Hugging Face's logo

hanamizuki-ai
/
pypi-wheels

Model card Files Community
Dividend9853 commited on
Commit
a056889
•
1 Parent(s): 8cf8037

Upload 5 files

Browse files
Files changed (5) hide show
  1. lws/index.html +24 -0
  2. lws/lws-1.2.8-cp310-cp310-win_amd64.whl +0 -0
  3. lws/lws-1.2.8-cp310-cp310-win_amd64.whl.metadata +174 -0
  4. lws/lws-1.2.8-cp311-cp311-win_amd64.whl +0 -0
  5. lws/lws-1.2.8-cp311-cp311-win_amd64.whl.metadata +174 -0
lws/index.html CHANGED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <!DOCTYPE html>
2
+ <html lang="en">
3
+ <head>
4
+ <meta name="generator" content="simple503 version 0.4.0" />
5
+ <meta name="pypi:repository-version" content="1.0" />
6
+ <meta charset="UTF-8" />
7
+ <title>
8
+ Links for lws
9
+ </title>
10
+ </head>
11
+ <body>
12
+ <h1>
13
+ Links for lws
14
+ </h1>
15
+ <a href="https://huggingface.co/hanamizuki-ai/pypi-wheels/resolve/main/lws/lws-1.2.8-cp311-cp311-win_amd64.whl#sha256=50ef93cf3ebc4005bbb3f7c6f1838a8a139a1ac553c4ae8d8b6c10f8e17060c3" data-dist-info-metadata="sha256=2f2413fad3f61684da1821265c3a376b4b150ad264127b8a121d9e93d051a4db">
16
+ lws-1.2.8-cp311-cp311-win_amd64.whl
17
+ </a>
18
+ <br />
19
+ <a href="https://huggingface.co/hanamizuki-ai/pypi-wheels/resolve/main/lws/lws-1.2.8-cp310-cp310-win_amd64.whl#sha256=e8fb971799e059dc2f76e083baf3aed6fb8df6848231af140b0561b8ad6f3b16" data-dist-info-metadata="sha256=2f2413fad3f61684da1821265c3a376b4b150ad264127b8a121d9e93d051a4db">
20
+ lws-1.2.8-cp310-cp310-win_amd64.whl
21
+ </a>
22
+ <br />
23
+ </body>
24
+ </html>
lws/lws-1.2.8-cp310-cp310-win_amd64.whl ADDED
Binary file (125 kB). View file
 
lws/lws-1.2.8-cp310-cp310-win_amd64.whl.metadata ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Metadata-Version: 2.1
2
+ Name: lws
3
+ Version: 1.2.8
4
+ Summary: Fast spectrogram phase reconstruction using Local Weighted Sums
5
+ Home-page: https://github.com/Jonathan-LeRoux/lws
6
+ Download-URL: https://github.com/Jonathan-LeRoux/lws/archive/1.2.8.tar.gz
7
+ Author: Jonathan Le Roux
8
+ Author-email: [email protected]
9
+ License: Apache 2.0
10
+ Keywords: phase,reconstruction,stft,short-term Fourier Transform,spectrogram
11
+ Classifier: Programming Language :: Python
12
+ Classifier: Intended Audience :: Developers
13
+ Classifier: Topic :: Multimedia :: Sound/Audio :: Analysis
14
+ Classifier: Programming Language :: Python :: 2
15
+ Classifier: Programming Language :: Python :: 2.7
16
+ Classifier: Programming Language :: Python :: 3
17
+ Classifier: Programming Language :: Python :: 3.5
18
+ Classifier: Programming Language :: Python :: 3.6
19
+ Classifier: Programming Language :: Python :: 3.7
20
+ Classifier: Programming Language :: Python :: 3.8
21
+ Classifier: Programming Language :: Python :: 3.9
22
+ Classifier: Programming Language :: Python :: 3.10
23
+ Classifier: Programming Language :: Python :: 3.11
24
+ Classifier: Programming Language :: Python :: 3.12
25
+ Classifier: Programming Language :: Python :: 3.13
26
+ Description-Content-Type: text/markdown
27
+ Requires-Dist: numpy
28
+
29
+ LWS
30
+ ===
31
+
32
+ ### Fast spectrogram phase recovery using Local Weighted Sums ###
33
+
34
+ Author: Jonathan Le Roux -- 2008-2023
35
+
36
+ [![PyPI version](https://badge.fury.io/py/lws.svg)](https://badge.fury.io/py/lws)
37
+
38
+ LWS is a C/C++ library for which this package is a Python wrapper.
