX6kB3Qja7nUv6AxSJ9vLW4o30XgiaM3YLKgbIhT4WeJb2TeEijyJ  
VGA1oGIkY4QQk1LPF0G.Ch3sQ5lIkfzq2Fka4VfjkxJaJpVEAkJLiPVmlUWV  
2zbAJ+HhVUk0TVGQQ2xqL6RYTsrnrP2KUcvwyZf4aDz5UQp1nIamjjmoRkCL  
dVIYXp2IOv2CKI1c7Qf.p8PudDkw6gQjtCHFtT0pO7LEeU9vXOX3Xe15pXU.  
MJHzSpPoH+pzuhOi9QWN+I8M03w6kIlwJCclYU098UkWmKVUbmbw7u.Wj6bl  
I4s6275DT9LIg9o4haF7EMWLWpdN5KY2ZtTc0WnQL7PgR5GFLG7IO9elQvdE  
iHFAQqRN6O3nXmTwcRbxzYk70kKTZmrqCTWNXtSZ4s8PHXXWs1bu9iebz+O+  
YNcwGG2CF6o2L3JHU4EcnzWuLHJUyhgWjZf8xn9UADtAK4NrdOYlKUGO0gQ8  
MgFExzlKUnkeewu.jzwz4.  
-----------end_max5_patcher-----------

For yin~ I use the standard settings (down: 2, lowest: 60, analyses: 5.)

#2

Hi Hans,

it is the slice that needs to be adapted for pipo.yin (you can look how gbr.yin help does it):

yin.minfreq 50, slice.size 1761, slice.hop 100

Maybe we should build the slicing into pipo.yin, as it also depends on the minfreq.

Best…
…Diemo

#3

Hi Diemo,

Do you mean that it is not possible to optimize this for pipo.yin at the moment? Using these gbr.yin values for slice does not give the expected results. It does help to change the yin.minfreq though in my case.

Hans.

#4

What is your input signal? Window size and minfreq are dependent on each other; the downsample parameter might also have an influence

#5

I mostly use this trumpet-sample of my own playing. Some flagioletti things, hi to low and with and without vibrato.

Hans-omfloerst.wav (1.09 MB)