Like many here, I have ROCK running on a NUC which serves Raspberry Pi endpoints on the network. This server can deliver bit-perfect music via RAAT to Roon Bridge software running on the RPi, or bit-perfect music via Roon’s implementation of LogiTech Media Server to Squeezebox software running on the same RPI. The endpoint operating system is DietPi.
Same music, same volume, same hardware, same room and speakers. No DSP, no leveling or other software adjustments, so still bit-perfect according to the DAC.
Other than expectation bias in my brain, why do they sound different?
Are you sending the PCM or the FLAC to the SqueezeLight endpoing?
To confirm that both remain bit-perfect, I have been listening to MQA-encoded FLAC from tidal.
Not what I meant I think,
I hear the difference running a Stack Audio Link II, which uses commercial version of RopieeeXL. I personally prefer the roon bridge endpoint. Find it has sharper focus and deeper soundstage. The Squeezelite endpoint is a little wider but sounds more diffused. It’s all kind of subtle though but I can hear it.
Yes I too have heard a difference when using either on the pi and ropieeee. It’s not massive but it is different. More like a filter change on a DAC, not saying one is better than the other just that there is a difference. I still use RAAT and won’t change its not enough really to worry about.
So if I shut off Sqeezelite support, I think it actually improves the roon bridge endpoint. It sounds more dynamic. Can hear more punch and better soundstage placement.
I’m using Focal Utopia’s out of Focal Arche. Prior to the Arche, I was using Chord Mscaler/Dave and heard the same changes. You hear people saying would only be due to poorly designed dac but Chord Dave and mscaler are supposed to be endgame. This was feeding it the Link II.
Strange because you also have people saying there should be no difference with source but I just added a linear power supply to the Link II and it made a noticeable improvement for the better. Deeper bass and more relaxed natural sound.
Basically think it all has to do with RF interference making a brighter sound and could be what’s happening with different software. If one is heavier processing, might subtly effect the sound.
I bring this up separately because the ongoing discussion about possible sonic improvement in upcoming version 1.8 has now veered off into camera technology. That lengthy topic had included some postings with the idea that different bit-perfect delivery schemes can alter sonics, with Squeezelite cited as an example.
Bits Are Bits proponents say that it’s all in our heads, and that, as long as the signal is bit-perfect, Roon has done everything possible to maximize SQ.
My take on it is the less processing the better. So that’s why hooking it up directly to my intel Nuc roon core sounds flatter and less engaging then Link II endpoint.
Also with Sqeezelite endpoint, you can change setting for sending uncompressed flac or not. Suppose to sound better uncompressed. As said, prefer roon bridge even over that setting.
Yes, I prefer the network endpoint over NUC-direct as well. My favorites so far are from Meridian and Allo.
This hobby compels me to keep pushing ahead, so I don’t want Roon to give up on finding ways to enhance sound quality.
Show us the signal path in both options. That may or may not give some insight.
Yes I get that but I don’t find the roon bridge lacking. Think sound quality just dependent on your source. Roon already recommends against hooking directly up to the core since they are not concerned about sound quality from that standpoint, just features. The roon bridge is already slimmed down and if used on a minimum streamer, you should be good. It’s not to say that they can’t figure out how to reduce the overhead on the bridge software even more.
Meridian DAC shows bit-perfect in both instances:
And what lights do you get on the explorer 2 with respect to MQA and sample rate?
And do you get similar differences in SQ with non MQA files?
The appropriate combination of white, green or blue lights come on to indicate sample rates and MQA status using either Roon Bridge or Squeezelite. Tidal offers two versions of this particular Shelby Lynne recording: 96 kHz with blue MQA and one white showing, or 192 kHz with blue MQA and two whites illuminated.
The differences are consistent with other sources, including local files and Qobuz. I only use MQA as an example because the DAC should not decode (and show green or blue) if the signal isn’t bit-perfect.
I can’t find the source in my quick search, but I’ve read in the past that so long as certain flags exist, the MQA light will illuminate without regard to being bitperfect. Someone did some specific testing on this. Sorry that I can’t locate the source. But just in case this is relevant.
I don’t have Tidal but I used a file in my library not streaming. I also had LMS to transcode flac to WAV to send on to Squeezelite o no flac processing on the device and as close to what Roon does as possible. Both players are bit perfect as I have performed a test with my RME DAC sending there test files which tests for this.
Well it does seem like the only difference is RAAT and squeezlite so I can’t explain why there is a difference in SQ.
There you go.
Hopefully, someone who knows this stuff better than I do can let us know if it’d conceivably apply to real-world usage.