Sansa Clip Firmware Update

@pickmorel wrote:

The most disgusting part of this whole thing is that, unless I’m really misreading what they posted before they discontinued the Clip, is that they went on record, in writing, committing to apply the fixes that _ they said they had already created. _

 

What fixes are you talking about, and can you provide links to any post in which SanDisk said they had already fixed them?

Sansa Clip Pitch Issue

Summary

Personal observation, as well as reports of other users, indicate that music files played on the Sansa Clip exhibit lower pitch, and thus longer duration, than expected. However, SanDisk technical support have stated as late as 2009/09/17 that this problem was fixed by firmware release 02.01.32. I tested SanDisk’s assertion by playing the same files on two Sansa Clips with the latest firmware version, as well as on two other MP3 players (an Apple iPod and a Philips GoGear).

The tests show that Sansa Clip with the latest firmware update continues to exhibit the pitch/duration error. The duration of an MP3 file sampled at 44.1kHz and played on a Sansa Clip was found to be >1% greater than the original, corresponding to a pitch lower by 19–20 cents than expected. By contrast, the other players exhibited pitch variations of < 4 cents (Philips) and < 1 cent (Apple), that is, below the threshold of human perception (5 cents).

The assertion of SanDisk technical support was thus found to be false.

However, the Clip’s pitch error is close to, but below, what might be regarded as the average pitch recognition value (25 cents). Hence, casual listeners, or those without musical training and/or without good pitch discrimination, are unlikely to be seriously affected by this issue. By contrast, serious listeners or those with musical training and/or good pitch discrimination will probably find the Sansa Clip (as well as other SanDisk products which exhibit this problem) highly unsatisfactory.

Introduction

Personal observation, as well as reports of other Sansa Clip users, indicate that music files played on this MP3 player are off-pitch [1]. For instance, DAVEK reported an error of -19.5 cents at 1000Hz on a file sampled at 44.1kHz on his Sansa Clip 8GB [2].

On the other hand, SanDisk technical support have stated that this problem was fixed by the latest firmware release (version 02.01.32 of February 2009) [3]. To test this assertion, I played the same audio file on two different Sansa Clips, on two other MP3 players, and on iTunes/PowerMac G4, and compared the results.

Players

I used the following MP3 players:

    Philips GoGear SA2115/37, f/w v4.16

    Apple iPod Classic 80G, v1.1.2 PC

    Sansa Clip 2GB, f/w V02.01.32A

    Sansa Clip 1GB, f/w V02.01.32A

Additionally, I also tested iTunes running on a PowerMac G4 under Mac OS X v10.3.9.

Source

I created a source file by ripping a track from an audio CD to AIFF, 16bits, 44.1kHz, stereo [4]. I edited the source with Sound Studio to insert two 1-second long 440Hz sine tones, exactly 360 seconds apart. Then I encoded the source to MP3 in 2 versions; version 1 with iTunes, 128k; version 2 with LAME v3.90.3, VBR (alt-preset-extreme).

Method

I copied the 2 MP3 files (128k and VBR) to the MP3 players from a PC in the usual manner [5].

I connected in turn each player’s headphone port to a Toshiba Satellite’s analogue audio-in port and recorded the player’s output with Audacity. I saved the result in Audacity’s format (AUP), and also exported it to WAV.

For each player, I measured in the WAV file the interval between the two 440Hz tones (which was exactly 360s in the source file).

Results

The results are listed below as xxx/yyy (zzz), where xxx is the interval for the 128k file in seconds, yyy for the VBR file, and zzz the percentage relative to 360, rounded to 2 decimals.

Note that only the GoGear generated different values for the two files.

    Philips GoGear … 360.528/359.361 (100.15/99.82)

    iPod Classic … 359.854/359.854 (99.96)

    iTunes … 360.002/360.002 (100)

    Sansa Clip 2GB … 364.084/364.084 (101.13)

    Sansa Clip 1GB … 364.085/364.085 (101.13)

In terms of pitch, these different intervals translate empirically [6] as follows (in absolute values):

    Philips GoGear … < 3 cents / < 4 cents

    iPod Classic … < 1 cent

    iTunes … << 1 cent

    Sansa Clip 2GB … 19 cents – 20 cents

    Sansa Clip 1GB … 19 cents – 20 cents

Conclusions

(1) The duration of 44.1kHz MP3 files played on the Sansa Clip is greater by >1% relative to the source, which is considerably higher than the variation exhibited by either the Philips GoGear or the iPod.

