This seems highly unlikely to me. There are significant changes in the sound of the player from firmware version to firmware version. I'm pretty convinced that the entire sound of the DX50 is the resu
lt of some pretty heavy DSP processing. Isn't it possible that the guy who answers the emails at iBasso has no idea what the programmers are doing?
Hello,
To me it is crazy When you get the player listen to it and like the sound, and then the manufactor keeps updating the software so there are more features and less bugs, but change the quality of the audio? I don't understand how the owners on this thread just learn to accept this and adapt to the lesser quality in sound? It may sound good but not as good as the prior releases? I do believe there are other owners on this thread that agree with me. Sorry for the ramble but I am just stating my opinion. When I update the firmware on my Anthem reciever the sound is never effected?
This seems highly unlikely to me. There are significant changes in the sound of the player from firmware version to firmware version. I'm pretty convinced that the entire sound of the DX50 is the result of some pretty heavy DSP processing. Isn't it possible that the guy who answers the emails at iBasso has no idea what the programmers are doing?
I have done some noodling with ffmpeg's audio filters (libavfilter). Little tweaks have subtle effects on what I hear. That's the point. While I have not tried to replicate DX50's changes with ffmpeg the changes that I have created are within the same scope as those that I've heard in my DX50. So yes, if iBasso is pulling from an actively developed filtering library like libavfilter then yes, upstream changes could can have a subtle (or not so subtle in the case of 1.2.3beta) changes on sound output.
Of course, the whole process of converting bits into sound waves is "heavy DSP" so I can't deny the technicality of your statement. Nevertheless, I maintain that the changes from 1.2.2 through 1.2.7 excluding 1.2.3b are subtle, not "significant" as you and others have described. You see, the problem with the claims of significant differences is that you're failing to tell us what your benchmark for "significant" is. When I put 1.2.2, 1.2.3b, 1.2.5 and 1.2.6 side by side as best I could I figured (based on fiddling with the EQ settings) the differences were less than +/- 3dB in the various frequency ranges with the most variance in 1.2.3b. Given that amplifiers like the Centrance HiFi-M8 and JDS C5 have bass and/or treble boost of +3dB to +6dB I call anything less than +3dB "subtle", anything from +3dB to +6dB "noticeable" and possibly "annoying" under certain circumstances. Get past +6dB and then we can talk about just how significant the changes really are.
I have done some noodling with ffmpeg's audio filters (libavfilter). Little tweaks have subtle effects on what I hear. That's the point. While I have not tried to replicate DX50's changes with ffmpeg the changes that I have created are within the same scope as those that I've heard in my DX50. So yes, if iBasso is pulling from an actively developed filtering library like libavfilter then yes, upstream changes could can have a subtle (or not so subtle in the case of 1.2.3beta) changes on sound output.
Of course, the whole process of converting bits into sound waves is "heavy DSP" so I can't deny the technicality of your statement. Nevertheless, I maintain that the changes from 1.2.2 through 1.2.7 excluding 1.2.3b are subtle, not "significant" as you and others have described. You see, the problem with the claims of significant differences is that you're failing to tell us what your benchmark for "significant" is. When I put 1.2.2, 1.2.3b, 1.2.5 and 1.2.6 side by side as best I could I figured (based on fiddling with the EQ settings) the differences were less than +/- 3dB in the various frequency ranges with the most variance in 1.2.3b. Given that amplifiers like the Centrance HiFi-M8 and JDS C5 have bass and/or treble boost of +3dB to +6dB I call anything less than +3dB "subtle", anything from +3dB to +6dB "noticeable" and possibly "annoying" under certain circumstances. Get past +6dB and then we can talk about just how significant the changes really are.
+1 This I completely agree. 1.2.3b is the odd firmware of the bunch and the sonic change is noticeable. Between 1.2.2 and 1.2.5 the change is very subtle. > 1.2.5 is really unnoticeable unless you really have a laser beam focus on each element of the sound which not you would do when normally listening to music.
TBH, if you can live with media scanning delays every time you turn on DX50 1.2.2 is good enough.
In that case, it's possible he considers their DSP system and its specific settings "proprietary" to iBasso and a perhaps one of their trade secrets, in which case he would be very reluctant to talk about it.
Of course, the whole process of converting bits into sound waves is "heavy DSP" so I can't deny the technicality of your statement. Nevertheless, I maintain that the changes from 1.2.2 through 1.2.7 excluding 1.2.3b are subtle, not "significant" as you and others have described. You see, the problem with the claims of significant differences is that you're failing to tell us what your benchmark for "significant" is. When I put 1.2.2, 1.2.3b, 1.2.5 and 1.2.6 side by side as best I could I figured (based on fiddling with the EQ settings) the differences were less than +/- 3dB in the various frequency ranges with the most variance in 1.2.3b. Given that amplifiers like the Centrance HiFi-M8 and JDS C5 have bass and/or treble boost of +3dB to +6dB I call anything less than +3dB "subtle", anything from +3dB to +6dB "noticeable" and possibly "annoying" under certain circumstances. Get past +6dB and then we can talk about just how significant the changes really are.
I would call a change of 1db significant, but I'm not sure how noticeable that would be. A 3db change would definitely be noticeable. But isn't the whole point of a "reference" player to have a flat frequency response with minimal colorations?
