Re: [mpg123-users] Calling MPG123 from Asterisk, version behavior changed
Brought to you by:
sobukus
From: Thomas O. <tho...@or...> - 2021-09-11 12:47:05
|
Am Sat, 11 Sep 2021 14:38:24 +0200 schrieb Thomas Orgis <tho...@or...>: > At the same time, I am looking into whether it might be good to revert > to the old behaviour of s16 stereo as default format for writing to > files. 1.25.0 still worked that way. Since stable distros are slower to > picking up new versions, some more people could still be surprised. The change is the enabling of the new resampler by default in 1.26. It works with floating point numbers instead dropping/repeating samples in the decoder. So mpg123 […] --resample fine > out.raw will produce f32 output and mpg123 […] --resample ntom > out.raw will produce s16 output like before (with horrible but cheap resampling quality, which you might not care about for your use case). I think it is surprising that the choice of sampling rate — worse: resampling method — suddenly changes the encoding, although it is technically reasonable not to waste time on downconversion to s16 unless forced to do so. I'll see to make a new release that fixes that. Perhaps also will provide some backport patch to push into distros like Debian … such errors bug me. In any case: It makes sense not to rely on this default encoding, so use -e s16 if you so desire. Alrighty then, Thomas |