Hmm… I think I might just update the OP with soxr-vhq… it works at least as well as speex-float-10 with lower latency to boot.
2 Likes
1 Like
OK, I’m confused now. The results at the bottom of that page seem to show that for most situations soxr-vhq uses less CPU but with a higher latency than speex-float-10. 
I’m running 24-bit at 48k with soxr-vhq and can report it sounds much better. My CPU is a 4th gen i3 and it now sounds as good as my Mac laptop. Very happy! 
For what it’s worth, I was a professional sound engineer for 26 years. Retired now and with some hearing damage and tinnitus but I can still (thankfully) hear the difference. Low level details and timing are much improved as well as low-end frequencies. I recommend reading the articles @sothis6881 mentioned and setting up your system with what suits it. Results may vary depending on your hardware. My listening was done using tracks with FLAC encoding using DeaDBeef.
If you want to go better then using an external DAC yields further improvements but that is extra expense of course. Happy tweaking!
By the way, my CPU is not running much harder than before, negligible difference.
5 Likes
Makes sense though (depending on your audio-interface too obviously), but in general:
More CPU = less latency, which is good, so you gotta keep the balance for your needs, if you just listen - you shouldn’t really care of latency, if it’s not some astronomical values like 2 seconds or something 
3 Likes
What was a DAC again?
Here my PC is connected to my amplifier which then play’s the sounds over a expensive 5.1 surround set. Floorstanding speakers with a big center speaker and 2 bookshelf speakers.
Maybe I should also get my hand on some FLAC files to hear the difference. I do remember using the speakers for the first time and immediately noticed how crappy YouTube audio sounded.
Digital to Analog Converter.
A chip on your motherboard that feeds the sound to your AMP in this case.
2 Likes
Yeah, a standalone dedicated DAC with it’s own power supply will sound better than the chip on your motherboard. Having said that some modern PCs are sounding pretty good. I have an old Desktop PC with a Core2 Duo and that sounds crap with whatever I do to it! My newer machines sound acceptable to me though, as a casual listening source, not pro. If was looking at pro stuff for home use I’d need to win the lottery!
1 Like
I’m confused as well. I might have mixed up soxr and speex… 
It would make sense that speex is tuned for low-latency if it’s intended for speech - you’d want to minimise delay if you’re talking to someone else.
At the moment, though, I’m just listening to audio and soxr is working just fine.
1 Like
For those that wish to look at some technical aspects:
https://lastique.github.io/src_test/
The sampling rate using soxr is much higher than speex. That would explain the improvement in resolution eg picking up fine details.
For those that don’t relate to technical specs, just use your ears!
By sampling rate I mean the amount of sampling done, not the bitrate (edited to make sense)
1 Like
WOW…what a great little tweak. I use PulseEffects also for an equalizer. ‘killall pulseaudio’ killed pulseeffects, so you have to reboot after applying the change to get it working again.
You should start a ‘Stupid Simple Tweaks’ thread with more helpful hints… I would tune in. 
2 Likes
A script added to login would set everything up on each reboot. I’m only just learning about script writing though so I can’t do this. If anyone has any ideas about a basic script that can be modified per user that would be fantastic.
Same here after doing some testing. 24-bit sounded fuller and clearer on my system than 16- or 32-bit.
Not sure about pulseeffects, but pulseaudio -k should do the trick, then you may also need to run pulseaudio --start
1 Like
Yeah, good news. I’m not running puelseeffects either. I think the modified settings on pulse are enough for my needs. Regarding music and movie playback DeaDBeef, Strawberry and SMplayer all have their own EQ capabilities. For me that is enough, but others may wish to make a system-wide solution. I prefer to keep the basic system as simple as possible and just use the core components. That way there is less to rely upon and less to go wrong. After all, who needs a calculator or a weather forecast for audio quality? I’ve got other apps for that! K.I.S.S principle. 
2 Likes
I missed the point, why do you need to do that on each boot? 
It’s one-time job to write config file to your home dir 
3 Likes
Yeah, point taken. It could be done either way as a config or autostart. It achieves the same thing but your suggestion is a neater way of doing it unless you want it for multiple users and root blah, blah. I suppose doing that would involve digging into systemd perhaps? Probably not such a good idea. Messing around with core system files gets…well…messy!
1 Like
Of course!
@sothis6881 Thanks for those links!
@jonathon and you solved a problem that has been bugging me for years.
Finally, decent sound in Linux!
1 Like