Comprehensive USB Bug List

[Taylor Sherman]

My 2017 restarts the song every single time I exit and return to the car. Like, I can get out and go over and open the passenger side door, and it'll restart the song.

The "shuffle" is also absolutely terrible. I've never encountered a great shuffle algorithm before, but this is the worst. Every day during my commute it will land on a song I heard the day before or the day before that, and the sequence of 10-15 songs after that (as I skip through them) is exactly the same. It'll finally land on something new, and I'll let that play. From there on I hear a couple new songs, then it's usually back to something it played the day before, and usually then an entire sequence. Instead of "shuffle" it should be called "deja vu".

I have 1,000 songs on my drive.

There are hundreds of songs it has never played, and it regularly repeats the same 2 or 3 10-15 song sequences. I've been using the same drive for a couple months now, for over an hour per day.

I understand that "make autopilot kinda work decently again" is a high priority, but this is really just sad. Every single day my car just demonstrates incompetence in a very obvious way.

 

[BertL]

^ Yes, the "begins at the beginning" is sadly an issue we've had for a long time -- not a new thing to 2017's or in recent firmware. You may find MP operates differently with some formats vs others (FLAC e.g. always has that problem from my experience, but at least one owner reported a while back that MP3 didn't have the same symptom. I've not tried to confirm or narrow all that down as Tesla hasn't documented which formats are or are not supported to begin with, so it's hard to then go at them with what is or isn't working. )

It's a similar thing with "shuffle" -- not a new issue at all ...and just to get everyone going I'd offer that I have never found anything in Tesla documentation that has ever stated what the universal crossed-arrow symbol is supposed to actually do in MP -- so we really don't know for a fact what Tesla intends this function to provide. IMHO, what we have been provided acts more like a dysfunctional "repeat" function as you describe, than a more sophisticated "shuffle" -- which if it were like nearly every other player, would randomly work it's way through 100% of the tracks in the set, playing each only once, until it either stops when it has played them all, or begins a new shuffle of the same set again.

SIDE NOTE: Yesterday I was starting to prepare documentation for when I take my MS in for service to report the "Random isn't Random" USB bug. I wanted to count the number of tracks that MP repeats, as my perception is that number is somewhere between 8-12 if I don't touch anything, no matter if I try to play all tracks (7K+), or select one of my 10 genre and have MP randomly play within those as I usually do. Instead of waiting for all tracks to play and begin the repeat to get my count, I used "forward" on my steering wheel to jump between songs to get what I thought would be a quick count of the repeated sequence. Maybe I was dreaming, but I found if I used "forward" with "random" on -- just clicking quickly through songs to count unique album art/titles vs listening to the media itself, the songs didn't seem to repeat -- at least as often. (I kept counting to 20 songs with no repeats which exceeds any sequence I remember.) It was strange. IDK if using "forward" perhaps influences how repeat does or does not work, or maybe it was just what happens with this buggy code if you forward through tracks without really listening to them. Does anyone have any observations in that regard?​

 

[Dameon]

Filesystem: seen with both fat32 and ext4.

FS organization: top_level_folder/{artist}/{album}/{mp3s}

Tag generation, done with Linux. Note, in an effort to establish a working baseline, I am not even attempting to handle compilation albums:

Create copy of the original mp3 files.
Sanitize all artist folder names (replace diacriticals, spaces, punctuation)
Sanitize all album folder names (replace diacriticals, spaces, punctuation)
Sanitize track names (replace diacriticals, spaces, punctuation, add/normalize leading track numbers 01-track_name.mp3, etc...)
Extract artwork from each track and convert to 300x300 jpeg.
Clear all tags on all tracks.
Set the following tags:

Artist (TPE1) - to the name of the artist folder
Album Artist (TPE2) - to the name of the artist folder
Album (TALB) - the name of the album folder
Title (TIT2) - the name of the track
Track (TRCK) - the leading two digits of the track name (01, 02, ... 17, etc)
Image (APIC) - the 300x300 jpeg

Set full access permissions on all folders and files (0777).

Your process is the closest to mine as any I've seen posted here. One addition in my case: I always use ID3 v1 and that started with the issues in my other vehicles. I have a 2009 Z06 and a 2009 Escalade (totally different infotainment systems for some reason) and they behave very differently with MP3 songs. Although the symptoms were different, the solution was the same: ID3 v1. So I make a copy of the songs I want to play in each vehicle and strip everything out and tag using ID3 v1.

My difference to most is that I don't put my entire 100K song collection on a USB drive. I usually buy the $0-10 Newegg weekly special USB drives (8-16GB mostly) and use them for the vehicles. I have a couple hundred songs on each drive which is usually way less than 50% capacity. I don't spend a lot of time in my vehicles, so that many songs can last a week without repeat. Works in my case.

