Clip+ Sound Quality: Original Firmware vs. Rockbox

There was a thread somewhere quite a while ago, possibly on the Head-fi forum,  where people had come to the conclusion that the Clip’s “Normal” setting for EQ sounded identical to going into its custom EQ and setting all the sliders to +4. If you think that is the case, then Rockbox with no EQ would indeed sound different, because it would truly be a flat setting.

The whole thing is irrelevant to me, as I never use the OF anymore…but by all means, feel free to debate on. :wink:

I imagine that it would help if listeners could be trained using standardized types/extents of audio degradation, to be able to discuss the varoius types/extents of degradation and deviation from fidelity they hear in different audio, as subjective evaluators…

Better yet, a freeware package that imparts a calibrated amount of each type of degradation to the tester’s chosen audio file, for comparison.  For example:  ___% THD.  Or, ____(fill in your favorite type of degradation and units).

Barring such standardization across subjective evaluations descriptions, good luck trying to understand what one or another evaluator is trying to express about their opinion of some particular audio.

As for me, the RB audio sounded a little harsher, maybe more distorted, than OF. I would guess intermodulation distortion because the distortion of low frequencies seemed to be beating on some order(s) of higher frequencies, and vice versa, and worsened when more frequencies present.  Also, the frequency response of RB was relatively lacking, somewhat attenuated at both low and high frequencies.  As previously stated, the OF clarity is better, you can hear both lower and higher frequencies more clearly and there is more musical detail.  Highs and lows are not attenuated.  In summary:  Better fidelity.  This is all subjective, of course, but I always compare to my decent home sound systems with freestanding speakers to help provide a baseline reference.

@marvin_martian wrote:

There was a thread somewhere quite a while ago, possibly on the Head-fi forum,  where people had come to the conclusion that the Clip’s “Normal” setting for EQ sounded identical to going into its custom EQ and setting all the sliders to +4.

 

Clip+: http://rmaa.elektrokrishna.com/Comparisons/16%20Ohm%20Dynamic%20-%20Samsung%20R0%2C%20Cowon%20J3%2C%20Sony%20A845%2C%20Sansa%20Clip%2B/fr.png.

@sandclip wrote:

 

As for me, the RB audio sounded a little harsher, maybe more distorted, than OF.

 

I asked you earlier, but you did not reply.  So I shall ask you again.  Why not submit a bug report?


http://download.rockbox.org/daily/manual/rockbox-sansaclipplus/rockbox-buildap5.html#x20-371000E.1.1.


Complaining about it here is probably not going to get the problem solved.  If, however, you submit a bug report and one of the Rockbox developers is able to confirm it, a solution might be possible.

@sandclip wrote:

 

Better yet, a freeware package that imparts a calibrated amount of each type of degradation to the tester’s chosen audio file, for comparison.  For example:  ___% THD.  Or, ____(fill in your favorite type of degradation and units).

 

 

 

Lots of people have done this:

http://rmaa.elektrokrishna.com/Temp/Clip%2B%20Rockbox%20HQ%20Patch%2065%20Ohm%20incl%200x1B.htm

(look at the  Vol 0x1B build compared to the OF.)   Output is basically identical, since both firmwares use nearly the same settings.


@sandclip wrote:

As for me, the RB audio sounded a little harsher, maybe more distorted, than OF. I would guess intermodulation distortion because the distortion of low frequencies seemed to be beating on some order(s) of higher frequencies, and vice versa, and worsened when more frequencies present.  Also, the frequency response of RB was relatively lacking, somewhat attenuated at both low and high frequencies.  As previously stated, the OF clarity is better, you can hear both lower and higher frequencies more clearly and there is more musical detail.  Highs and lows are not attenuated.  In summary:  Better fidelity.  This is all subjective, of course, but I always compare to my decent home sound systems with freestanding speakers to help provide a baseline reference.


And this is why no one should care about subjective comparisons :slight_smile:  They’re always wrong.  

@lobak wrote:

Hi, just Rockboxed my Sansa Clip+ yesterday, so I could play 24/96 flac files without having to downsample them. The difference in audio quality is quite huge. The OF has a smoother and rounder sound which I think sounds a lot better.

