"""
Fast Fourier Transform.
A Fast Fourier Transform (FFT) is an efficient algorithm to compute
the discrete Fourier transform (DFT) and its inverse (IFFT).
The objects below can be used to perform sound processing in the
spectral domain.
"""
"""
Copyright 2009-2015 Olivier Belanger
This file is part of pyo, a python module to help digital signal
processing script creation.
pyo is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as
published by the Free Software Foundation, either version 3 of the
License, or (at your option) any later version.
pyo is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with pyo. If not, see <http://www.gnu.org/licenses/>.
"""
from ._core import *
from ._maps import *
[docs]class FFT(PyoObject):
"""
Fast Fourier Transform.
FFT analyses an input signal and converts it into the spectral
domain. Three audio signals are sent out of the object, the
`real` part, from bin 0 (DC) to bin size/2 (Nyquist), the
`imaginary` part, from bin 0 to bin size/2-1, and the bin
number, an increasing count from 0 to size-1. `real` and
`imaginary` buffer's left samples up to size-1 are filled
with zeros. See notes below for an example of how to retrieve
each signal component.
:Parent: :py:class:`PyoObject`
:Args:
input: PyoObject
Input signal to process.
size: int {pow-of-two > 4}, optional
FFT size. Must be a power of two greater than 4.
The FFT size is the number of samples used in each
analysis frame. Defaults to 1024.
overlaps: int, optional
The number of overlaped analysis block. Must be a
positive integer. More overlaps can greatly improved
sound quality synthesis but it is also more CPU
expensive. Defaults to 4.
wintype: int, optional
Shape of the envelope used to filter each input frame.
Possible shapes are :
0. rectangular (no windowing)
1. Hamming
2. Hanning
3. Bartlett (triangular)
4. Blackman 3-term
5. Blackman-Harris 4-term
6. Blackman-Harris 7-term
7. Tuckey (alpha = 0.66)
8. Sine (half-sine window)
.. note::
FFT has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
FFT has no `mul` and `add` attributes.
Real, imaginary and bin_number parts are three separated set
of audio streams. The user should call :
| FFT['real'] to retrieve the real part.
| FFT['imag'] to retrieve the imaginary part.
| FFT['bin'] to retrieve the bin number part.
>>> s = Server().boot()
>>> s.start()
>>> a = Noise(.25).mix(2)
>>> fin = FFT(a, size=1024, overlaps=4, wintype=2)
>>> t = ExpTable([(0,0),(3,0),(10,1),(20,0),(30,.8),(50,0),(70,.6),(150,0),(512,0)], size=512)
>>> amp = TableIndex(t, fin["bin"])
>>> re = fin["real"] * amp
>>> im = fin["imag"] * amp
>>> fout = IFFT(re, im, size=1024, overlaps=4, wintype=2).mix(2).out()
"""
def __init__(self, input, size=1024, overlaps=4, wintype=2):
pyoArgsAssert(self, "oiIi", input, size, overlaps, wintype)
PyoObject.__init__(self)
self._real_dummy = []
self._imag_dummy = []
self._bin_dummy = []
self._input = input
self._size = size
self._overlaps = overlaps
self._wintype = wintype
self._in_fader = InputFader(input)
in_fader, size, wintype, lmax = convertArgsToLists(self._in_fader, size, wintype)
self._base_players = []
for j in range(overlaps):
for i in range(lmax):
hopsize = wrap(size, i) * j // overlaps
self._base_players.append(FFTMain_base(wrap(in_fader, i), wrap(size, i), hopsize, wrap(wintype, i)))
self._real_objs = []
self._imag_objs = []
self._bin_objs = []
for j in range(len(self._base_players)):
self._real_objs.append(FFT_base(wrap(self._base_players, j), 0, self._mul, self._add))
self._imag_objs.append(FFT_base(wrap(self._base_players, j), 1, self._mul, self._add))
self._bin_objs.append(FFT_base(wrap(self._base_players, j), 2, self._mul, self._add))
self._base_objs = [Sig(0)] # Dummy objs to prevent PyoObjectBase methods to fail.
self._init_play()
def __len__(self):
return len(self._real_objs)
def __getitem__(self, str):
if str == "real":
self._real_dummy.append(Dummy([obj for i, obj in enumerate(self._real_objs)]))
return self._real_dummy[-1]
if str == "imag":
self._imag_dummy.append(Dummy([obj for i, obj in enumerate(self._imag_objs)]))
return self._imag_dummy[-1]
if str == "bin":
self._bin_dummy.append(Dummy([obj for i, obj in enumerate(self._bin_objs)]))
return self._bin_dummy[-1]
[docs] def get(self, identifier="real", all=False):
"""
Return the first sample of the current buffer as a float.
Can be used to convert audio stream to usable Python data.
"real", "imag" or "bin" must be given to `identifier` to specify
which stream to get value from.
:Args:
identifier: string {"real", "imag", "bin"}
Address string parameter identifying audio stream.
Defaults to "real".
all: boolean, optional
If True, the first value of each object's stream
will be returned as a list. Otherwise, only the value
of the first object's stream will be returned as a float.
Defaults to False.