I'm old enough to remember when CD players first arrived in cars. They pretty much sucked. Most required a big square "caddy" to install your CD into. Kinda made it like a big 3.5" floppy. The only nice thing was that it offered a level of protection as the discs bounced around in your console. I spent the money for individual caddies for each CD. Back then, every time you cycled the car (turn it off and back on), the CD would just start over. If you were lazy, you heard track one over and over again. Plus, even with the improved stability of the caddy, they skipped horribly. Eventually they improved the head units to eliminate the caddy and improved skipping, but the best they could do on "resume" was to start the last playing track over again. Finally, they improved to the point where it started playing exactly where it left off. I had a 1988 RX7 Turbo II with a factory head unit that resumed precisely where it left off. 1988. Nice to see Tesla's media player is catching up to 1988 technology.

 

[amosnomor]

Your process is the closest to mine as any I've seen posted here. One addition in my case: I always use ID3 v1 and that started with the issues in my other vehicles. I have a 2009 Z06 and a 2009 Escalade (totally different infotainment systems for some reason) and they behave very differently with MP3 songs. Although the symptoms were different, the solution was the same: ID3 v1. So I make a copy of the songs I want to play in each vehicle and strip everything out and tag using ID3 v1.

When I follow my recipe properly, I haven't seen an issue with the v2 tags. I don't remember the exact numbers, but I have around 130 artists, 220 albums, and 2200 songs. The nice thing with the v2 tags is being able to add the image.
[/QUOTE]

 

[outie]

One year in and my model X still does not resume mp3 playback when I get in the car or it randomly starts from my phone using Bluetooth. Ridiculous.

 

[BertL]

Tracks are now correctly restarting where they left off, every time I reenter my MS.
I've had one week of zero rescans or loading errors -- which has not happened in a very long time.
...and it's not because of a Tesla firmware change, but because of a choice I decided to make.

BACKGROUND

While I swore off doing another round of complete USB MP testing as I've reported here several times, I remain convinced that Tesla's poor management of constrained memory conditions is a primary reason some of us have more or less USB playback problems than others. It's why for more than a year, I have gone to such extremes extracting less than 7K tracks from my curated master music library of primarily ALAC (AAC Lossless) tracks into FLAC. I then let my Mac spend hours processing those tracks to remove extraneous tagging, standardize Album Art into JPG-only that is roughly the same size as the largest image displayed anywhere within the MP UI, and apply a few other transformations to resolve lack of MP functionality where I am able. Except for testing, I've used FLAC for all my MP tracks since I took delivery of my MS nearly 2 years ago: I had the space on my large USB/SSD devices, conversion time from ALAC is super-fast, and I get the best possible quality going into MP that I can provide. The result isn't perfect as I still have my Tesla Twilight Zone moments, but my problems have been minimized despite long-outstanding MP bugs and missing features.

We've discussed here before that ID3 tagging has created a number of variants over time, but those variants also multiply as different codecs are used -- e.g. MP3 tagging isn't 100% the same as what happens with FLAC for the same content. Album Art also has unlimited possibilities on sizes, image compression formats, etc. which I've discussed in posts before. I personally have not found any issue using any ID3 variation up to ID3V2.4 in my MS, but I believe we've proven in this thread more than once, that ill-structured tagging can make MP go berserk or cause more intermittent failures. The problem with that is the damage could have been done years ago inside your source tracks without you even realizing it, and the error may not be apparent in all players. To me, ID3 tags just open up a lot of possible places where any media player can fail if it's not designed assuming an owner may give it tracks with errors to begin with.​

MY LATEST AH-HA MOMENT

I have not been able to get @f-stop's post a few weeks ago out of my head suggesting filetype (aka "codec") is yet another variable in the equation. Thanks @f-stop! Beyond specific testing I have documented here in the past, I have had nothing but 100% FLAC in my MS for daily use ...but what if I could improve my not-so-perfect MP environment even further?​

THE NET

After a quick test to ensure MP would handle the output, I reran my dBpoweramp conversion process against my same 6,708 ALAC tracks with all the same transformation options I've been using, except instead of FLAC, I encoded all tracks as MP3 VBR -V0 (aka, Variable Bit Rate with highest quality).

