You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The Stack Overflow narrowed down the problem to the SOXR_NO_DITHER flag, as it indeed uses a seed that depends on time() and the random address of a variable:
The seed value is only used if the soxr output is 16 samples. In that case, some dithering process is applied. This means some noise is added to the signal to somehow mask quantization errors. Hence this is not a bug but a feature.
This being said, using a fixed seed to produce a deterministic result won't harm the result. What is essential is the apparent randomization from sample to sample.
Each run of soxr will produce different results.
In certain contexts this can be undesirable.
We stumbled on this problem today.
As far as I could research on the Internet, this problem has been brought up at least twice:
The Stack Overflow narrowed down the problem to the
SOXR_NO_DITHER
flag, as it indeed uses a seed that depends on time() and the random address of a variable:The solution is obvious & straightforward: Allow to provide the seed externally, instead of setting it to a different value on each run:
The text was updated successfully, but these errors were encountered: