Constant clicking problem (ipod 5g)

[saratoga]

Because of the sound noise problems I have found in both flavors of rockbox (both on my H140 and the Ipod 5G), and the fact that even some people (including at least 1 rockbox developer) seemed to think the X5 sounded fine when it was playing back with a 10khz lowpass filter make me a bit skeptical about rockbox audio quality claims...

Many people cannot hear high frequencies well or at all, and do not use good headphones.  Amoung those that can hear them, sensitivity to them is generally very low.  This is one of the principles used in MP3 encodeing. 

Its really not surprising that many people didn't notice the sample rate problem.

[PinkZeppelin]

This is one of the principles used in MP3 encodeing. 

Its really not surprising that many people didn't notice the sample rate problem.

Mp3 encoding is supposed to have a lowpass around 18khz, not 10khz.  Either people listen to music at ridiculously low bitrates, or have no clue on how to properly encode mp3s. 

[lostlogic]

The only claim I know of that rockbox makes about audio quality is that unlike many stock firmwares, we do not futz with the audio output of the codec, we simply play it back to the hardware as is (unless you turn on the various 'enhancement' options available).

As for the H140 buzz, that occurs with stock and rockbox firmwares as far as I know (at least it did on my H340), so don't try to pin that on rockbox, it's a problem with isolation on the iRiver players.

Finally, I do not doubt that the clicking problem occurs, as I said, we have a known issue with our I2C driver which could cause such a problem.  When using Rockbox, you need to realize that it is a _free software_ project.  It is developed for fun.  If you have a problem with it, fine, but don't act _entitled_ to a fix or  change to rockbox behavior.  You are more than welcome to develop your own solution, as the source is there for your modification.

I (unfortunately) do not have the software, hardware, nor knowledge to do RMAA analysis of my player, but I would be interested in other datapoints on the clicking issue.

[Llorean]

Also, keep in mind that Rockbox has not been RELEASED yet. On ANY of the software codec targets. While it's at version 2.5 for the Archoses, the overall 3.0 will be version 1 for the iRivers, and 3.1 will be the first release for the iPods. Until then, you shouldn't *expect* anything to work a specific way, or even at all. So while bug reports are welcome, you are using a prerelease software, and should keep that in mind at all points in time.

[saratoga]

This is one of the principles used in MP3 encodeing. 

Its really not surprising that many people didn't notice the sample rate problem.

Mp3 encoding is supposed to have a lowpass around 18khz, not 10khz.  

Please don't misquote me.  What I said was:

Amoung those that can hear them, sensitivity to them is generally very low.  This is one of the principles used in MP3 encodeing. 

Note that I said people had low sensitivity, not that they were lowpassing all high frequency sounds.  Theres a very big difference between the two. 

Anyway, MP3 lowpass is configurable, but the spec clearly intends you to use no higher then ~15-16KHz, not 18KHz.  Even if you force the lowpass higher, very little high frequency content is encoded (or rather very little bitrate is allocated due to the sfb21 issue in the MP3 format).  For this reason, when you FFT MP3 encoded audio, even with LAME, much of the energy above 16KHz is actually noise.

Either people listen to music at ridiculously low bitrates, or have no clue on how to properly encode mp3s. 

Or, as I already suggested, they're insensitive to higher frequency sound.

You may be interested in looking at the equal loudness curves here:

http://en.wikipedia.org/wiki/Absolute_Threshold_of_Hearing

or the information here:

http://en.wikipedia.org/wiki/Psychoacoustics

Note that high frequency sensitivity drops off VERY quickly with age.  You're already well past your prime by the time you're 20.  By age 40, the average male has already lost 10 fold as much high frequency sensitivity as they've lost at low frequencies. 

Its perfectly reasonable that many people using Rockbox would never notice a 22Khz sample rate, even listening to lossless audio.

[PinkZeppelin]

The only claim I know of that rockbox makes about audio quality is that unlike many stock firmwares, we do not futz with the audio output of the codec, we simply play it back to the hardware as is (unless you turn on the various 'enhancement' options available).

As for the H140 buzz, that occurs with stock and rockbox firmwares as far as I know (at least it did on my H340), so don't try to pin that on rockbox, it's a problem with isolation on the iRiver players.

Finally, I do not doubt that the clicking problem occurs, as I said, we have a known issue with our I2C driver which could cause such a problem.  When using Rockbox, you need to realize that it is a _free software_ project.  It is developed for fun.  If you have a problem with it, fine, but don't act _entitled_ to a fix or  change to rockbox behavior.  You are more than welcome to develop your own solution, as the source is there for your modification.

I (unfortunately) do not have the software, hardware, nor knowledge to do RMAA analysis of my player, but I would be interested in other datapoints on the clicking issue.

I just ran some tests on my H140 and the bass hum does indeed to effect both original and rockbox firmware, so I did erraneously pin that on rockbox. 

I am not upset at rockbox development, I am just upset when people claim that the audio sounds fine when it is not true, or when I make threads like this and don't get answers until I actually measure the problem.  It would have been nice to know that the H140s had a buzzing problem before I purchased it last year (but that is not really anyones fault).

Anyway, MP3 lowpass is configurable, but the spec clearly intends you to use no higher then ~15-16KHz, not 18KHz.

I don't think that is quite right.  Using the standard (and recommended) Lame 3.97b encoding setting, the resulting mp3s have a lowpass around 19khz.

Its perfectly reasonable that many people using Rockbox would never notice a 22Khz sample rate, even listening to lossless audio.

I think virtually nobody could discern anything above even 18khz, which is why many lossy encoders aim at transparency by using a lowpass around those frequencies.   


Nevertheless,

Thank you lostlogic for explaining the clicking problem.  I'm now looking into a purchase of an X5, just need to have someone post a spectral view like I did to confirm that there are no audio anomalies.

[saratoga]

I am not upset at rockbox development, I am just upset when people claim that the audio sounds fine when it is not true, or when I make threads like this and don't get answers until I actually measure the problem. 

Then you're welcome to hire someone to document your bug reports for you. 

Anyway, MP3 lowpass is configurable, but the spec clearly intends you to use no higher then ~15-16KHz, not 18KHz.

I don't think that is quite right.  Using the standard (and recommended) Lame 3.97b encoding setting, the resulting mp3s have a lowpass around 19khz.

Not to be an ass, but read the whole paragraph.  The very next sentence literally explains why you're misunderstanding the issue:

Even if you force the lowpass higher, very little high frequency content is encoded (or rather very little bitrate is allocated due to the sfb21 issue in the MP3 format). 

As I said, you can set the lowpass in LAME to whatever you want.  That doesn't mean you're getting much but noise above ~15-16k.  If you're still confused, google sfb21.  Its a well documented flaw in the MP3 format, and there are plenty of people who have explained it better then I.

Its perfectly reasonable that many people using Rockbox would never notice a 22Khz sample rate, even listening to lossless audio.

I think virtually nobody could discern anything above even 18khz, which is why many lossy encoders aim at transparency by using a lowpass around those frequencies.   

I don't disagree, but I don't think you understood my post.  I was talking about sample rate, not low pass (sample rate must be > 2x lowpass).

[Llorean]

Zep: A lot of time when you complain about audio quality, it's something you can only here with high impedence phones or other situations where not everyone notices it. Also, a lot of the developers use their Archoses, or other older players, as their primary listening, possibly even while in the process of testing things on-target for newer devices. So, sound related issues that are faint (and the ticking is something I don't hear, though I have heard the volume clicks) don't get noticed.

Then there's the fact that unless it's something very noticeable to *everyone* it's almost impossible to pin down what causes it. If and when you put up full documentation like the graph you did, it is *incredibly* helpful for finding out what's wrong. In any project it's always asked that you provide as much information about a bug, and a means of reproducing it. So, yeah, you didn't get a response until you measured the problem. But many MANY people have complained about the sound in Rockbox, and then when asked to measure it in the past, have come back with results that show it was identical. So, you'll have to forgive the developers here for taking complaints about sound quality with a grain of salt, but when you're used to people telling you Rockbox sounds like a "bag of shit" you start to want some evidence as to what *exactly* is wrong with the audio, so that not only you have proof it is wrong (if it doesn't show up in your phones) but also so you know where to begin looking for the problem.

[lostlogic]

Patience is of the essence here... the I2C driver problem that I believe to cause the clicking on ipod is something that needs fixing for many reasons (all of which are better than a -100dB clicking that is only audible with 100ohm earphones), I'm sure that I or someone will be working on the issue, but it could be a year before it changes.

It would probably make sense to use a headphone amp with your ipod in the meantime as the low impedence of the amp would eliminate the click.

[travishayes89]

all i know about the clicking is that if you turn the appropriate option off in the stock firmware you should get no clicking in it, and i never heard a clicking in Rockbox even with a pair of Sony "studio monitors" (i didn't feel like looking for the number).

[PinkZeppelin]

Patience is of the essence here... the I2C driver problem that I believe to cause the clicking on ipod is something that needs fixing for many reasons (all of which are better than a -100dB clicking that is only audible with 100ohm earphones), I'm sure that I or someone will be working on the issue, but it could be a year before it changes.

haven't been here in a while, just checking to see if any progress has been made so far?