I wouldn't use "significant" or "noticeable" to describe the bass boost of 1.2.3b. I would call that "ridiculous." I would probably use the same word to describe the FiiO X3's mid-bass bump as well.
To my ears, the DX50's sound in general is pleasing, but artificial. Its sound reminds me of those early DSP receivers that had preprommed settings like "jazz club," "arena," "concert hall," "stadium," etc. This is especially noticeable when it comes to the width of the soundstage.
The FiiO X3 sounds more honest to me, but the mid-bass boost makes the player completely unlistenable. My hunch (pure speculation) is that the Wolfson DAC is the cause of the mid-bass bump in the FiiO. The reason it's not a problem with the DX50 is becuase iBasso is using a lot of digital processing to "tame" the midbass.
I just don't buy the "library" explanation by itself. If that were the case, why does no other player have these noticeable changes in sound with firmware updates (with the suspicious exception of iBasso's own DX100)?
If my memory serves me well, official reply from iBasso have always maintained the sound signature never changed from FW to FW. Why the change of stance now? Have they succumb to the pressure to admit the changes? or have they suddenly realised that the sound signature does infact changes?
Also, whenever I update driver for my PC soundcard, etc, the sound signature have always been the same. Only bug fixes and occasionally, extra features added.
+1 This I completely agree. 1.2.3b is the odd firmware of the bunch and the sonic change is noticeable. Between 1.2.2 and 1.2.5 the change is very subtle. > 1.2.5 is really unnoticeable unless you really have a laser beam focus on each element of the sound which not you would do when normally listening to music.
TBH, if you can live with media scanning delays every time you turn on DX50 1.2.2 is good enough.
Nah, 1.2.5 has a much more noticeable midbass hump. It's ALWAYS going, whereas 1.2.2 depends more on the individual song selected. Thing is, it's nice because it doesn't bleed into the mids. Quite fun, actually.
iBasso's explanation makes perfect sense and it is actually the only explanation that makes sense.
AK100 has had the same issue. It is not unique.
As for a neutral rendering of digital to analog, it is does not really exist. Decisions have to be made. That is why the CODECs can differ.
A good analogy is a digital camera. When processing raw digital files to a TIFF or JPEG, decisions on white balance, contrast and saturation, among others, have to be made. You cannot ignore these decisions and arrive at neutral. You can make them and then argue over whether or not what you achieved is actually neutral. It is the same with CODECs and digital filtering etc during the DAC process. Decisions have to be made.
As for sound quality getting worse, that is a matter of taste. The CODECs are being continually updated in whatever open source iBasso is using. It seems that the updates can be a bit experimental. 1.2.3beta is an example of what I consider to be some unusual values (excessively wide sound stage - not natural).
There is no conspiracy here. (I will leave it at that...)
+- 3 dB is relatively huge. DSP is likely the cause of these perceptive changes, but I would bet that if someone actually measured the DX50 FR across firmware versions, it would graph flat on every single one.
Originally Posted by cooperpwc /img/forum/go_quote.gif
The CODECs are being continually updated in whatever open source iBasso is using. It seems that the updates can be a bit experimental. 1.2.3beta is an example of what I consider to be some unusual values (excessively wide sound stage - not natural). 1.2.2 and 1.2.7 are my favourites but obviously some disagree.
I don't profess to know much about these open source codecs, but if that's the case, then doesn't the fact that they're being continually updated (with the result being noticeable differences in sound quality with LOSSLESS audio formats) mean that they're pretty bad codecs to begin with?
It makes no sense to me that only the two DAPs made by iBasso would exhibit these noticeable differences in sound with firmware updates, yet no other player I've heard (incluing Foobar, Winamp, Rockbox iPod, Rockbox Clip, etc.) behaves the same way.
I don't profess to know much about these open source codecs, but if that's the case, then doesn't the fact that they're being continually updated (with the result being noticeable differences in sound quality with LOSSLESS audio formats) mean that they're pretty bad codecs to begin with?
It makes no sense to me that only the two DAPs made by iBasso would exhibit these noticeable differences in sound with firmware updates, yet no other player I've heard (incluing Foobar, Winamp, Rockbox iPod, Rockbox Clip, and frankly every other player I've used) behaves the same way.
Here I partly agree. Not so much that they are using bad CODECs but rather that iBasso might be well advised to use the latest "stable" or "official" CODEC releases rather than daily updates. (That is how these open source projects are usually structured.) That would produce more uniform and less experimental results.
I think you have hit the nail on the head. The nicest sound i have experienced (dac wise) is from the metrum acoustics octave. I have listened to quite a few and owned a few different dac's and the octave to my ears is my preference. Still i want to give the dx50 a chance, i shall hold onto it and wait for a couple more firmware revisions. As some folk have mentioned in this thread that the sound has actually changed from one firmware revision to the next. Perhaps the next firmware might just be to my liking.
It makes no sense to me that only the two DAPs made by iBasso would exhibit these noticeable differences in sound with firmware updates, yet no other player I've heard (incluing Foobar, Winamp, Rockbox iPod, Rockbox Clip, etc.) behaves the same way.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.