"""
if not all:
return self.__getitem__(identifier)[0]._getStream().getValue()
else:
return [obj._getStream().getValue() for obj in self.__getitem__(identifier).getBaseObjects()]
[docs] def play(self, dur=0, delay=0):
dur, delay, lmax = convertArgsToLists(dur, delay)
self._autoplay(dur, delay)
self._in_fader.play(dur, delay)
self._base_players = [obj.play(wrap(dur, i), wrap(delay, i)) for i, obj in enumerate(self._base_players)]
self._real_objs = [obj.play(wrap(dur, i), wrap(delay, i)) for i, obj in enumerate(self._real_objs)]
self._imag_objs = [obj.play(wrap(dur, i), wrap(delay, i)) for i, obj in enumerate(self._imag_objs)]
self._bin_objs = [obj.play(wrap(dur, i), wrap(delay, i)) for i, obj in enumerate(self._bin_objs)]
return self
[docs] def stop(self, wait=0):
self._autostop(wait)
self._in_fader.stop(wait)
[obj.stop(wait) for obj in self._base_players]
[obj.stop(wait) for obj in self._real_objs]
[obj.stop(wait) for obj in self._imag_objs]
[obj.stop(wait) for obj in self._bin_objs]
return self
[docs] def out(self, chnl=0, inc=1, dur=0, delay=0):
return self.play(dur, delay)
[docs] def setSize(self, x):
"""
Replace the `size` attribute.
:Args:
x: int
new `size` attribute.
"""
pyoArgsAssert(self, "i", x)
self._size = x
x, lmax = convertArgsToLists(x)
poly = len(self._base_players) // self._overlaps
for j in range(self._overlaps):
for i in range(poly):
hopsize = wrap(x, i) * j // self._overlaps
self._base_players[j * poly + i].setSize(wrap(x, i), hopsize)
[docs] def setWinType(self, x):
"""
Replace the `wintype` attribute.
:Args:
x: int
new `wintype` attribute.
"""
pyoArgsAssert(self, "i", x)
self._wintype = x
x, lmax = convertArgsToLists(x)
[obj.setWinType(wrap(x, i)) for i, obj in enumerate(self._base_players)]
@property
def input(self):
"""PyoObject. Input signal to process."""
return self._input
@input.setter
def input(self, x):
self.setInput(x)
@property
def size(self):
"""int. FFT size."""
return self._size
@size.setter
def size(self, x):
self.setSize(x)
@property
def wintype(self):
"""int. Windowing method."""
return self._wintype
@wintype.setter
def wintype(self, x):
self.setWinType(x)
[docs]class IFFT(PyoObject):
"""
Inverse Fast Fourier Transform.
IFFT takes a signal in the spectral domain and converts it to a
real audio signal using an inverse fast fourier transform.
IFFT takes two signals in input, the `real` and `imaginary` parts
of an FFT analysis and returns the corresponding real signal.
These signals must correspond to `real` and `imaginary` parts
from an FFT object.
:Parent: :py:class:`PyoObject`
:Args:
inreal: PyoObject
Input `real` signal.
inimag: PyoObject
Input `imaginary` signal.
size: int {pow-of-two > 4}, optional
FFT size. Must be a power of two greater than 4.
The FFT size is the number of samples used in each
analysis frame. This value must match the `size`
attribute of the former FFT object. Defaults to 1024.
overlaps: int, optional
The number of overlaped analysis block. Must be a
positive integer. More overlaps can greatly improved
sound quality synthesis but it is also more CPU
expensive. This value must match the `overlaps`
atribute of the former FFT object. Defaults to 4.
wintype: int, optional
Shape of the envelope used to filter each output frame.
Possible shapes are :
0. rectangular (no windowing)
1. Hamming
2. Hanning
3. Bartlett (triangular)
4. Blackman 3-term
5. Blackman-Harris 4-term
6. Blackman-Harris 7-term
7. Tuckey (alpha = 0.66)
8. Sine (half-sine window)
.. note::
The number of streams in `inreal` and `inimag` attributes
must be egal to the output of the former FFT object. In
most case, it will be `channels of processed sound` * `overlaps`.
The output of IFFT must be mixed to reconstruct the real
signal from the overlapped streams. It is left to the user
to call the mix(channels of the processed sound) method on
an IFFT object.
>>> s = Server().boot()
>>> s.start()
>>> a = Noise(.25).mix(2)
>>> fin = FFT(a, size=1024, overlaps=4, wintype=2)
>>> t = ExpTable([(0,0),(3,0),(10,1),(20,0),(30,.8),(50,0),(70,.6),(150,0),(512,0)], size=512)
>>> amp = TableIndex(t, fin["bin"])
>>> re = fin["real"] * amp
>>> im = fin["imag"] * amp
>>> fout = IFFT(re, im, size=1024, overlaps=4, wintype=2).mix(2).out()
"""
def __init__(self, inreal, inimag, size=1024, overlaps=4, wintype=2, mul=1, add=0):
pyoArgsAssert(self, "ooiIiOO", inreal, inimag, size, overlaps, wintype, mul, add)
PyoObject.__init__(self, mul, add)
self._inreal = inreal
self._inimag = inimag
self._size = size
self._overlaps = overlaps
self._wintype = wintype
self._in_fader = InputFader(inreal)
self._in_fader2 = InputFader(inimag)
in_fader, in_fader2, size, wintype, mul, add, lmax = convertArgsToLists(
self._in_fader, self._in_fader2, size, wintype, mul, add
)
self._base_objs = []
ratio = lmax // overlaps
for i in range(lmax):
hopsize = wrap(size, i) * ((i // ratio) % overlaps) // overlaps
self._base_objs.append(
IFFT_base(
wrap(in_fader, i),
wrap(in_fader2, i),
wrap(size, i),
hopsize,
wrap(wintype, i),
wrap(mul, i),
wrap(add, i),
)
)
self._init_play()
def __len__(self):
return len(self._inreal)
[docs] def setInReal(self, x, fadetime=0.05):
"""
Replace the `inreal` attribute.
:Args:
x: PyoObject
New input `real` signal.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inreal = x
self._in_fader.setInput(x, fadetime)
[docs] def setInImag(self, x, fadetime=0.05):
"""
Replace the `inimag` attribute.
:Args:
x: PyoObject
New input `imag` signal.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inimag = x
self._in_fader2.setInput(x, fadetime)
[docs] def setSize(self, x):
"""
Replace the `size` attribute.