(2) This deviation corresponds to a pitch error of between 19 and 20 cents, which is an excellent match to the value reported by DAVEK [2].

(3) According to data quoted in Wikipedia, 5 cents can be taken as the threshold of human pitch perception, and most normal adults can “very reliably” recognise pitch differences of 25 cents or more [7]. The Sansa Clip’s pitch error is close to, but below the latter value. This would suggest that many users – in particular, casual listeners, those without musical training, or those with average and below average pitch discrimination – would be little affected by this issue. On the other hand, dedicated listeners or users with musical training and/or above average pitch discrimination would indeed perceive this issue as a problem.

(4) The fact that this kind of bug was allowed to occur in the final product would suggest that the Clip development team did not include anyone musically trained or with good pitch recognition. This would further suggest that the Clip (as well as other SanDisk products which exhibit this error) is not intended for customers with discriminating ears.

(5) According to SanDisk technical support,

“There is an available firmware update for your Sansa Clip. This should fix the issue with the pitch.” [3]

This assertion is clearly in flagrant contradiction with the facts; moreover, it is also in contradiction with statements of other representatives of SanDisk, eg,

“The SanDisk Firmware developers are aware of this issue and a fix is in the works for the Sansa Fuze and the Sansa Clip+. At this time due to development targeted on the Clip+, firmware for the original Clip is not currently under development. This may change in the future as time permits.” [8]

These contradictions raise doubts regarding either the expertise or the veracity of SanDisk technical support – or perhaps both.

File Availability

I can make available the files used to anyone who desires, provided a suitable arrangement can be made. The whole project is 2.19GB, of which the test files are 0.086GB, the recordings in WAV format 0.72GB, and the same in AUP format 1.41GB.

Notes

[1] See <http://en.wikipedia.org/wiki/Sansa_Clip#Music_Pitch_Bug>, retrieved 2009/09/19.

[2] See <http://forums.sandisk.com/sansa/board/message?board.id=sansafuse&message.id=18012#M18012>, retrieved 2009/09/19.

[3] SanDisk Technical Support, personal communication, 2009/09/17.

[4] Specifically, Angela Hewitt’s “Bach: The Well-Tempered Clavier”, Hyperion CDA67741/4 (2009), CD 2, Track 24, BWV 869: Fugue.

[5] Sansa Clip 1GB was connected in MSC mode, the other in MTP mode.

[6] That is, using Peak to change the pitch of the measured interval until its duration was close enough to 360s.

[7] See <http://en.wikipedia.org/wiki/Cent_(music)>, retrieved 2009/09/19.

[8] See <http://forums.sandisk.com/sansa/board/message?board.id=sansafuse&view=by_date_ascending&message.id=32947#M32947>, retrieved 2009/09/19.


[Last modified: 2009/09/20]

Thanks you for your in-depth and extensive test, Malix.

Quantifiable, verifiable, and unbiased proof of how bad the problem is indeed.  If I could give you rep points Malix, I would.

I second that Malix.  That’s a nice engineering analysis.  I hope someone at Sandisk is listening.

 

@slotmonsta wrote:

All,

 

At this time due to development targeted on the Clip+, firmware for the original Clip is not currently under development. This may change in the future as time permits.

 

 

Forum Admin

slotmonsta 

Message Edited by slotmonsta on 09-13-2009 02:46 PM
 

And so has there been any word back on this soltmonsta?  For a firmware upgrade and pitch fix for both the Clip and the Clip+?  Thank you.

   

Over the last few days there have been several meetings with Engineering, Marketing, and Product Management regarding the pitch issue some users have seen. Please see below a statement regarding the decision that was reached as a result of these meetings.

 

At SanDisk, our goal with our Sansa MP3 product line is to provide products which deliver a quality consumer experience at price points which are accessible to the majority of the population.  Our “value” positioning has served us well historically, although we acknowledge that occasionally our products do notlive up to some users’ expectations.  The issues raised on this Forum regarding sound fidelity are important to us however due to trade-off decisions that were made in engineering these products to deliver superior consumer value at what we believe are extremely attractive price points, our sound fidelity isn’t perfect.  We have re-evaluated the possibility of reducing the pitch variation and due to the engineering trade-offs the decision was made to stay with the current design. Very few listeners, however, have noticed or complained about it as an issue in actual practice.  For those who can detect sound differences with their naked ears during actual use and not via frequency analysis, our products may not be the best choice for them.

