Welcome to SSRC Discussions!

[shibatch]

Please feel free to write your thoughts about this project here.
If you have any feedback or requests regarding the project, please let us know.

[leonardmule]

Hello,
I'm a user of ssrc-1.33, which I've used regularly for years to convert files from 44100 Hz to 48000 Hz.
This morning, I wanted to try ssrc-2.4.0 and noticed that it introduces a silence (>100ms) at the beginning of the output file.
The parameters I use are : ssrc.exe --rate 48000 --bits 24
Please tell me what I'm doing wrong :-)

[shibatch]

That's strange. How about --profile lightning?

[leonardmule]

Much better : just 1ms is added. With or without --minPhase. Not perfect still.

[shibatch]

The 100ms silence addition is by design. If you don't need it, please remove it yourself.

[leonardmule]

Okay, thanks, but I'm going back to version 1.33 because I need the generated files to be exactly the same length without any intervention from me. By the way, is version 1.33 the last update for version 1.X?

[shibatch]

Yes

[scitrin2]

Sorry to bother you, but so you are aware, and if you have any advice....

I recently followed the instructions in (ssrc) API_DOCUMENTATION.md, section 3, which essentially fakes out (or re-defines) the code to use ssrc upsampling rather than soxr. This was for Linux program, mpd. (Subjectively there is a significant improvement of sound using ssrc.) The problem is, is that attempting to properly terminate mpd by "mpd --kill" does not terminate mpd. To terminate mpd, you need to do a " killall -9 mpd". This is on mpd version 0.24.8. If you do the modification (soxr to ssrc) to mpd 0.24.9, mpd will not run because it hits the following block in main.cxx:

#ifdef linux
if (ProcStatusThreads() > 1)
/* threads created by libraries before main() can
cause all sorts of bugs that are difficult to
analyze; bail out quickly and refuse to run if we
detect such a thing */
throw "Background threads were detected before MPD could initialize. This is likely caused by a misbehaving library. Aborting to prevent erroneous behavior. Please report.";
#endif

If you comment out this block of code in main(), the program then runs (as in 0.24.8). I suspect the culprit is line 101 in libssrc.cpp, "BGExecutorStatic bgExecutorStatic;", which apparently allocates bgExecutorStatic on the data segment and perhaps sets threads in motion before main() has a chance to execute, but I could be wrong. In any event, I haven't figure out an easy way to fix this, or whether if really matters that you have to "kill -9" the program.