39
+ A Matlab/Mex wrapper is also available.
40
+
41
+ License
42
+ -------
43
+
44
+ Copyright (C) 2008-2023 Jonathan Le Roux
45
+ Apache 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
46
+
47
+ Citing this code
48
+ ----------------
49
+
50
+ If you use this code, please cite the following papers:
51
+
52
+ ### Batch LWS ###
53
+
54
+ Jonathan Le Roux, Hirokazu Kameoka, Nobutaka Ono, Shigeki Sagayama,
55
+ "Fast Signal Reconstruction from Magnitude STFT Spectrogram Based on Spectrogram Consistency,"
56
+ in Proc. International Conference on Digital Audio Effects (DAFx), pp. 397--403, Sep. 2010.
57
+
58
+ @InProceedings{LeRoux2010DAFx09,
59
+ author = {Jonathan {Le Roux} and Hirokazu Kameoka and Nobutaka Ono and Shigeki Sagayama},
60
+ title = {Fast Signal Reconstruction from Magnitude {STFT} Spectrogram Based on Spectrogram Consistency},
61
+ booktitle = {Proc. International Conference on Digital Audio Effects (DAFx)},
62
+ year = 2010,
63
+ pages = {397--403},
64
+ month = sep
65
+ }
66
+
67
+
68
+ ### Online LWS, "No future" LWS ###
69
+
70
+ Jonathan Le Roux, Hirokazu Kameoka, Nobutaka Ono, Shigeki Sagayama,
71
+ "Phase initialization schemes for faster spectrogram-consistency-based signal reconstruction,"
72
+ in Proc. of ASJ Autumn Meeting, 3-10-3, Sep. 2010.
73
+
74
+ @InProceedings{LeRoux2010ASJ09,
75
+ author = {Jonathan {Le Roux} and Hirokazu Kameoka and Nobutaka Ono and Shigeki Sagayama},
76
+ title = {Phase Initialization Schemes for Faster Spectrogram-Consistency-Based Signal Reconstruction},
77
+ year = 2010,
78
+ booktitle = {Proc. Acoustical Society of Japan Autumn Meeting (ASJ)},
79
+ number = {3-10-3},
80
+ month = mar
81
+ }
82
+
83
+
84
+ Installation:
85
+ -------------
86
+
87
+ 1) The easiest way to install `lws` is via `pip`:
88
+
89
+ ```sh
90
+ pip install lws
91
+ ```
92
+
93
+ 2) To compile from source using cython (required if one modifies the code):
94
+
95
+ ```sh
96
+ cd python
97
+ LWS_USE_CYTHON=1 make
98
+ ```
99
+
100
+ 3) To compile from source using the pre-generated c source file (which was obtained with cython):
101
+
102
+ ```sh
103
+ cd python
104
+ make
105
+ ```
106
+
107
+ 4) Alternatively, one can first use cython to create a tarball, which can then be installed with pip:
108
+
109
+ ```sh
110
+ cd python
111
+ make sdist
112
+ pip install dist/lws-1.2.7.tar.gz
113
+ ```
114
+
115
+ **Note:** On Windows, the Microsoft Visual C++ Compiler for your version of Python needs to be installed. See [this page](https://wiki.python.org/moin/WindowsCompilers) for more details.
116
+
117
+ Usage
118
+ -----
119
+
120
+ ```python
121
+ import lws
122
+ import numpy as np
123
+
124
+ lws_processor=lws.lws(512,128, mode="speech") # 512: window length; 128: window shift
125
+ X = lws_processor.stft(x) # where x is a single-channel waveform
126
+ X0 = np.abs(X) # Magnitude spectrogram
127
+ print('{:6}: {:5.2f} dB'.format('Abs(X)', lws_processor.get_consistency(X0)))
128
+ X1 = lws_processor.run_lws(X0) # reconstruction from magnitude (in general, one can reconstruct from an initial complex spectrogram)
129
+ print('{:6}: {:5.2f} dB'.format('LWS', lws_processor.get_consistency(X1)))
130
+ ```
131
+
132
+ Options
133
+ -------
134
+
135
+ ```python
136
+ lws_processor=lws.lws(awin_or_fsize, fshift, L = 5, swin = None, look_ahead = 3,
137
+ nofuture_iterations = 0, nofuture_alpha = 1, nofuture_beta = 0.1, nofuture_gamma = 1,
138
+ online_iterations = 0, online_alpha = 1, online_beta = 0.1, online_gamma = 1,
139
+ batch_iterations = 100, batch_alpha = 100, batch_beta = 0.1, batch_gamma = 1,
140
+ symmetric_win = True, mode= None, fftsize=None, perfectrec=True)
141
+ ```
142
+
143
+ * `awin_or_fsize`: either the analysis window, or a window length (in which case the sqrt(hann) window is used); the analysis window should be symmetric for the computations to be correct.
