Hello Tanu, > I did scan every patch. My criteria for inclusion were: good to have this as guidelines what should go into NEWS I am not particularly keen on having any of those listed > > e66e8464183e9916b248a9dedd4d6dcf3e1a38fa Change/fix conversion to/from float32 > > This seemed like a minor improvement, and falls partially to the bug fix > category. If you disagree about this having only a minor impact, I'll > add this item. it potentially changes the audio data going through PA; not that one should depend on a precise float32 value :) might bite people doing automated testing > > 930654a3af266ab112d1d1f6b1083197ee4bb373 Generate normalized rows in calc_map_table() > A bug fix. Perhaps generally interesting, though, so I might add this. it potentially changes the audio mixing; audio may sound differently > > 92bb9fb8b5aeebb87c4df7416e75db1782e2dd3a Switch to speex-float-1 by default > > I thought about adding a "changed defaults" section, but the file seemed > to be pretty long already... I'm not really against adding this, though. > Another changed default is that the default default.pa loads > module-jackdbus-detect now with "channels=2" argument. speex-float-1 is quite noticeable on ARM > > echo-cancel: support multi-channel EC implementation with asymmetric number of channels in record/playback/output stream specs > > Yes, this should probably be mentioned. I don't remember how this > exactly works and what are the implications, so I need to think a bit > what to write. maybe nobody cares except me; the implication is that the EC module is now more flexible and supports e.g. a 4-channels microphone signal going into the echo canceller (say for noise reduction) yet output and playback (towards the application) are single channel; no EC implementation shipping with PA needs this for the moment p. -- Peter Meerwald +43-664-2444418 (mobile)