Yes, the Clip hardware cannot play back 96kHz files, so rockbox will use a low quality (but high battery life) resampler to convert them to 44.1kHz.  You will get substantially better quality if you properly resample your files on the PC (i recommend SoX or foobar, both have very high quality resamplers).

@contrapuntal wrote:

 

@sandclip wrote:

 

As for me, the RB audio sounded a little harsher, maybe more distorted, than OF.

 

I asked you earlier, but you did not reply.  So I shall ask you again.  Why not submit a bug report?


http://download.rockbox.org/daily/manual/rockbox-sansaclipplus/rockbox-buildap5.html#x20-371000E.1.1.


Complaining about it here is probably not going to get the problem solved.  If, however, you submit a bug report and one of the Rockbox developers is able to confirm it, a solution might be possible.

 

Who’s complaining?  I simply observed that RB sounded inferior to OF in my opinion, and posted a thread to poll other users for their opinions and possibly data.

If RB is signficantly (noticeably) inferior in sound quality compared to OF, as my own listening indicates, this may not even qualify as a “bug”, it may just be a “characteristic”.

If the RB coders do not hear the difference and/or consider it worth addressing, then chances are they are not equipped to refine or improve the sound quality, even if they WERE to attempt to do so.  (Technical measurements by themselves are not enough, if an evaluator has dull hearing and/or sub-par headphones.)

I am just a simple user.  If one setup sounds inferior to another, I don’t really have time/resources to engage in extensive testing to produce reams of data and collaborate with developers to assist them in making improvements.  I’ll just keep using OF until I see info indicating that RB sound quality issues have been improved and verified, no big deal.  The Clip+ with OF sounds phenomenally good to me, on my personal-favorite headphones.  :slight_smile:

@saratoga wrote:

 

<snip>

 

 

 

 

And this is why no one should care about subjective comparisons :slight_smile:  They’re always wrong.  

 

Respectfully:  Subjective comparisions are not “always wrong”.  Well, maybe YOURS are, but mine are not.

It is a gross oversimplification and fallacy to assume that subjective comparisons are “always wrong”.  They do vary by evaluator because evaluators are different, they are subject to evaluator bias that sometimes includes hidden agendas or motives, and they are difficult (to impossible) to verify and replicate.  But that does not mean they are “always wrong”.

If I cannot trust my own subjective discernment because it is “always wrong”, then I might as well give up on trying to obtain better-quality in most areas of my life.  Music/audio, food, coffee, vehicles, housing, air quality…the list goes on and on.

This RB vs. OF sound quality thread is interesting in illuminating individual emotional responses rooted in misconception and fallacy, but its really not very productive until someone links to data, or a posting says “News:  Latest RB firmware update includes sound quality improvements”.  Meanwhile, I am just fine with OF.  It sounds better to my brutally-honest subjective opinion.  :slight_smile:

For me, Rockbox resulted in vastly improved sound quality, due to the vastly superior parametric equalizer. How’s that for subjective? :stuck_out_tongue:

@sandclip wrote:

I simply observed that RB sounded inferior to OF in my opinion, and posted a thread to poll other users for their opinions and possibly data.

The majority’s opinion is that there is nothing wrong with the sound quality of Rockbox’s firmware.  What kind of data are you looking for?

If RB is signficantly (noticeably) inferior in sound quality compared to OF, as my own listening indicates, this may not even qualify as a “bug”, it may just be a “characteristic”.

If it sounds worse, it is a bug.  Please see what the Rockbox developers have to say about the version-two Clip.  


http://www.rockbox.org/wiki/SansaAMS#Port_Status: “There is some very annoying noise which can be heard especially at low volume.” 


It was reported as a problem by the Rockbox developers.  A problem is a bug. 

If the RB coders do not hear the difference and/or consider it worth addressing, then chances are they are not equipped to refine or improve the sound quality, even if they WERE to attempt to do so. 

How do you know this is true if you are unwilling to help them by submitting a bug report?