:Args:
x: int
new `size` attribute.
"""
pyoArgsAssert(self, "i", x)
self._size = x
x, lmax = convertArgsToLists(x)
ratio = len(self._base_objs) // self._overlaps
for i, obj in enumerate(self._base_objs):
hopsize = wrap(x, i) * ((i // ratio) % self._overlaps) // self._overlaps
self._base_objs[i].setSize(wrap(x, i), hopsize)
[docs] def setWinType(self, x):
"""
Replace the `wintype` attribute.
:Args:
x: int
new `wintype` attribute.
"""
pyoArgsAssert(self, "i", x)
self._wintype = x
x, lmax = convertArgsToLists(x)
[obj.setWinType(wrap(x, i)) for i, obj in enumerate(self._base_objs)]
[docs] def ctrl(self, map_list=None, title=None, wxnoserver=False):
self._map_list = [SLMapMul(self._mul)]
PyoObject.ctrl(self, map_list, title, wxnoserver)
@property
def inreal(self):
"""PyoObject. Real input signal."""
return self._inreal
@inreal.setter
def inreal(self, x):
self.setInReal(x)
@property
def inimag(self):
"""PyoObject. Imaginary input signal."""
return self._inimag
@inimag.setter
def inimag(self, x):
self.setInImag(x)
@property
def size(self):
"""int. FFT size."""
return self._size
@size.setter
def size(self, x):
self.setSize(x)
@property
def wintype(self):
"""int. Windowing method."""
return self._wintype
@wintype.setter
def wintype(self, x):
self.setWinType(x)
[docs]class CarToPol(PyoObject):
"""
Performs the cartesian to polar conversion.
The Cartesian system locates points on a plane by measuring the horizontal and
vertical distances from an arbitrary origin to a point. These are usually denoted
as a pair of values (X,Y).
The Polar system locates the point by measuring the straight line distance, usually
denoted by R, from the origin to the point and the angle of an imaginary line from
the origin to the point measured counterclockwise from the positive X axis.
:Parent: :py:class:`PyoObject`
:Args:
inreal: PyoObject
Real input signal.
inimag: PyoObject
Imaginary input signal.
.. note::
Polar coordinates can be retrieve by calling :
| CarToPol['mag'] to retrieve the magnitude part.
| CarToPol['ang'] to retrieve the angle part.
CarToPol has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
>>> s = Server().boot()
>>> snd1 = SfPlayer(SNDS_PATH+"/transparent.aif", loop=True, mul=.7).mix(2)
>>> snd2 = FM(carrier=[75,100,125,150], ratio=[.499,.5,.501,.502], index=20, mul=.1).mix(2)
>>> fin1 = FFT(snd1, size=1024, overlaps=4)
>>> fin2 = FFT(snd2, size=1024, overlaps=4)
>>> # get magnitudes and phases of input sounds
>>> pol1 = CarToPol(fin1["real"], fin1["imag"])
>>> pol2 = CarToPol(fin2["real"], fin2["imag"])
>>> # times magnitudes and adds phases
>>> mag = pol1["mag"] * pol2["mag"] * 100
>>> pha = pol1["ang"] + pol2["ang"]
>>> # converts back to rectangular
>>> car = PolToCar(mag, pha)
>>> fout = IFFT(car["real"], car["imag"], size=1024, overlaps=4).mix(2).out()
>>> s.start()
"""
def __init__(self, inreal, inimag, mul=1, add=0):
pyoArgsAssert(self, "ooOO", inreal, inimag, mul, add)
PyoObject.__init__(self, mul, add)
self._mag_dummy = []
self._ang_dummy = []
self._inreal = inreal
self._inimag = inimag
self._in_fader = InputFader(inreal)
self._in_fader2 = InputFader(inimag)
in_fader, in_fader2, mul, add, lmax = convertArgsToLists(self._in_fader, self._in_fader2, mul, add)
self._base_objs = []
for i in range(lmax):
self._base_objs.append(CarToPol_base(wrap(in_fader, i), wrap(in_fader2, i), 0, wrap(mul, i), wrap(add, i)))
self._base_objs.append(CarToPol_base(wrap(in_fader, i), wrap(in_fader2, i), 1, wrap(mul, i), wrap(add, i)))
self._init_play()
def __len__(self):
return len(self._inreal)
def __getitem__(self, str):
if str == "mag":
self._mag_dummy.append(Dummy([obj for i, obj in enumerate(self._base_objs) if i % 2 == 0]))
return self._mag_dummy[-1]
if str == "ang":
self._ang_dummy.append(Dummy([obj for i, obj in enumerate(self._base_objs) if i % 2 == 1]))
return self._ang_dummy[-1]
[docs] def get(self, identifier="mag", all=False):
"""
Return the first sample of the current buffer as a float.
Can be used to convert audio stream to usable Python data.
"mag" or "ang" must be given to `identifier` to specify
which stream to get value from.
:Args:
identifier: string {"mag", "ang"}
Address string parameter identifying audio stream.
Defaults to "mag".
all: boolean, optional
If True, the first value of each object's stream
will be returned as a list. Otherwise, only the value
of the first object's stream will be returned as a float.
Defaults to False.
"""
if not all:
return self.__getitem__(identifier)[0]._getStream().getValue()
else:
return [obj._getStream().getValue() for obj in self.__getitem__(identifier).getBaseObjects()]
[docs] def setInReal(self, x, fadetime=0.05):
"""
Replace the `inreal` attribute.
:Args:
x: PyoObject
New signal to process.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inreal = x
self._in_fader.setInput(x, fadetime)
[docs] def setInImag(self, x, fadetime=0.05):
"""
Replace the `inimag` attribute.
