Hi guys, I stumbled upon your presentation/application today. I noticed that you were mentioning issues with SoX 14.3.2. I'd just like to make sure I get this right: 1) Training was done using SoX < 14.3.2; 2) Using SoX >= 14.3.2 requires retraining to get great accuracy.
Thus, if I get this right, it's a small change in sox's behaviour that breaks your training. If my quick scan serves me right, you're using only three effects: spectrogram, trim and stat. Out of curiosity, I'd like to isolate the issue if easily possible. There were modifications to all three of them: - spectrogram: the "-r" option was added to get raw spectrogram without legend; - stats: the "-w" option was fixed; - trim: added support specifying an absolute end location for trim effect instead only an offset from trim begin location. The last change is most likely the culprit... trim has been rewritten in 14.4.0 most notably fixing backward compatibility. Which brings me to... did you have the chance to try out 14.4.0? It's available in Debian testing and unstable, and in Ubuntu Quantal. Congrats for your work and cheers, -Pascal -- Homepage (http://organact.mine.nu) Debian GNU/Linux (http://www.debian.org) COMunité/LACIME: École de technologie supérieure (http://www.comunite.ca) Integrated Microsystems Laboratory: McGill (http://www.iml.ece.mcgill.ca) ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ SoX-devel mailing list SoX-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sox-devel