(Technical measurements by themselves are not enough, if an evaluator has dull hearing and/or sub-par headphones.)

You have yet to provide even the tiniest shred of evidence of your so-called superior hearing.  Why?

I am just a simple user.

This is the problem.  You do not understand the pitfalls of subjective evaluations of sound quality.  Did you even bother to read those articles I linked to? 


http://forums.sandisk.com/t5/Clip-Clip/Highest-FLAC-quality-playable-on-the-clip/td-p/220134/page/4#M35147


Because if you did, these pitfalls should have been made very clear to you.  Do you know why ABX tests are done?  Do you know what the placebo effect is?  Did you ABX?

If one setup sounds inferior to another, I don’t really have time/resources to engage in extensive testing to produce reams of data and collaborate with developers to assist them in making improvements. 

I think the more plausible explanation is that you know Rockbox’s firmware sounds as good as SanDisk’s firmware.  But, because of confirmation bias, you do not want to publicly admit that you were mistaken.

I’ll just keep using OF until I see info indicating that RB sound quality issues have been improved and verified, no big deal. 

How can this be an issue if you do not consider it a bug.  So you are unwilling to help the Rockbox developers by submitting a bug report yet remain hopeful that this “characteristic” changes in the future.  You really think this is realistic?

Respectfully:  Subjective comparisions are not “always wrong”.  Well, maybe YOURS are, but mine are not.

No one is immune from the placebo effect.

It is a gross oversimplification and fallacy to assume that subjective comparisons are “always wrong”.  They do vary by evaluator because evaluators are different, they are subject to evaluator bias that sometimes includes hidden agendas or motives, and they are difficult (to impossible) to verify and replicate. 

Impossible?  Sounds like your claims. 

If I cannot trust my own subjective discernment because it is “always wrong”, then I might as well give up on trying to obtain better-quality in most areas of my life.  Music/audio, food, coffee, vehicles, housing, air quality…the list goes on and on.

For the purposes of this thread, let us just discuss your so-called superior hearing.  Please ABX some files at –V0 using LAME 3.98.4 against their FLACs and report back to us. 

This RB vs. OF sound quality thread is interesting in illuminating individual emotional responses rooted in misconception and fallacy, but its really not very productive until someone links to data, or a posting says “News:  Latest RB firmware update includes sound quality improvements”. 

Yes—like confirmation bias.  Sandclip, the person with the most misconceptions and fallacies in this thread is you.  If you do not believe me, please read those articles that I linked to for the explanation.

Meanwhile, I am just fine with OF.  It sounds better to my brutally-honest subjective opinion.  :slight_smile:

There is no such thing as a brutally-honest subjective opinion when evaluating the sound quality of MP3 players.  For such an individual to be taken seriously, this individual must show humility, at least some knowledge of audio, a very strong understanding of the placebo effect and how to guard against it, a willingness to admit to mistakes, and a desire to learn, especially from one’s mistakes.

@sandclip wrote:

 

 

Respectfully:  Subjective comparisions are not “always wrong”.  Well, maybe YOURS are, but mine are not.

 

Well do your own RMAA tests and prove it then.  Its always possible something is broken on your player that makes a real difference, but that doesn’t happen to other people’s devices.  Really unlikely, but possible.  Let me know if you find something.

@sandclip wrote:

 

 

If I cannot trust my own subjective discernment because it is “always wrong”, then I might as well give up on trying to obtain better-quality in most areas of my life.  Music/audio, food, coffee, vehicles, housing, air quality…the list goes on and on.

 

FWIW it is a very common misconception amoung people that hearing subtle differences between devices is somehow simple or natural.  It is not.  It takes enormous amounts of practice.  Exceptionally few people are good at this, and most are professional acoustic engineers, rather then casual listeners on the internet.

@sandclip wrote:

 

 

This RB vs. OF sound quality thread is interesting in illuminating individual emotional responses rooted in misconception and fallacy, but its really not very productive until someone links to data, or a posting says “News:  Latest RB firmware update includes sound quality improvements”.   