:Args:
x: PyoObject
New signal to process.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inimag = x
self._in_fader2.setInput(x, fadetime)
@property
def inreal(self):
"""PyoObject. Real input signal."""
return self._inreal
@inreal.setter
def inreal(self, x):
self.setInReal(x)
@property
def inimag(self):
"""PyoObject. Imaginary input signal."""
return self._inimag
@inimag.setter
def inimag(self, x):
self.setInImag(x)
[docs]class PolToCar(PyoObject):
"""
Performs the polar to cartesian conversion.
The Polar system locates the point by measuring the straight line distance, usually
denoted by R, from the origin to the point and the angle of an imaginary line from
the origin to the point measured counterclockwise from the positive X axis.
The Cartesian system locates points on a plane by measuring the horizontal and
vertical distances from an arbitrary origin to a point. These are usually denoted
as a pair of values (X,Y).
:Parent: :py:class:`PyoObject`
:Args:
inmag: PyoObject
Magintude input signal.
inang: PyoObject
Angle input signal.
.. note::
Cartesians coordinates can be retrieve by calling :
| PolToCar['real'] to retrieve the real part.
| CarToPol['imag'] to retrieve the imaginary part.
PolToCar has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
>>> s = Server().boot()
>>> snd1 = SfPlayer(SNDS_PATH+"/transparent.aif", loop=True, mul=.7).mix(2)
>>> snd2 = FM(carrier=[75,100,125,150], ratio=[.499,.5,.501,.502], index=20, mul=.1).mix(2)
>>> fin1 = FFT(snd1, size=1024, overlaps=4)
>>> fin2 = FFT(snd2, size=1024, overlaps=4)
>>> # get magnitudes and phases of input sounds
>>> pol1 = CarToPol(fin1["real"], fin1["imag"])
>>> pol2 = CarToPol(fin2["real"], fin2["imag"])
>>> # times magnitudes and adds phases
>>> mag = pol1["mag"] * pol2["mag"] * 100
>>> pha = pol1["ang"] + pol2["ang"]
>>> # converts back to rectangular
>>> car = PolToCar(mag, pha)
>>> fout = IFFT(car["real"], car["imag"], size=1024, overlaps=4).mix(2).out()
>>> s.start()
"""
def __init__(self, inmag, inang, mul=1, add=0):
pyoArgsAssert(self, "ooOO", inmag, inang, mul, add)
PyoObject.__init__(self, mul, add)
self._real_dummy = []
self._imag_dummy = []
self._inmag = inmag
self._inang = inang
self._in_fader = InputFader(inmag)
self._in_fader2 = InputFader(inang)
in_fader, in_fader2, mul, add, lmax = convertArgsToLists(self._in_fader, self._in_fader2, mul, add)
self._base_objs = []
for i in range(lmax):
self._base_objs.append(PolToCar_base(wrap(in_fader, i), wrap(in_fader2, i), 0, wrap(mul, i), wrap(add, i)))
self._base_objs.append(PolToCar_base(wrap(in_fader, i), wrap(in_fader2, i), 1, wrap(mul, i), wrap(add, i)))
self._init_play()
def __len__(self):
return len(self._inmag)
def __getitem__(self, str):
if str == "real":
self._real_dummy.append(Dummy([obj for i, obj in enumerate(self._base_objs) if i % 2 == 0]))
return self._real_dummy[-1]
if str == "imag":
self._imag_dummy.append(Dummy([obj for i, obj in enumerate(self._base_objs) if i % 2 == 1]))
return self._imag_dummy[-1]
[docs] def get(self, identifier="real", all=False):
"""
Return the first sample of the current buffer as a float.
Can be used to convert audio stream to usable Python data.
"real" or "imag" must be given to `identifier` to specify
which stream to get value from.
:Args:
identifier: string {"real", "imag"}
Address string parameter identifying audio stream.
Defaults to "mag".
all: boolean, optional
If True, the first value of each object's stream
will be returned as a list. Otherwise, only the value
of the first object's stream will be returned as a float.
Defaults to False.
"""
if not all:
return self.__getitem__(identifier)[0]._getStream().getValue()
else:
return [obj._getStream().getValue() for obj in self.__getitem__(identifier).getBaseObjects()]
[docs] def setInMag(self, x, fadetime=0.05):
"""
Replace the `inmag` attribute.
:Args:
x: PyoObject
New signal to process.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inmag = x
self._in_fader.setInput(x, fadetime)
[docs] def setInAng(self, x, fadetime=0.05):
"""
Replace the `inang` attribute.
:Args:
x: PyoObject
New signal to process.
fadetime: float, optional
Crossfade time between old and new input. Default to 0.05.
"""
pyoArgsAssert(self, "oN", x, fadetime)
self._inang = x
self._in_fader2.setInput(x, fadetime)
@property
def inmag(self):
"""PyoObject. Magnitude input signal."""
return self._inmag
@inmag.setter
def inmag(self, x):
self.setInMag(x)
@property
def inang(self):
"""PyoObject. Angle input signal."""
return self._inang
@inang.setter
def inang(self, x):
self.setInAng(x)
[docs]class FrameDelta(PyoObject):
"""
Computes the phase differences between successive frames.
The difference between the phase values of successive FFT frames for a given bin
determines the exact frequency of the energy centered in that bin. This is often
known as the phase difference (and sometimes also referred to as phase derivative
or instantaneous frequency if it's been subjected to a few additional calculations).
In order to reconstruct a plausible playback of re-ordered FFT frames, we need to
calculate the phase difference between successive frames and use it to construct a
`running phase` (by simply summing the successive differences with FrameAccum) for
the output FFT frames.
:Parent: :py:class:`PyoObject`
:Args:
input: PyoObject
Phase input signal, usually from an FFT analysis.
framesize: int, optional
Frame size in samples. Usually the same as the FFT size.