After one week, the result is sound quality where I can't tell a difference from my FLAC files -- especially with road noise -- and now whenever I get into my MS, MP resumes playing the same track from the same spot it left off, every single time without any interaction on my part. No pause, no funny business. It has worked 100% of the time with short errands, deeper sleep, and through a scheduled charging cycle. I also have had no unexpected rescans or loading errors in this same week -- although given how intermittent both of these problems are, I am not ready to declare success on either of these concerns.​

CONCLUSION

As some of us have thought for a long time, MP3 seems to be more of a sweet spot for how MP best operates. I am now personally convinced that is the case -- at least with firmware 2017.32 9ea02cb I am currently running. I still believe the Tesla Engineers that design and test MP are mostly using small numbers of MP3 files, with sparse ID3 tagging and probably a single piece of art for a complete album, vs what some of us "music enthusiasts" have with multi-disc albums, compilations, highly curated and extensive ID3 tagging, perhaps different art for different tracks, and a desire to use lossless codecs with a large number of tracks from our collection.

Converting to MP3 VBR has become my new standard. Some of you can debate if you believe MP3 VBR is of sufficient quality in your MS. (Note I am not suggesting use of lesser CBR or ABR targets. I admit I'm surprised Tesla's code supports MP3 VBR as it's more complex and not something every MP3 player will handle, but it works -- probably because of tag-along code in open source libraries.) While it does not require any effort on my part except to pay the energy bill, the conversion from a lossless format to MP3 VBR, with the rest of the dBpoweramp transformation I do to hopefully reduce CID memory usage is highly compute intensive. For my 6,708 ALAC tracks, once I have a copy extracted from my iTunes Library, it takes a whopping 53 hours for my 4.0GHz quad core to accomplish the task.​

- - - - - - - - - -

BERT'S dBpoweramp MEDIA PLAYER 8.1 WORKAROUND 09-17-2017

It's been nearly a year since I posted the last version of this nearly 1,000 posts upthread. For those that didn't catch that and care about an overview of the transformations I make on my master music library files for use in my Tesla MS, and the basic steps I use with dBpoweramp -- either the Mac or Windows version -- to automate most of the effort, here you go. I have no affiliation with the products themselves.

My settings accomplish the following:

• Use a temporary directory containing copies of the source audio files I want to have converted for use in my MS
• Puts the contents of ALBUMARTIST into TRACKARTIST on every track
• Changes all dBpoweramp-supported Album Art to JPG, and reduces it’s size to a maximum 300x300
• Changes TRACKTITLE to “DISCNUMBER-TRACKNUMBER TRACKTITLE” on every track
• Performs Volume Normalize analysis against each track, physically changing the contents of the track copy so the Tesla driver does not have to fiddle as much with the volume level as soft/loud parts of a track, or across tracks of disparate albums, are played.
  • This performs a methodical analysis of each track’s volume, looking across 6000ms windows of time and adjusting the volume up/down just as you would trying to keep relative playback volume closer to the same during very soft and loud parts of the track. This seems to most closely emulate what I know as ASL or Automatic Sound Leveling options that are done real-time in other vehicles I’ve owned, except here it’s physically preprocessed into a copy of every track for playback later in my Tesla. I'm a pretty critical listener, and do not find this processing takes away from my content enjoyment or introduces some unnatural annoyance. The side benefit is the relative volume level, as you switch from USB Media Source to e.g. FM, is pretty close to being the same, which in my experience isn't what happens otherwise.
  • THE BIG DRAWBACK is this takes an extraordinary amount of hands-off compute time to accomplish. It increases the conversion time by 5-16X. If you don't want to use it, just delete the Volume Normalize line from the DSP Effects once you load my settings.

• Converts every source audio file of varying encoding types to MP3 VBR -V 0 Quality
• Places the newly created version of the MP3 files into a 2-level directory structure: One folder off of the root for each album with related tracks inside, correctly handling track sequencing and multi-disc albums (assuming you have them tagged correctly in your master library to being with).

The process

• Extract files you want to convert from your master library to a temporary directory on your Mac or PC. (I use "Export for iTunes" available on the Mac App Store.)
• Start dBpoweramp Music Converter
  • Point it at that temporary directory you just made and filled with source tracks
  • Click the Convert icon
  • On the conversion menu:
    • Encoding: MP3 (Lame)
      • Target: Quality (VBR)
      • Move the slider all the way to the right (high quality / larger file)
      • Encoding: Normal
    • Output To: Select "Edit Dynamic Naming" in the dropdown
      • Base Location: Specify the folder you want to put the converted tracks into
      • Dynamic Naming: [album] - [album artist]/[disc]-[track] [title]
        • If you have a real need and think through the implications, change that directory structure as you desire for use in Folder View, noting pros and cons of too shallow and too deep folder structures discussed elsewhere in this thread, and implications that may have on Tesla's other Views if you choose to use them
    • Unzip, then Load the attached DSP Effects / Actions. (Don't be concerned the filename says "FLAC". Your encoding selection above is what counts.)
    • Click Convert and you’re on your way. Let your computer do it’s thing and go do something more productive.