 Perhaps you missed it, but I did link data above.  The problems you describe did not occur with the devices under test.  If you think this is a mistake, please test your device.  

Have to agree. The SQ of the original Sansa firmware is better to my ears too. The difference I feel is in lack of bass extension. Slight but noticible. That was good enough for me to revert back to the OF to listen to music and then RB when I wanted to play games. :smiley:

“Lack of bass extension” would be measurable.

If you really want to do an actual comparison–a blind testing–do it and see if you can hear the difference when you don’t know what the source is.

http://sourceforge.net/projects/abchr/

You can also compare “bass extension”–I assume you mean that you don’t hear equally deep bass? so some low frequencies are missing or muted?–by looking at the waveforms in Audacity.

http://audacity.sourceforge.net/

I think VST plugins work in Audacity so if you really want to get scientific you could try this:

http://www.voxengo.com/product/span/

Otherwise, as Saratoga (a Rockbox developer) pointed out, Rockbox may have a flatter equalization curve (that is, not boosting any frequencies for boom and sizzle–closer to the actual waveform) than the OF. Or one may be more compatible with  your particular headphones than the other.

But as Contrapuntal very patiently and fruitlessly pointed out, it would help to be able to precisely describe the difference and measure it–and explain how you’re making the measurements (like “it sounds way better through my $10 earbuds”)–before  making the big generalization.

Subjective impressions do have a point–everyone has a different pair of favorite headphones. But if you’re talking about the signal that’s coming out of the headphone jack–rather than how your headphones sound in your ears–then there are ways to measure that.

Note that I haven’t tried Rockbox so far, nor have I compared it to the Original Sansa firmware. I’m just fascinated by the communication dynamics within this thread. And of course a little bit emotionally involved, since I’m someone who supports the existence of sonic differences among audio electronics even without corresponding proof in the form of measuring data or ABX tests.


Black-Rectangle wrote:

“Lack of bass extension” would be measurable.


Maybe. But for all-embracing results it’s not sufficient to measure sine waves – which is the norm. You’d also have to measure the dynamic behavior of the device, possibly influenced by the current-supply capabilities (–> power supply) – which is never done (and admittedly hard to do). Also the amplitude response at low impedances, for that matter (–> coupling capacitors on the headphone output).


Marvin_Martian wrote:

There was a thread somewhere quite a while ago, possibly on the Head-fi forum, where people had come to the conclusion that the Clip’s “Normal” setting for EQ sounded identical to going into its custom EQ and setting all the sliders to +4. If you think that is the case, then Rockbox with no EQ would indeed sound different, because it would truly be a flat setting.


That’s a claim made out of thin air, and that’s definitely not the case – as some measurements have shown in the very same thread (in which I’ve been involved).


Contrapuntal wrote:

No ABX = bias.


I’m with Sandclip here. It would be more adequate to speak of a «potential bias» instead of a definitive bias. There’s no proof that any possibly existing bias has to definitely and decisively influence the subjective judgment in the form of the perception of a nonexistent sonic difference based on non-audio factors.


Contrapuntal wrote:

What do you mean by smoother and rounder?  It should sound neutral.  Please be as specific as possible.


«Neutral» is a term of wide comprehension. Most amps (and CD, DVD… players) without activated EQ sound and measure «neutral», without sounding the same to passably sensitive ears. For the description of sonic characteristics you can’t make scientific statements – that would be mere speculation –, you have to restrict yourself to what you effectively and subjectively hear. E.g., a minor dose of harmonic distortion isn’t audible as such, but may appear as additional warmth or colorfulness or harshness, depending on the characteristic of the distortion. In this context a statement such as «sounds like added distortion» even if correct by accident, would be of no use for people interested in the perceived sound characteristic.


Black-Rectangle wrote:

“Lack of bass extension” would be measurable.


Maybe. But for all-embracing results it’s not sufficient to measure sine waves – which is the norm. You’d also have to measure the dynamic behavior of the device, possibly influenced by the current-supply capabilities (–> power supply) – which is never done (and admittedly hard to do). Also the amplitude response at low impedances, for that matter (–> coupling capacitors on the headphone output).


saratoga wrote:

And this is why no one should care about subjective comparisons :slight_smile:  They’re always wrong.