Defaults to 1024.
overlaps: int, optional
Number of overlaps in incomming signal. Usually the same
as the FFT overlaps. Defaults to 4.
.. note::
FrameDelta has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
>>> s = Server().boot()
>>> s.start()
>>> snd = SNDS_PATH + '/transparent.aif'
>>> size, hop = 1024, 256
>>> nframes = int(sndinfo(snd)[0] / size) + 1
>>> a = SfPlayer(snd, mul=.3)
>>> m_mag = [NewMatrix(width=size, height=nframes) for i in range(4)]
>>> m_pha = [NewMatrix(width=size, height=nframes) for i in range(4)]
>>> fin = FFT(a, size=size, overlaps=4)
>>> pol = CarToPol(fin["real"], fin["imag"])
>>> delta = FrameDelta(pol["ang"], framesize=size, overlaps=4)
>>> m_mag_rec = MatrixRec(pol["mag"], m_mag, 0, [i*hop for i in range(4)]).play()
>>> m_pha_rec = MatrixRec(delta, m_pha, 0, [i*hop for i in range(4)]).play()
>>> m_mag_read = MatrixPointer(m_mag, fin["bin"]/size, Sine(freq=0.25, mul=.5, add=.5))
>>> m_pha_read = MatrixPointer(m_pha, fin["bin"]/size, Sine(freq=0.25, mul=.5, add=.5))
>>> accum = FrameAccum(m_pha_read, framesize=size, overlaps=4)
>>> car = PolToCar(m_mag_read, accum)
>>> fout = IFFT(car["real"], car["imag"], size=size, overlaps=4).mix(1).out()
>>> right = Delay(fout, delay=0.013).out(1)
"""
def __init__(self, input, framesize=1024, overlaps=4, mul=1, add=0):
pyoArgsAssert(self, "oiiOO", input, framesize, overlaps, mul, add)
PyoObject.__init__(self, mul, add)
self._input = input
self._framesize = framesize
self._overlaps = overlaps
self._in_fader = InputFader(input)
in_fader, framesize, overlaps, mul, add, lmax = convertArgsToLists(
self._in_fader, framesize, overlaps, mul, add
)
num_of_mains = len(self._in_fader) // self._overlaps
self._base_players = []
for j in range(num_of_mains):
objs_list = []
for i in range(len(self._in_fader)):
if (i % num_of_mains) == j:
objs_list.append(self._in_fader[i])
self._base_players.append(FrameDeltaMain_base(objs_list, wrap(framesize, j), wrap(overlaps, j)))
self._base_objs = []
for i in range(lmax):
base_player = i % num_of_mains
overlap = i // num_of_mains
self._base_objs.append(
FrameDelta_base(self._base_players[base_player], overlap, wrap(mul, i), wrap(add, i))
)
self._init_play()
[docs] def out(self, chnl=0, inc=1, dur=0, delay=0):
return self.play(dur, delay)
[docs] def setFrameSize(self, x):
"""
Replace the `framesize` attribute.
:Args:
x: int
new `framesize` attribute.
"""
pyoArgsAssert(self, "i", x)
self._framesize = x
x, lmax = convertArgsToLists(x)
[obj.setFrameSize(wrap(x, i)) for i, obj in enumerate(self._base_players)]
@property
def input(self):
"""PyoObject. Phase input signal."""
return self._input
@input.setter
def input(self, x):
self.setInput(x)
@property
def framesize(self):
"""PyoObject. Frame size in samples."""
return self._framesize
@framesize.setter
def framesize(self, x):
self.setFrameSize(x)
[docs]class FrameAccum(PyoObject):
"""
Accumulates the phase differences between successive frames.
The difference between the phase values of successive FFT frames for a given bin
determines the exact frequency of the energy centered in that bin. This is often
known as the phase difference (and sometimes also referred to as phase derivative
or instantaneous frequency if it's been subjected to a few additional calculations).
In order to reconstruct a plausible playback of re-ordered FFT frames, we need to
calculate the phase difference between successive frames, with FrameDelta, and use
it to construct a `running phase` (by simply summing the successive differences) for
the output FFT frames.
:Parent: :py:class:`PyoObject`
:Args:
input: PyoObject
Phase input signal.
framesize: int, optional
Frame size in samples. Usually same as the FFT size.
Defaults to 1024.
overlaps: int, optional
Number of overlaps in incomming signal. Usually the same
as the FFT overlaps. Defaults to 4.
.. note::
FrameAccum has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
>>> s = Server().boot()
>>> s.start()
>>> snd = SNDS_PATH + '/transparent.aif'
>>> size, hop = 1024, 256
>>> nframes = int(sndinfo(snd)[0] / size) + 1
>>> a = SfPlayer(snd, mul=.3)
>>> m_mag = [NewMatrix(width=size, height=nframes) for i in range(4)]
>>> m_pha = [NewMatrix(width=size, height=nframes) for i in range(4)]
>>> fin = FFT(a, size=size, overlaps=4)
>>> pol = CarToPol(fin["real"], fin["imag"])
>>> delta = FrameDelta(pol["ang"], framesize=size, overlaps=4)
>>> m_mag_rec = MatrixRec(pol["mag"], m_mag, 0, [i*hop for i in range(4)]).play()
>>> m_pha_rec = MatrixRec(delta, m_pha, 0, [i*hop for i in range(4)]).play()
>>> m_mag_read = MatrixPointer(m_mag, fin["bin"]/size, Sine(freq=0.25, mul=.5, add=.5))
>>> m_pha_read = MatrixPointer(m_pha, fin["bin"]/size, Sine(freq=0.25, mul=.5, add=.5))
>>> accum = FrameAccum(m_pha_read, framesize=size, overlaps=4)
>>> car = PolToCar(m_mag_read, accum)
>>> fout = IFFT(car["real"], car["imag"], size=size, overlaps=4).mix(1).out()
>>> right = Delay(fout, delay=0.013).out(1)
"""
def __init__(self, input, framesize=1024, overlaps=4, mul=1, add=0):
pyoArgsAssert(self, "oiiOO", input, framesize, overlaps, mul, add)
PyoObject.__init__(self, mul, add)
self._input = input
self._framesize = framesize
self._overlaps = overlaps
self._in_fader = InputFader(input)
in_fader, framesize, overlaps, mul, add, lmax = convertArgsToLists(
self._in_fader, framesize, overlaps, mul, add
)
num_of_mains = len(self._in_fader) // self._overlaps
self._base_players = []
for j in range(num_of_mains):
objs_list = []
for i in range(len(self._in_fader)):
if (i % num_of_mains) == j:
objs_list.append(self._in_fader[i])
self._base_players.append(FrameAccumMain_base(objs_list, wrap(framesize, j), wrap(overlaps, j)))
self._base_objs = []
for i in range(lmax):
base_player = i % num_of_mains
overlap = i // num_of_mains
self._base_objs.append(
FrameAccum_base(self._base_players[base_player], overlap, wrap(mul, i), wrap(add, i))
)
self._init_play()
[docs] def out(self, chnl=0, inc=1, dur=0, delay=0):
return self.play(dur, delay)
[docs] def setFrameSize(self, x):
"""
Replace the `framesize` attribute.
:Args:
x: int
new `framesize` attribute.
"""
pyoArgsAssert(self, "i", x)
self._framesize = x
x, lmax = convertArgsToLists(x)
[obj.setFrameSize(wrap(x, i)) for i, obj in enumerate(self._base_players)]
@property
def input(self):
"""PyoObject. Phase input signal."""
return self._input
@input.setter
def input(self, x):
self.setInput(x)
@property
def framesize(self):
"""PyoObject. Frame size in samples."""
return self._framesize
@framesize.setter
def framesize(self, x):
self.setFrameSize(x)
[docs]class Vectral(PyoObject):
"""
Performs magnitude smoothing between successive frames.
Vectral applies filter with different coefficients for increasing
and decreasing magnitude vectors, bin by bin.
:Parent: :py:class:`PyoObject`
:Args:
input: PyoObject
Magnitude input signal, usually from an FFT analysis.
framesize: int, optional
Frame size in samples. Usually the same as the FFT size.
Defaults to 1024.
overlaps: int, optional
Number of overlaps in incomming signal. Usually the same
as the FFT overlaps. Defaults to 4.
up: float or PyoObject, optional
Filter coefficient for increasing bins, between 0 and 1.
Lower values results in a longer ramp time for bin magnitude.
Defaults to 1.
down: float or PyoObject, optional
Filter coefficient for decreasing bins, between 0 and 1.
Lower values results in a longer decay time for bin magnitude.
Defaults to 0.7
damp: float or PyoObject, optional
High frequencies damping factor, between 0 and 1. Lower values
mean more damping. Defaults to 0.9.
.. note::
Vectral has no `out` method. Signal must be converted back to time domain,
with IFFT, before being sent to output.
>>> s = Server().boot()
>>> snd = SNDS_PATH + '/accord.aif'
>>> size, olaps = 1024, 4
>>> snd = SfPlayer(snd, speed=[.75,.8], loop=True, mul=.3)
>>> fin = FFT(snd, size=size, overlaps=olaps)
>>> pol = CarToPol(fin["real"], fin["imag"])
>>> vec = Vectral(pol["mag"], framesize=size, overlaps=olaps, down=.2, damp=.6)
>>> car = PolToCar(vec, pol["ang"])
>>> fout = IFFT(car["real"], car["imag"], size=size, overlaps=olaps).mix(2).out()
>>> s.start()
"""
def __init__(self, input, framesize=1024, overlaps=4, up=1.0, down=0.7, damp=0.9, mul=1, add=0):
pyoArgsAssert(self, "oiiOOOOO", input, framesize, overlaps, up, down, damp, mul, add)
PyoObject.__init__(self, mul, add)
self._input = input
self._framesize = framesize
self._overlaps = overlaps
self._up = up
self._down = down
self._damp = damp
self._in_fader = InputFader(input)
in_fader, framesize, overlaps, up, down, damp, mul, add, lmax = convertArgsToLists(
self._in_fader, framesize, overlaps, up, down, damp, mul, add
)
num_of_mains = len(self._in_fader) // self._overlaps
self._base_players = []
for j in range(num_of_mains):
objs_list = []
for i in range(len(self._in_fader)):
if (i % num_of_mains) == j:
objs_list.append(self._in_fader[i])
self._base_players.append(
VectralMain_base(
objs_list, wrap(framesize, j), wrap(overlaps, j), wrap(up, j), wrap(down, j), wrap(damp, j)
)
)
self._base_objs = []
for i in range(lmax):
base_player = i % num_of_mains
overlap = i // num_of_mains
self._base_objs.append(Vectral_base(self._base_players[base_player], overlap, wrap(mul, i), wrap(add, i)))
self._init_play()
[docs] def out(self, chnl=0, inc=1, dur=0, delay=0):
return self.play(dur, delay)
[docs] def setFrameSize(self, x):
"""
Replace the `framesize` attribute.
:Args:
x: int
new `framesize` attribute.
"""
pyoArgsAssert(self, "i", x)
self._framesize = x
x, lmax = convertArgsToLists(x)
[obj.setFrameSize(wrap(x, i)) for i, obj in enumerate(self._base_players)]
[docs] def setUp(self, x):
"""
Replace the `up` attribute.
:Args:
x: float or PyoObject
new `up` attribute.
