Re: [Audacity-devel] SoX resampling
A free multi-track audio editor and recorder
Brought to you by:
aosiniao
From: Gale A. <ga...@au...> - 2012-10-15 21:31:37
|
| From Rob Sykes <aq...@ya...> | Mon, 15 Oct 2012 12:07:19 +0100 (BST) | Subject: [Audacity-devel] SoX resampling > > From: Steve the Fiddle <ste...@gm...> > > To: aud...@li... > > Cc: > > Sent: Monday, 15 October 2012, 8:49 > > Subject: Re: [Audacity-devel] SoX resampling > > > >T he thing that made the CRLF issue so difficult to track down is that > > once you have successfully built Audacity with libsoxr, it doesn't > > matter if there is CRLF, even if you delete the entire libsoxr > > directory and replace it and run make distclean. OK, it must have worked for me then, because I first modified lib-src/libsoxr/configure on Linux, so was able to build the first time. > Yes, it seems that there are a couple more Windows/Unix > differences to resolve: 1) lib-src/libsoxr/configure needs to > have the executable flag set Is chmod 755 ( -rwxr-xr-x ) what's wanted for that file? > the Windows-style <CR><LF> line-endings that have crept in to this > file must be removed. Both of these might be difficult for Vaughan to > tackle from the Windows side If on Windows, it can be done (at least) with Notepad++. > Also worth mentioning, is that the top-level 'configure' file, though > generated, must be committed prior to any release, so this will > certainly need to be done by a unix dev. Does the configure I posted look right? > Hopefully, these are the last remaining build issues, and when fixed, > will allow more folk to have a play with the resampler. The 6 qualities > presented in the current libsoxr snapshot are fine for comparison > against existing versions of Audacity (and with SoX), but I don't > think that they make sense for a release. So some thought is needed > as to how-many/which qualities are needed (and if the user should be > allowed to configure their own quality by varying parameters). The KISS > principle might suggest to offer just 4 standard SoX qualities ('quick', 'low', > 'high'similar to LSR's linear, fast sinc, medium sinc; and 'very high' > similar-ish to LSR best sinc, but not showing any spectrogram artefacts > @ float32), and only to expose more features if users request them. I'm not aware of any requests in the past for users to vary parameters. The Windows and Mac releases (libresample) now only have two quality choices so I agree six may be too many and four about right. I think it's important we have a "best" equivalent if that is now much faster than the LSR "best" was. Gale |