A surprisingly subjective (and certainly wrong) statement from a self-confessed objectivist!


Black-Rectangle wrote:

Subjective impressions do have a point–everyone has a different pair of favorite headphones. But if you’re talking about the signal that’s coming out of the headphone jack–rather than how your headphones sound in your ears–then there are ways to measure that.



Contrapuntal wrote:

So far I have yet to see any objective data presented in this thread.


But keep in mind that data can only show measuring differences, they don’t tell if the latter are responsible for perceived sonic differences or not.

I’ve been following a Head-Fi thread some time ago with the same topic. Interestingly some people heard no difference, some heard a difference – but if, it was always in favor of the original firmware. A coincidence?


Sandclip wrote:

I compared Rockbox 3.7.1 vs. original firmware 01.02.15 on the Clip+. It sounds to my ears like the original firmware is slightly better fidelity/quality output than Rockbox.  I did not do a true blind ABX comparison or anything like that, just switched back and forth between OF and RB repeatedly, on the same passages of certain music FLAC s.



saratoga wrote:

Yes, the Clip hardware cannot play back 96kHz files, so rockbox will use a low quality (but high battery life) resampler to convert them to 44.1kHz.  You will get substantially better quality if you properly resample your files on the PC (i recommend SoX or foobar, both have very high quality resamplers).


This may be one of the key statements in this thread, and I wonder why it hasn’t got more attention. What if the difference only shows up with FLAC s?

However, with a bit more tolerance and less subjective bias the thread would have been more fruitful. In first place I’m addressing the unfortunate claim for proof and supporting data, whereas the thread starter just wanted to get some (subjective) feedback from the community or alternatively some relevant data – with a hail of subjectivist criticism for posting such a heretic topic at all. – Imagine you’re not allowed to post your listening impressions in a public audio forum!

 

Black-Rectangle wrote:

 

“Lack of bass extension” would be measurable.

 

Maybe. But for all-embracing results it’s not sufficient to measure sine waves – which is the norm. 

 

 

Actually, yes, it is sufficient to measure sin waves.  They form an orthogonal basis set, so why not?

 

 

 

Contrapuntal wrote:

 

No ABX = bias.

 

I’m with Sandclip here. It would be more adequate to speak of a «potential bias» instead of a definitive bias. There’s no proof that any possibly existing bias has to definitely and decisively influence the subjective judgment in the form of the perception of a nonexistent sonic difference based on non-audio factors.

 

 

I don’t care about things that are potientially true.  Either show that there is a problem, or do not waste my time with nonsense bug reports.  

 

 

 

 

Sandclip wrote:

 

I compared Rockbox 3.7.1 vs. original firmware 01.02.15 on the Clip+. It sounds to my ears like the original firmware is slightly better fidelity/quality output than Rockbox.  I did not do a true blind ABX comparison or anything like that, just switched back and forth between OF and RB repeatedly, on the same passages of certain music FLAC s.

saratoga wrote:

 

Yes, the Clip hardware cannot play back 96kHz files, so rockbox will use a low quality (but high battery life) resampler to convert them to 44.1kHz.  You will get substantially better quality if you properly resample your files on the PC (i recommend SoX or foobar, both have very high quality resamplers).

 

This may be one of the key statements in this thread, and I wonder why it hasn’t got more attention. What if the difference only shows up with FLAC s?

 

 

I have no idea what you’re getting at.  If you think there is a problem with the FLAC decoder, it is trivial to confirm that the output is lossless.

 

FWIW, over the years since this thread was posted, no one else has found these problems, and I’ve confirmed that they do not exist probably 10 or 15 times while working on unrelated bits of the code, so its probably a safe bet that they’re imaginary. 

 

 

@saratoga wrote:


@contrapuntal wrote:

 

No ABX = bias.