144
+ * `fshift`: window shift
145
+ * `L`: approximation order in the phase reconstruction algorithm, 5 should be good.
146
+ * `swin`: synthesis window (if None, it gets computed from the analysis window for perfect reconstruction)
147
+ * `look_ahead`: number of look-ahead frames in RTISI-LA-like algorithm, 3 should be good.
148
+ * `xxx_iterations`, `xxx_alpha`, `xxx_beta`, `xxx_gamma`: number of iterations of algorithm xxx (where xxx is one of `nofuture`, `online`, or `batch`), and parameters alpha/beta/gamma of the decreasing sparsity curve that is used to determine which bins get updated at each iteration. Any bin with magnitude larger than a given threshold is updated, others are ignored (`thresholds = alpha * np.exp(- beta * np.arange(iterations)**gamma)`)
149
+ * `symmetric_win`: determines whether to use a symmetric hann window or not
150
+ * `mode`: `None`, `'speech'`, or `'music'`. This sets default numbers of iterations of each algorithm that seem to be good for speech and music signals. Disclaimer: your mileage may vary.
151
+ * `fftsize`: can be set longer than frame size to do 0-padding in the FFT. Note that 0-padding will be done symmetrically on the left and right of the window to enforce symmetry in the analysis window.
152
+ * `perfectrec`: whether to pad with zeros on each side to ensure perfect reconstruction at the boundaries too.
153
+
154
+ Three steps are implemented, and they can be turned on/off independently by appropriately setting the corresponding number of iterations:
155
+
156
+ * "no future" LWS: phase initialization using LWS updates that only involve past frames
157
+ * online LWS: phase estimation using online LWS updates, corresponding to a fast time-frequency domain version of RTISI-LA
158
+ * LWS: phase estimation using batch LWS updates on the whole spectrogram
159
+
160
+
161
+
162
+
163
+ Remarks
164
+ -------
165
+
166
+ 1) The .cpp files are actually C code with some C99 style comments, but the .cpp extension is needed on Windows for mex to acknowledge the c99 flag (with .c, it is discarded, and -ansi used instead, leading to compilation errors)
167
+
168
+ 2) Because the module is a C extension, it cannot be reloaded (see <http://bugs.python.org/issue1144263>). In Jupyter Notebooks, in particular, autoreload will not work, and the kernel has to be restarted.
169
+
170
+
171
+ Acknowledgements
172
+ ----------------
173
+
174
+ The recipe to wrap the LWS C code as a python module was largely inspired by Martin Sosic's post: http://martinsosic.com/development/2016/02/08/wrapping-c-library-as-python-module.html
lws/lws-1.2.8-cp311-cp311-win_amd64.whl ADDED
Binary file (135 kB). View file
 
lws/lws-1.2.8-cp311-cp311-win_amd64.whl.metadata ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Metadata-Version: 2.1
2
+ Name: lws
3
+ Version: 1.2.8
4
+ Summary: Fast spectrogram phase reconstruction using Local Weighted Sums
5
+ Home-page: https://github.com/Jonathan-LeRoux/lws
6
+ Download-URL: https://github.com/Jonathan-LeRoux/lws/archive/1.2.8.tar.gz
7
+ Author: Jonathan Le Roux
8
+ Author-email: [email protected]
9
+ License: Apache 2.0
10
+ Keywords: phase,reconstruction,stft,short-term Fourier Transform,spectrogram
11
+ Classifier: Programming Language :: Python
12
+ Classifier: Intended Audience :: Developers
13
+ Classifier: Topic :: Multimedia :: Sound/Audio :: Analysis
14
+ Classifier: Programming Language :: Python :: 2
15
+ Classifier: Programming Language :: Python :: 2.7
16
+ Classifier: Programming Language :: Python :: 3
17
+ Classifier: Programming Language :: Python :: 3.5
18
+ Classifier: Programming Language :: Python :: 3.6
19
+ Classifier: Programming Language :: Python :: 3.7
20
+ Classifier: Programming Language :: Python :: 3.8
21
+ Classifier: Programming Language :: Python :: 3.9
22
+ Classifier: Programming Language :: Python :: 3.10
23
+ Classifier: Programming Language :: Python :: 3.11
24
+ Classifier: Programming Language :: Python :: 3.12
25
+ Classifier: Programming Language :: Python :: 3.13
26
+ Description-Content-Type: text/markdown
27
+ Requires-Dist: numpy
28
+
29
+ LWS
30
+ ===
31
+
32
+ ### Fast spectrogram phase recovery using Local Weighted Sums ###
33
+
34
+ Author: Jonathan Le Roux -- 2008-2023
35
+
36
+ [![PyPI version](https://badge.fury.io/py/lws.svg)](https://badge.fury.io/py/lws)
37
+
38
+ LWS is a C/C++ library for which this package is a Python wrapper.