Forum Admin

slotmonsta 

Message Edited by slotmonsta on 09-24-2009 09:33 AM

What does that mean, exactly? Are you abandoning work on fixing the pitch bug on ALL devices now?

@slotmonsta wrote:

Over the last few days there have been several meetings withEngineering, Marketing, and Product Management regarding the pitch issue someusers have seen. Please see below a statement regarding the decision that wasreached as a result of these meetings.

 

At SanDisk, our goal with our Sansa MP3 product line is toprovide products which deliver a quality consumer experience at price pointswhich are accessible to the majority of the population.  Our “value” positioning has served us wellhistorically, although we acknowledge that occasionally our products do notlive up to some users’ expectations.  Theissues raised on this Forum regarding sound fidelity are important to ushowever due to trade-off decisions that were made in engineering these productsto deliver superior consumer value at what we believe are extremely attractiveprice points, our sound fidelity isn’t perfect.  We have re-evaluated the possibility of reducingthe pitch variation and due to the engineering trade-offs the decision was madeto stay with the current design. Very few listeners, however, have noticed orcomplained about it as an issue in actual practice.  For those who can detect sound differenceswith their naked ears during actual use and not via frequency analysis, ourproducts may not be the best choice for them.

 

Forum Admin

slotmonsta 

Slotmonsta,   does this mean Engineering couldn’t fix the issue by adjusting something with the code?  or that management decided not to spend the time to see if it was possible.    What are the engineering trade offs?

I would suggest to those who can hear the difference to re-encode things to 48k samples  instead of 44.1k…  

Thanks Slot for your reply.

niko

Message Edited by niko_sama on 09-23-2009 12:57 PM

" We have re-evaluated the possibility of reducingthe pitch variation and due to the engineering trade-offs the decision was madeto stay with the current design."

I guess I’m unclear, now, as to whether the pitch problem is due to hardware or firmware.  What could be the “engineering trade-offs” involved in upgrading the firmware?

@slotmonsta wrote:

Over the last few days there have been several meetings withEngineering, Marketing, and Product Management regarding the pitch issue someusers have seen. Please see below a statement regarding the decision that wasreached as a result of these meetings.

 

At SanDisk, our goal with our Sansa MP3 product line is toprovide products which deliver a quality consumer experience at price pointswhich are accessible to the majority of the population.  Our “value” positioning has served us wellhistorically, although we acknowledge that occasionally our products do notlive up to some users’ expectations.  Theissues raised on this Forum regarding sound fidelity are important to ushowever due to trade-off decisions that were made in engineering these productsto deliver superior consumer value at what we believe are extremely attractiveprice points, our sound fidelity isn’t perfect.  We have re-evaluated the possibility of reducingthe pitch variation and due to the engineering trade-offs the decision was madeto stay with the current design. Very few listeners, however, have noticed orcomplained about it as an issue in actual practice.  For those who can detect sound differenceswith their naked ears during actual use and not via frequency analysis, ourproducts may not be the best choice for them.

 

Forum Admin

slotmonsta 

I haven’t really noticed the pitch problem myself but there are going to be A LOT of unhappy people on the forums lol.

Amazing:  already posted on Wikipedia.  Does Wikipedia not sleep?   ;)   http://en.wikipedia.org/wiki/SanDisk_Sansa

@slotmonsta wrote:

Over the last few days there have been several meetings withEngineering, Marketing, and Product Management regarding the pitch issue someusers have seen. Please see below a statement regarding the decision that wasreached as a result of these meetings.

 

At SanDisk, our goal with our Sansa MP3 product line is toprovide products which deliver a quality consumer experience at price pointswhich are accessible to the majority of the population.  Our “value” positioning has served us wellhistorically, although we acknowledge that occasionally our products do notlive up to some users’ expectations.  Theissues raised on this Forum regarding sound fidelity are important to ushowever due to trade-off decisions that were made in engineering these productsto deliver superior consumer value at what we believe are extremely attractiveprice points, our sound fidelity isn’t perfect.  We have re-evaluated the possibility of reducingthe pitch variation and due to the engineering trade-offs the decision was madeto stay with the current design. Very few listeners, however, have noticed orcomplained about it as an issue in actual practice.  For those who can detect sound differenceswith their naked ears during actual use and not via frequency analysis, ourproducts may not be the best choice for them.

 

Forum Admin

slotmonsta 

If you have noticed this message has been posted in several threads relating to this issue. Unfortunately during the copy and pasting process I did not notice the errors in this particular post. I have edited the original post to correct the grammatical errors and I thank you for pointing them out. 

