https://github.com/zeroby0/mdctn

Multidimensional Modified Discrete Cosine Transforms in Python
https://github.com/zeroby0/mdctn

dct lapped-orthogonal-transform mdct signal-processing

Last synced: about 2 months ago
JSON representation

Multidimensional Modified Discrete Cosine Transforms in Python

• Host: GitHub
• URL: https://github.com/zeroby0/mdctn
• Owner: zeroby0
• Created: 2022-12-09T08:16:32.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2022-12-14T06:59:47.000Z (about 2 years ago)
• Last Synced: 2024-10-10T21:21:05.456Z (3 months ago)
• Topics: dct, lapped-orthogonal-transform, mdct, signal-processing
• Language: Python
• Homepage:
• Size: 30.3 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# MDCTN :yarn:

Multidimensional [Modified Discrete Cosine Transforms](https://en.wikipedia.org/wiki/Modified_discrete_cosine_transform)

```bash

pip install mdctn

```

- [x] 1-D MDCT & IMDCT

- [ ] n-D MDCT & IMDCT

- [ ] Windowing support

- [x] Helper functions for signals

### 1-D MDCT on signals

Signals are [wrapped around](https://github.com/zeroby0/mdctn/discussions/1)

so all the data is stored in the same number of bits.

``` python

import numpy as np

from mdctn import mdct, imdct

x = np.arange(24)

y =  mdct(x, N=16)

z = imdct(y, N=16)

np.allclose(x, z) # True

```

### 1-D Pure MDCT

The core MDCT function

``` python

import numpy as np

from mdctn import core

x = np.arange(6) # [0, 1, 2, 3, 4, 5]

y_1 = core.mdct(x[0:4]) # [-2.50104055, -0.49476881]

y_2 = core.mdct(x[2:6]) # [-4.34879961, -1.26013568]

z_1 = core.imdct(y_1) # [-0.5,  0.5,  2.5,  2.5]

z_2 = core.imdct(y_2) #             [-0.5,  0.5,  4.5,  4.5]

z = (z_1[2:4] + z_2[0:2]) # [2.0, 3.0]

```

### Benchmarks

See [benchmarks.ipynb](./benchmarks.ipynb)