39
+ A Matlab/Mex wrapper is also available.
40
+
41
+ License
42
+ -------
43
+
44
+ Copyright (C) 2008-2023 Jonathan Le Roux
45
+ Apache 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
46
+
47
+ Citing this code
48
+ ----------------
49
+
50
+ If you use this code, please cite the following papers:
51
+
52
+ ### Batch LWS ###
53
+
54
+ Jonathan Le Roux, Hirokazu Kameoka, Nobutaka Ono, Shigeki Sagayama,
55
+ "Fast Signal Reconstruction from Magnitude STFT Spectrogram Based on Spectrogram Consistency,"
56
+ in Proc. International Conference on Digital Audio Effects (DAFx), pp. 397--403, Sep. 2010.
57
+
58
+ @InProceedings{LeRoux2010DAFx09,
59
+ author = {Jonathan {Le Roux} and Hirokazu Kameoka and Nobutaka Ono and Shigeki Sagayama},
60
+ title = {Fast Signal Reconstruction from Magnitude {STFT} Spectrogram Based on Spectrogram Consistency},
61
+ booktitle = {Proc. International Conference on Digital Audio Effects (DAFx)},
62
+ year = 2010,
63
+ pages = {397--403},
64
+ month = sep
65
+ }
66
+
67
+
68
+ ### Online LWS, "No future" LWS ###
69
+
70
+ Jonathan Le Roux, Hirokazu Kameoka, Nobutaka Ono, Shigeki Sagayama,
71
+ "Phase initialization schemes for faster spectrogram-consistency-based signal reconstruction,"
72
+ in Proc. of ASJ Autumn Meeting, 3-10-3, Sep. 2010.
73
+
74
+ @InProceedings{LeRoux2010ASJ09,
75
+ author = {Jonathan {Le Roux} and Hirokazu Kameoka and Nobutaka Ono and Shigeki Sagayama},
76
+ title = {Phase Initialization Schemes for Faster Spectrogram-Consistency-Based Signal Reconstruction},
77
+ year = 2010,
78
+ booktitle = {Proc. Acoustical Society of Japan Autumn Meeting (ASJ)},
79
+ number = {3-10-3},
80
+ month = mar
81
+ }
82
+
83
+
84
+ Installation:
85
+ -------------
86
+
87
+ 1) The easiest way to install `lws` is via `pip`:
88
+
89
+ ```sh
90
+ pip install lws
91
+ ```
92
+
93
+ 2) To compile from source using cython (required if one modifies the code):
94
+
95
+ ```sh
96
+ cd python
97
+ LWS_USE_CYTHON=1 make
98
+ ```
99
+
100
+ 3) To compile from source using the pre-generated c source file (which was obtained with cython):
101
+
102
+ ```sh
103
+ cd python
104
+ make
105
+ ```
106
+
107
+ 4) Alternatively, one can first use cython to create a tarball, which can then be installed with pip:
108
+
109
+ ```sh
110
+ cd python
111
+ make sdist
112
+ pip install dist/lws-1.2.7.tar.gz
113
+ ```
114
+
115
+ **Note:** On Windows, the Microsoft Visual C++ Compiler for your version of Python needs to be installed. See [this page](https://wiki.python.org/moin/WindowsCompilers) for more details.