 

I’m with Sandclip here. It would be more adequate to speak of a «potential bias» instead of a definitive bias. There’s no proof that any possibly existing bias has to definitely and decisively influence the subjective judgment in the form of the perception of a nonexistent sonic difference based on non-audio factors.

 

I don’t care about things that are potientially true. Either show that there is a problem, or do not waste my time with nonsense bug reports.

Yes, and that’s the weakness in your argumentation: it’s black or white. Moreover, in the meantime the thread starter has made clear that he’s not speaking of a bug – but a characteristic. From my perspective I absolutely wouldn’t state the Clip+ to have a sonic bug just because it sounds inferior to the Clip Zip to my ears. And of course I would dare to report this perception of mine in this public forum, without delivering ABX protocols and measuring data. If you don’t want to waste your time in a thread like this, you certainly know what’s there to do.

@saratoga wrote:


@sandclip wrote:

 

I compared Rockbox 3.7.1 vs. original firmware 01.02.15 on the Clip+. It sounds to my ears like the original firmware is slightly better fidelity/quality output than Rockbox.  I did not do a true blind ABX comparison or anything like that, just switched back and forth between OF and RB repeatedly, on the same passages of certain music FLAC s.



@saratoga wrote:

 

Yes, the Clip hardware cannot play back 96kHz files, so rockbox will use a low quality (but high battery life) resampler to convert them to 44.1kHz.  You will get substantially better quality if you properly resample your files on the PC (i recommend SoX or foobar, both have very high quality resamplers).


 

This may be one of the key statements in this thread, and I wonder why it hasn’t got more attention. What if the difference only shows up with FLAC s?

 

I have no idea what you’re getting at.  If you think there is a problem with the FLAC decoder, it is trivial to confirm that the output is lossless.

Well, maybe I was a bit premature in my reasoning – since the «low-quality» resampling process only takes action in the case of high sampling rates, which Sandclip didn’t explicitly mention. But it’s still possible that it’s involved in the problem.

I’m aware that this is an old thread, but after all it’s been woken up by a fresh post recently. Although I tend to think that the reports of Rockbox sounding minimally worse might have a serious background (see also the one-sided tendency of reports), I certainly won’t pretend it to be fact. My main concern was the messages between the lines in this thread anyway. :wink:

@jazz wrote:

Yes, and that’s the weakness in your argumentation: it’s black or white. Moreover, in the meantime the thread starter has made clear that he’s not speaking of a bug

 

 


 

If it actually sounds different, that is a bug.  Black and white.  You can try and make grey from that, but it doesn’t make any sense to do so.

 

 

@jazz wrote:

But it’s still possible that it’s involved in the problem.

 


 

 Its also possible rainbows will fly out of your butt.  And probably about as likely.

 

 

@jazz wrote:

Although I tend to think that the reports of Rockbox sounding minimally worse might have a serious background

 


 

 

Given how many years have passed, and the fact that no one has ever been able to reproduce this, I think its more likely that the original reporter was just mistaken.  But by all means, go ahead, get out RMAA and check for yourself.  

Well, I’ll be generous and overlook your increasing sarcasm. After all you belong to the (absolutely appreciated and commendable) Rockbox developers. How would I react in your situation?

@jazz wrote:

Well, I’ll be generous and overlook your increasing sarcasm. 

To be clear, lecturing me about things you do not understand is not “being generous”.  It is being ignorant, and a waste of everyone’s time.  I realize most people have no real understanding of how software or hardware work, and I don’t hold that against them, but at the same time if your’e going to tell me about “the weakness of my arguement” or how debug code, I expect you to at least know something about what you are saying.  If you do not understand the topic at the same level I do (and clearly you do not), then please do not presume to lecture me on it.

@jazz wrote:

After all you belong to the (absolutely appreciated and commendable) Rockbox developers. How would I react in your situation?

You should be more concerned with how foolish you have actually acted, not with how much better you might have acted in other circumstances.    

I still stick to everything I’ve stated, save for the thing(s) I’ve taken back. Moreover most of my objections weren’t of a technical kind, so I don’t accept any natural superiority of your standpoint. I’ve never criticized your field of activity or work.