Forum Admin

slotmonsta 

This is so classic.

Like the sound problem, I didn’t notice at first the absolutely awful composition above – and how ironic that they’d be SO sloppy in a lawyer-speak drop-dead-letter crafted for the purpose of explaining how they are content to deliver cruddy audio rendition.

And, like the sound problem, once you DO notice it, you can’t help NOT noticing it.

(If this irony point ends up in wikipedia too (I wouldn’t be surprised), don’t blame ME for it. I don’t edit there.  Tried a few times, minor things, to correct some errors, notHing at all out of line, but they were inevitably reversed by the cult there, so screw it.)

But, I guess that having yourwords posted in a waythat makes themrun together is OK in a productthat doesn’t really pretendto be anything better than runof the mill.

The only real problem is that actual “run of the mill”  stuff does NOT exhibit this really sad failure.

 So explain to me again how this amounts to “superior consumer value at what we believe are extremely attractiveprice [sic] points”?  A no-name “MP4 Player” that costs much less than the Sansa gear CAN carry a tune – so what is the motivation for paying a premium for the Sansa brand?

By the way, I find it curious that after my previous post, I can no longer access this forum using that computer.  I kept checking for several hours and each time the connection would hang.  I thought they were probably having server problems – then I wondered if they’d taken the forum offline as a hamfisted method of dealing with the dissent.  I then checked some other forums to see if anyone was discussing the forum being offline, and found no such commentary.  So, I logged in with a different computer, and here I am, posting. Hopefully this post will go through.

I still cannot access the forum from that other browser – I can reach other sites from there, but THIS site is a dead end.

So, am I being partially blocked?  Are OTHER posters being blocked too, if that’s the case?  I hope not, but if so, that sort of heavyhanded nonsense can only backfire.

Message Edited by slotmonsta on 09-24-2009 04:30 PM

The legendary Sansa quality shines through yet again – theirforum software rejected my color highlighting – which I applied using the color button in the forumeditor – as having invalid HTML, which it did me the favor of stripping out.  But, when I RE-applied it, it accepted it.  I guessthis is what you’d call a “work-around” – one of those nuisances we have to tolerate when we wishto use a Sansa product?

PS – I took the liberty of (I mean, “I tookthe liberty of”) proof-reading my paragraph above, and editing it to reflect the Sansa standard of quality.

I would not object if everyone began doing Sansa-style word coupling when posting about this issue.  It is the sort of thing that helps point out the absurdity of their actions.

And to think, WE are the ones who not too long ago were the most fervent “unofficial salesmen” for this stuff – and, the ONLY reason we’re still hanging around is because we DO like the products, and only want to have glaring design defects repaired.

And any time they try telling me that the design defects are NOT glaring, I will simply provide a link to the “red letter edition” of their lawyer letter – and remind them that THAT series of errors, while certainly glaring, somehow eluded notice until a few minutes ago.

This post is simply distasteful and obviously nothing more than a direct insult. This will be your one and only warning so please take some time to read our forum rules and guidelines. If your posts continue in this nature with this attitude you will be ban from accessing this forum. 

 

Forum Admin

slotmonsta 

Message Edited by slotmonsta on 09-24-2009 09:56 AM

@gwk1967 wrote:
What does that mean, exactly? Are you abandoning work on fixing the pitch bug on ALL devices now?

Well, look at his exact words:

"For those who can detect sound differenceswith [sic] their naked ears during actual use and not via frequency analysis, ourproducts [sic] may not be the best choice for them."

Note that he used the plural of “product” rather than say “Clip” – I think it would take a major “whistling past the graveyard” effort to see this in any way other than as it reads.

When a company talls me to avoid their products – without specifying any particular model – if I care about quality, what grade of naive would I need to be to not take them at their word?

I realize the paradox here – we DID take them at their word when they said that the fix was there, ready to be dropped into the next firmware release, so we should just be patient.  But, I think it’s just an “apparent paradox” – I think we’re OK if we take them at their word when they are delivering BAD news.  It only seems to be when they are making “campaign promises” that we should not get our hopes up.  (I think of it in those terms, having been trained by the best politicians in the land not to believe the promises they mke when trying to gain my support.)

Message Edited by PickMorel on 09-24-2009 05:51 AM

Since this issue can be fixed by resampling your _ files _ to 48kHz, it’s probably also possible to implement a fixed-point software resampler in the player. This would trade-off battery time and a decrease in SNR for correct sampling of the audio.