116
+
117
+ Usage
118
+ -----
119
+
120
+ ```python
121
+ import lws
122
+ import numpy as np
123
+
124
+ lws_processor=lws.lws(512,128, mode="speech") # 512: window length; 128: window shift
125
+ X = lws_processor.stft(x) # where x is a single-channel waveform
126
+ X0 = np.abs(X) # Magnitude spectrogram
127
+ print('{:6}: {:5.2f} dB'.format('Abs(X)', lws_processor.get_consistency(X0)))
128
+ X1 = lws_processor.run_lws(X0) # reconstruction from magnitude (in general, one can reconstruct from an initial complex spectrogram)
129
+ print('{:6}: {:5.2f} dB'.format('LWS', lws_processor.get_consistency(X1)))
130
+ ```
131
+
132
+ Options
133
+ -------
134
+
135
+ ```python
136
+ lws_processor=lws.lws(awin_or_fsize, fshift, L = 5, swin = None, look_ahead = 3,
137
+ nofuture_iterations = 0, nofuture_alpha = 1, nofuture_beta = 0.1, nofuture_gamma = 1,
138
+ online_iterations = 0, online_alpha = 1, online_beta = 0.1, online_gamma = 1,
139
+ batch_iterations = 100, batch_alpha = 100, batch_beta = 0.1, batch_gamma = 1,
140
+ symmetric_win = True, mode= None, fftsize=None, perfectrec=True)
141
+ ```
142
+
143
+ * `awin_or_fsize`: either the analysis window, or a window length (in which case the sqrt(hann) window is used); the analysis window should be symmetric for the computations to be correct.
144
+ * `fshift`: window shift
145
+ * `L`: approximation order in the phase reconstruction algorithm, 5 should be good.
146
+ * `swin`: synthesis window (if None, it gets computed from the analysis window for perfect reconstruction)
147
+ * `look_ahead`: number of look-ahead frames in RTISI-LA-like algorithm, 3 should be good.
148
+ * `xxx_iterations`, `xxx_alpha`, `xxx_beta`, `xxx_gamma`: number of iterations of algorithm xxx (where xxx is one of `nofuture`, `online`, or `batch`), and parameters alpha/beta/gamma of the decreasing sparsity curve that is used to determine which bins get updated at each iteration. Any bin with magnitude larger than a given threshold is updated, others are ignored (`thresholds = alpha * np.exp(- beta * np.arange(iterations)**gamma)`)
149
+ * `symmetric_win`: determines whether to use a symmetric hann window or not
150
+ * `mode`: `None`, `'speech'`, or `'music'`. This sets default numbers of iterations of each algorithm that seem to be good for speech and music signals. Disclaimer: your mileage may vary.
151
+ * `fftsize`: can be set longer than frame size to do 0-padding in the FFT. Note that 0-padding will be done symmetrically on the left and right of the window to enforce symmetry in the analysis window.
152
+ * `perfectrec`: whether to pad with zeros on each side to ensure perfect reconstruction at the boundaries too.
153
+
154
+ Three steps are implemented, and they can be turned on/off independently by appropriately setting the corresponding number of iterations:
155
+
156
+ * "no future" LWS: phase initialization using LWS updates that only involve past frames
157
+ * online LWS: phase estimation using online LWS updates, corresponding to a fast time-frequency domain version of RTISI-LA
158
+ * LWS: phase estimation using batch LWS updates on the whole spectrogram
159
+
160
+
161
+
162
+
163
+ Remarks
164
+ -------
165
+
166
+ 1) The .cpp files are actually C code with some C99 style comments, but the .cpp extension is needed on Windows for mex to acknowledge the c99 flag (with .c, it is discarded, and -ansi used instead, leading to compilation errors)
167
+
168
+ 2) Because the module is a C extension, it cannot be reloaded (see <http://bugs.python.org/issue1144263>). In Jupyter Notebooks, in particular, autoreload will not work, and the kernel has to be restarted.
169
+
170
+
171
+ Acknowledgements
172
+ ----------------
173
+
174
+ The recipe to wrap the LWS C code as a python module was largely inspired by Martin Sosic's post: http://martinsosic.com/development/2016/02/08/wrapping-c-library-as-python-module.html