"""
pyoArgsAssert(self, "O", x)
self._up = x
x, lmax = convertArgsToLists(x)
[obj.setUp(wrap(x, i)) for i, obj in enumerate(self._base_players)]
[docs] def setDown(self, x):
"""
Replace the `down` attribute.
:Args:
x: float or PyoObject
new `down` attribute.
"""
pyoArgsAssert(self, "O", x)
self._down = x
x, lmax = convertArgsToLists(x)
[obj.setDown(wrap(x, i)) for i, obj in enumerate(self._base_players)]
[docs] def setDamp(self, x):
"""
Replace the `damp` attribute.
:Args:
x: float or PyoObject
new `damp` attribute.
"""
pyoArgsAssert(self, "O", x)
self._damp = x
x, lmax = convertArgsToLists(x)
[obj.setDamp(wrap(x, i)) for i, obj in enumerate(self._base_players)]
[docs] def ctrl(self, map_list=None, title=None, wxnoserver=False):
self._map_list = [
SLMap(0.0, 1.0, "lin", "up", self._up),
SLMap(0.0, 1.0, "lin", "down", self._down),
SLMap(0.0, 1.0, "lin", "damp", self._damp),
SLMapMul(self._mul),
]
PyoObject.ctrl(self, map_list, title, wxnoserver)
@property
def input(self):
"""PyoObject. Magnitude input signal."""
return self._input
@input.setter
def input(self, x):
self.setInput(x)
@property
def framesize(self):
"""int. Frame size in samples."""
return self._framesize
@framesize.setter
def framesize(self, x):
self.setFrameSize(x)
@property
def up(self):
"""float or PyoObject. Filter coefficient for increasing bins."""
return self._up
@up.setter
def up(self, x):
self.setUp(x)
@property
def down(self):
"""float or PyoObject. Filter coefficient for decreasing bins."""
return self._down
@down.setter
def down(self, x):
self.setDown(x)
@property
def damp(self):
"""float or PyoObject. High frequencies damping factor."""
return self._damp
@damp.setter
def damp(self, x):
self.setDamp(x)
[docs]class CvlVerb(PyoObject):
"""
Convolution based reverb.
CvlVerb implements convolution based on a uniformly partitioned overlap-save
algorithm. This object can be used to convolve an input signal with an
impulse response soundfile to simulate real acoustic spaces.
:Parent: :py:class:`PyoObject`
:Args:
input: PyoObject
Input signal to process.
impulse: string, optional
Path to the impulse response soundfile. The file must have the same
sampling rate as the server to get the proper convolution. Available at
initialization time only. Defaults to 'IRMediumHallStereo.wav', located
in pyo SNDS_PATH folder.
size: int {pow-of-two}, optional
The size in samples of each partition of the impulse file. Small size means
smaller latency but more computation time. If not a power-of-2, the object
will find the next power-of-2 greater and use that as the actual partition size.
This value must also be greater or equal than the server's buffer size.
Available at initialization time only. Defaults to 1024.
bal: float or PyoObject, optional
Balance between wet and dry signal, between 0 and 1. 0 means no
reverb. Defaults to 0.25.
.. seealso::
:py:class:`WGVerb`, :py:class:`STRev`, :py:class:`Freeverb`
>>> s = Server().boot()
>>> s.start()
>>> sf = SfPlayer(SNDS_PATH+"/transparent.aif", loop=True, mul=0.5)
>>> cv = CvlVerb(sf, SNDS_PATH+"/IRMediumHallStereo.wav", size=1024, bal=0.4).out()
"""
def __init__(self, input, impulse=SNDS_PATH + "/IRMediumHallStereo.wav", bal=0.25, size=1024, mul=1, add=0):
pyoArgsAssert(self, "osOiOO", input, impulse, bal, size, mul, add)
PyoObject.__init__(self, mul, add)
self._input = input
self._impulse = impulse
self._bal = bal
self._size = size
self._in_fader = InputFader(input)
in_fader, bal, size, mul, add, lmax = convertArgsToLists(self._in_fader, bal, size, mul, add)
impulse, lmax2 = convertArgsToLists(impulse)
self._base_objs = []
for file in impulse:
_size, _dur, _snd_sr, _snd_chnls, _format, _type = sndinfo(file, raise_on_failure=True)
lmax3 = max(lmax, _snd_chnls)
self._base_objs.extend(
[
CvlVerb_base(
wrap(in_fader, i),
stringencode(file),
wrap(bal, i),
wrap(size, i),
i % _snd_chnls,
wrap(mul, i),
wrap(add, i),
)
for i in range(lmax3)
]
)
self._init_play()
[docs] def setBal(self, x):
"""
Replace the `bal` attribute.
:Args:
x: float or PyoObject
new `bal` attribute.
"""
pyoArgsAssert(self, "O", x)
self._bal = x
x, lmax = convertArgsToLists(x)
[obj.setBal(wrap(x, i)) for i, obj in enumerate(self._base_objs)]
[docs] def ctrl(self, map_list=None, title=None, wxnoserver=False):
self._map_list = [SLMap(0.0, 1.0, "lin", "bal", self._bal), SLMapMul(self._mul)]
PyoObject.ctrl(self, map_list, title, wxnoserver)
@property
def input(self):
"""PyoObject. Input signal to process."""
return self._input
@input.setter
def input(self, x):
self.setInput(x)
@property
def bal(self):
"""float or PyoObject. Wet / dry balance."""
return self._bal
@bal.setter
def bal(self, x):
self.setBal(x)
[docs]class IFFTMatrix(PyoObject):
"""
Inverse Fast Fourier Transform with a PyoMatrixObject as input.
IFFTMatrix takes a matrix as input and read it as it is a sonogram.
On the current column, given by the `index` argument, the cells
at the bottom represent the lower frequencies of the spectrum and
the cells at the top, the higher frequencies of the spectrum.
