This package is a tool to restore sound from waveform images.
The general procedure is:
- Find the max and min points of each column of waveform pixles
- Upsample to a higher sample rate (44.1kHz by default) by 3rd-order spline interpolation given the condition that the derivative at all the max and min points is 0
- Apply a lowpass filter to reduce the frequencies that are impossible to be restored from the waveform image.
julia <path/to/the/project>/cli/readwf.jl [OPTIONS]... IMAGE_FILE R G B DURATION
positional arguments:
IMAGE_FILE Path to the waveform image file
R Red value of the waveform colour (0~255)
G Green value of the waveform colour (0~255)
B Blue value of the waveform colour (0~255)
DURATION Duration of the sound in seconds
options:
--max-colour-delta MAX-COLOUR-DELTA
Maximum colour difference allowed for a pixel to be counted
as part of the waveform. Try increasing if the waveform is
not read in whole
-o, --output OUTPUT_FILE
Path to the output audio file
-q, --quiet Turn off printing
Assuming the current path is the project home path
julia ./cli/readwf.jl ./examples/eg1.png 59 153 105 0.692
The above will generate a WAV file at the current path
Or you can specify the output file path (and the project environment):
julia --project=. ./cli/readwf.jl -o ./examples/eg1.wav ./examples/eg1.png 59 153 105 0.692
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent