Finally got round to do some listening.I'd be interested to know how you think the changes affected the sound, if at all. Positive or negative.
Tested with combinations of Pi Zero, Pi B+, generic 'Sabre' ES9023 async (50gHz) Dac, generic TI PCM5102 async Dac, PiFi dac (modified Chinese clone with TI PCM5122 chipset) and a HifiberryDac+ (with TI PCM5122 chipset).
First performance, no problems running SoXr on a Pi Zero or B+. Both have more than 50% processor headroom when MPD/SoXr are working hardest (output bitrate/depth “192000:32:2”).
Like you, most of my music is 44.1kHz, mostly with 24bit depth. Just for reference, CD encoding quality is 44.1kHz @ 16bit depth.
With the PiFi dac and the HifiberryDac+ I could not find any combinations of setting which added anything to sound quality. Less sharp, less details, muffled…
With the ‘Sabre’ also no improvement. This Chinese clone has relatively sound poor quality even without using SoXr.
The big surprise came with the cheapest Dac of the selection, a generic TI PCM5102 async Dac costing less than 5 Euro. To begin with the sound from the Dac is OK (but it’s not as good as the Dac’s using the PMC5122 chipsets but better than the ‘Sabre’). I found that using “96000:32:2” is really interesting. Details and clarity are a little less, but the soundstage opens up. Wider and seems to have more precision in the positioning of the players/instruments. For me this is now the standard setting for this Dac.
What I also found when experimenting is that the output bitrate should be set to one of the standard bit rates which the Dac supports and not a multiple of the input rate. These are generally 8, 16, 32, 44.1, 48, 96 and 192kHz. Don’t use 384kHz, it is not supported by the Pi (ALSA). I have a feeling that the Dac’s are optimised for processing these frequencies.
Setting the bitrate higher than 96kHz (e.g. 192kHz) does not seem to improve anything.
Increasing the depth to 32bits (if the Dac supports it) seems to be the key to what SoXr does to improve the soundstage. But this seems only to work when the input and output bitrate are different. It also works fine with output bitrates lower than the input, e.g. “32000:32:2”.
To summarise, SoXr changes the sound, usually negatively, unexpected soundstage improvements are possible, the effects are Dac specific, on/off changes are easy to hear, output bitrate and depth changes are extremely subtle (are they really there?) and most importantly it will be your personal preferences which will determine if it is better or not.
janui