lauchmelder/spectralyze
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity

Updated readme

Browse source
This commit is contained in:
Robert 2021-04-29 00:02:05 +02:00
parent b7e7790d86
commit 35620d3395
1 changed files with 33 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
README.md
View file

@ -10,6 +10,39 @@ spectralyze -i 50 coolSong.wav
This would read in a file called `coolSong.wav`, split it into 50ms long audio segments and then transform each of those individually. The resulting spectrums will be stored in `coolSong.json`
## Setting a frequency range
By default, spectralyze will output the entire frequency spectrum, all the way up to the nyquist limit (which is half the sampling rate). If you are not interested in the entire spectrum you can tell the program the only output frequencies in the range you specify:
```
spectralyze -f 0,2500 coolSong.wav
```
This command would only output frequencies ranging from 0kHz-2kHz, greatly decreasing file size.
## Disabling channels
By default this program will analyze all channels in the given audio file, if you are only interested in noe specific channel you can tell the program that via the `-m` flag:
```
spectralyze -m 1 coolSong.wav
```
will only analyze the first audio channel
## Zero-padding
The FFT algorithm implemented here can only work if the number of samples is a power of 2. So by default, before performing the transformation, this program will zero-pad the signal until we reach such a sample size. Essentially, it appends a bunch of zeros to the end until it is a power of two. By using the `-p` flag you can go further than this. `-p 2` will tell the program to pad up until the power of two *after* the next one, essentially doubling the sample size. This results in a higher resolution in the frequency spectrum
```
spectralyze -p 3 coolSong.wav
```
This will tell the program to pad to the 3rd-next power of 2! This means, if the number if samples given is 100, it would pad it to 256 by default, and due to the `-p` switch all the way to 1024.
**RESOLUTION (and thus file size) SCALES WITH 2^p**
## Example command
```
spectralyze -i 20 -f 0,1000 -p 3 coolSong.wav
```
After the file coolSong.wav is read, the audio signal is split into 20ms long clips, which are then each individually given to the transformation function.
Due to `-p 3`, the zero padding will go two powers of two higher than it normally would, essentially quadrupling output resolution.
`-f 0,1000` will limit the outputted spectrum to a range between 0kHz and 1kHz
## Supported Formats
* WAV
* AIFF