@kohlrabi wrote:
Since this issue can be fixed by resampling your _ files _ to 48kHz, it’s probably also possible to implement a fixed-point software resampler in the player. This would trade-off battery time and a decrease in SNR for correct sampling of the audio.

anytime u make the processor work harder,  u usually have to bump up the clock … and when u bump up the clock… u end up with less battery…      not saying i know for sure but the battery test vs bitrate/encoders do show dramatic differences in battery live…    when they clock up the cpu. 

What bothers me after reading everything about the pitch issue, is at one time  Sansa Engineers looked at the issue and had reduced the pitch deviation to less than a few cents.    That tells me it is fixable.   And why did it never get released?  Is it because Sansa Engineering and Management can’t get their facts straight?   Were the right people in that meeting ?   The ones who worked on it before or were those engineers part of the group that got displaced?   I really think Sansa took the easy way out on this, and I hope customers don’t allow that to happen.    If there was a fix what happened to it ?    Sansafix gave detailed information about how much of an improvement they could accomplish, I really dough he would have given this detailed info if they had not done the work.  

I hope Sansa reconsiders their decision on this.  It’s making the management there look uninformed and not in touch with the history of the issue and the work that was already done.   Either that or they need to provide the exact details of the engineering trade offs.   It wouldn’t be the first time management made an uninformed decision in hopes that the issue would just disappear and underestimate how loud thier customers can be.   If you had the fix before what happened to it?  Why Sansa?

Please bang the drums till we get more details from Sansa regarding this issue.  And don’t let them take the easy way out.  If Sansa wants to make a second rate product they are making the right decisions about this issue.  Please Sansa don’t take the easy way out when you already had a fix for this.

Reference:  http://forums.sandisk.com/sansa/board/message?board.id=sansafuse&message.id=18197

Sansafix wrote:

 

All,

Good news, We have reduced the pitch error by one order of magnitude with little to no effect on battery life (<3%).

The optimization will be included in the next firmware release due out this quarter.

We have optimized for 44Khz and the pitch error is < 0.14%

For all other samples rates its <0.18%

Message Edited by sansafix on 02-19-2009 08:37 AM
 

Message Edited by niko_sama on 09-24-2009 09:41 AM

Message Edited by niko_sama on 09-24-2009 09:47 AM

This is all so odd–given the above, why wasn’t the fix made?  Perhaps a concern over the lower battery duration, from a marketing perspective?

Having said that, it is unclear to me that SanDisk will reverse itself–its various committees already convened and made a determination, presumably with, among other things, the above information in hand.   The company would need to be given a reason and a motivation to go the other way.

@miikerman wrote:

This is all so odd–given the above, why wasn’t the fix made?  Perhaps a concern over the lower battery duration, from a marketing perspective?

 

Having said that, it is unclear to me that SanDisk will reverse itself–its various committees already convened and made a determination, presumably with, among other things, the above information in hand.   The company would need to be given a reason and a motivation to go the other way.

i dough thats true.    They probably didn’t have the right people in the meeting or they weren’t asking the right questions.     Maybe they were too busy thinking about lunch or golf on friday.       If the above is copyied to the right level,   vp engineering directors etc.   maybe the right questions will be asked. and answered.

committees?  lol wtf…  its was probably one engineering director or manager asking a couple engineers and maybe those engineers present didn’t have the answers.    Or very possible it was just a managers response in email to his boss saying the issue was not important and they were maxed out on on other issues / work load.  

If Sansafix said there was no noticable effect on battery life, i believe what he said.  

If i was the director I would find out what happened to the fix.   And it would get done.    It would be nice to copy the VP’s and Directors some of the history on this issue…  its possible we know more about it than they do.

@miikerman wrote:

This is all so odd–given the above, why wasn’t the fix made?  Perhaps a concern over the lower battery duration, from a marketing perspective?

 

 

Could be the person(s) that wrote the fix got laid off.

 As far as being a “value” brand, they do have to pick what things they invest in.  But I can’t help notice that of the other players people tested for comparison, none came anywhere near the error displayed by Clip and Fuze.  Do they really want the reputation of being “worst on the market by a long shot” wrt pitch?  

I’m curious to know if ANY other player at any price point has equivalent or greater error.  I don’t know what other brand uses AMS chips or if that’s a key factor.

Has anyone tested Fuze or Clip with Rockbox for this?  

 Speaking of Rockbox, a new release just came out with enhancements to pitch correction, and a tuner plugin that displays the note and cents error of the note coming in the microphone.