Because a matrix is usually used to store bipolar signals (with the
amplitude between -1 and 1), a cell value of -1 represent a frequency
bin with no amplitude and a cell value of 1 represents the maximum
amplitude for the given frequency bin.
The instantaneous angle value (in polar coordinates) of each frequency
bin is given by the current sample in the audio signal given to the
`phase` argument. Generally speaking, the more noisy this signal is,
the more energy a bin with a positive amplitude value will have.
:Parent: :py:class:`PyoObject`
:Args:
matrix: PyoMatrixObject
The matrix used like a sonogram.
index: PyoObject
Normalized horizontal position in the matrix. 0 is the
first column and 1 is the last. Positions between two
columns are interpolated. If this signal is a Phasor,
the matrix is read from left to right.
phase: PyoObject
Instantaneous angle value used to compute the inverse
FFT. Try different signals like white noise or an oscillator
with a frequency slightly detuned in relation to the
frequency of the FFT (sr / fftsize).
size: int {pow-of-two > 4}, optional
FFT size. Must be a power of two greater than 4.
The FFT size is the number of samples used in each
analysis frame. This value must match the `size`
attribute of the former FFT object. Defaults to 1024.
overlaps: int, optional
The number of overlaped analysis block. Must be a
positive integer. More overlaps can greatly improved
sound quality synthesis but it is also more CPU
expensive. This value must match the `overlaps`
atribute of the former FFT object. Defaults to 4.
wintype: int, optional
Shape of the envelope used to filter each output frame.
Possible shapes are :
0. rectangular (no windowing)
1. Hamming
2. Hanning
3. Bartlett (triangular)
4. Blackman 3-term
5. Blackman-Harris 4-term
6. Blackman-Harris 7-term
7. Tuckey (alpha = 0.66)
8. Sine (half-sine window)
.. note::
The output of IFFTMatrix must be mixed to reconstruct the real
signal from the overlapped streams. It is left to the user
to call the mix(number of channels) method on an IFFTMatrix object.
>>> s = Server().boot()
>>> s.start()
>>> m = NewMatrix(512, 512)
>>> m.genSineTerrain(1, 0.15)
>>> index = Phasor([0.4, 0.5])
>>> phase = Noise(0.7)
>>> fout = IFFTMatrix(m, index, phase, size=2048, overlaps=16, wintype=2).mix(2).out()
"""
def __init__(self, matrix, index, phase, size=1024, overlaps=4, wintype=2, mul=1, add=0):
pyoArgsAssert(self, "mooiIiOO", matrix, index, phase, size, overlaps, wintype, mul, add)
PyoObject.__init__(self, mul, add)
self._matrix = matrix
self._index = index
self._phase = phase
self._size = size
self._overlaps = overlaps
self._wintype = wintype
matrix, index, phase, size, wintype, mul, add, self._lmax = convertArgsToLists(
matrix, index, phase, size, wintype, mul, add
)
self._base_objs = []
for j in range(overlaps):
for i in range(self._lmax):
hopsize = int(wrap(size, i) / overlaps) * j
self._base_objs.append(
IFFTMatrix_base(
wrap(matrix, i),
wrap(index, i),
wrap(phase, i),
wrap(size, i),
hopsize,
wrap(wintype, i),
wrap(mul, i),
wrap(add, i),
)
)
self._init_play()
def __len__(self):
return int(len(self._base_objs) / self._overlaps)
[docs] def setIndex(self, x):
"""
Replace the `index` attribute.
:Args:
x: PyoObject
new `index` attribute.
"""
pyoArgsAssert(self, "o", x)
self._index = x
x, lmax = convertArgsToLists(x)
for j in range(overlaps):
for i in range(self._lmax):
self._base_objs[j * self._overlaps + i].setIndex(wrap(x, i))
[docs] def setPhase(self, x):
"""
Replace the `phase` attribute.
:Args:
x: PyoObject
new `phase` attribute.
"""
pyoArgsAssert(self, "o", x)
self._phase = x
x, lmax = convertArgsToLists(x)
for j in range(overlaps):
for i in range(self._lmax):
self._base_objs[j * self._overlaps + i].setPhase(wrap(x, i))
[docs] def setSize(self, x):
"""
Replace the `size` attribute.
:Args:
x: int
new `size` attribute.
"""
pyoArgsAssert(self, "i", x)
self._size = x
x, lmax = convertArgsToLists(x)
for j in range(overlaps):
for i in range(self._lmax):
hopsize = int(wrap(x, i) / self._overlaps) * j
self._base_objs[j * self._overlaps + i].setSize(wrap(x, i), hopsize)
[docs] def setWinType(self, x):
"""
Replace the `wintype` attribute.
:Args:
x: int
new `wintype` attribute.
"""
pyoArgsAssert(self, "i", x)
self._wintype = x
x, lmax = convertArgsToLists(x)
for j in range(overlaps):
for i in range(self._lmax):
self._base_objs[j * self._overlaps + i].setWinType(wrap(x, i))
[docs] def ctrl(self, map_list=None, title=None, wxnoserver=False):
self._map_list = [SLMapMul(self._mul)]
PyoObject.ctrl(self, map_list, title, wxnoserver)
@property
def index(self):
"""PyoObject. Normalized horizontal position."""
return self._index
@index.setter
def index(self, x):
self.setIndex(x)
@property
def phase(self):
"""PyoObject. Instantaneous bin angle value."""
return self._phase
@phase.setter
def phase(self, x):
self.setPhase(x)
@property
def size(self):
"""int. FFT size."""
return self._size
@size.setter
def size(self, x):
self.setSize(x)
@property
def wintype(self):
"""int. Windowing method."""
return self._wintype
@wintype.setter
def wintype(self, x):
self.setWinType(x)