I suggest you give the process a try with a small number of tracks from various albums to see how it works in your MS. Make changes to fit your own needs and desire, then let the Converter rip against your larger set of tracks. Oh, and if you change the settings, remember to save the DSP Effects so you don't have to go through the trial-and-error again one day.​

Enjoy!

 

[aesculus]

Converting to MP3 VBR has become my new standard. Some of you can debate if you believe MP3 VBR is of sufficient quality in your MS.

This is probably OK for an MS, but my MX could never tolerate such a compromise.

Thanks for the tip though. Oddly I have the same experience mostly but with no resume within track. It only starts the same track over again. And there are many times when it just does not remember anything but the track and does not seem to realize it was playing an album at the time. Once the track was over that was it.

 

[msnow]

Tracks are now correctly restarting where they left off, every time I reenter my MS.
I've had one week of zero rescans or loading errors -- which has not happened in a very long time.
...and it's not because of a Tesla firmware change, but because of a choice I decided to make.

BACKGROUND

While I swore off doing another round of complete USB MP testing as I've reported here several times, I remain convinced that Tesla's poor management of constrained memory conditions is a primary reason some of us have more or less USB playback problems than others. It's why for more than a year, I have gone to such extremes extracting less than 7K tracks from my curated master music library of primarily ALAC (AAC Lossless) tracks into FLAC. I then let my Mac spend hours processing those tracks to remove extraneous tagging, standardize Album Art into JPG-only that is roughly the same size as the largest image displayed anywhere within the MP UI, and apply a few other transformations to resolve lack of MP functionality where I am able. Except for testing, I've used FLAC for all my MP tracks since I took delivery of my MS nearly 2 years ago: I had the space on my large USB/SSD devices, conversion time from ALAC is super-fast, and I get the best possible quality going into MP that I can provide. The result isn't perfect as I still have my Tesla Twilight Zone moments, but my problems have been minimized despite long-outstanding MP bugs and missing features.

We've discussed here before that ID3 tagging has created a number of variants over time, but those variants also multiply as different codecs are used -- e.g. MP3 tagging isn't 100% the same as what happens with FLAC for the same content. Album Art also has unlimited possibilities on sizes, image compression formats, etc. which I've discussed in posts before. I personally have not found any issue using any ID3 variation up to ID3V2.4 in my MS, but I believe we've proven in this thread more than once, that ill-structured tagging can make MP go berserk or cause more intermittent failures. The problem with that is the damage could have been done years ago inside your source tracks without you even realizing it, and the error may not be apparent in all players. To me, ID3 tags just open up a lot of possible places where any media player can fail if it's not designed assuming an owner may give it tracks with errors to begin with.​

MY LATEST AH-HA MOMENT

I have not been able to get @f-stop's post a few weeks ago out of my head suggesting filetype (aka "codec") is yet another variable in the equation. Thanks @f-stop! Beyond specific testing I have documented here in the past, I have had nothing but 100% FLAC in my MS for daily use ...but what if I could improve my not-so-perfect MP environment even further?​

THE NET

After a quick test to ensure MP would handle the output, I reran my dBpoweramp conversion process against my same 6,708 ALAC tracks with all the same transformation options I've been using, except instead of FLAC, I encoded all tracks as MP3 VBR -V0 (aka, Variable Bit Rate with highest quality).

After one week, the result is sound quality where I can't tell a difference from my FLAC files -- especially with road noise -- and now whenever I get into my MS, MP resumes playing the same track from the same spot it left off, every single time without any interaction on my part. No pause, no funny business. It has worked 100% of the time with short errands, deeper sleep, and through a scheduled charging cycle. I also have had no unexpected rescans or loading errors in this same week -- although given how intermittent both of these problems are, I am not ready to declare success on either of these concerns.​

CONCLUSION

As some of us have thought for a long time, MP3 seems to be more of a sweet spot for how MP best operates. I am now personally convinced that is the case -- at least with firmware 2017.32 9ea02cb I am currently running. I still believe the Tesla Engineers that design and test MP are mostly using small numbers of MP3 files, with sparse ID3 tagging and probably a single piece of art for a complete album, vs what some of us "music enthusiasts" have with multi-disc albums, compilations, highly curated and extensive ID3 tagging, perhaps different art for different tracks, and a desire to use lossless codecs with a large number of tracks from our collection.

Converting to MP3 VBR has become my new standard. Some of you can debate if you believe MP3 VBR is of sufficient quality in your MS. (Note I am not suggesting use of lesser CBR or ABR targets. I admit I'm surprised Tesla's code supports MP3 VBR as it's more complex and not something every MP3 player will handle, but it works -- probably because of tag-along code in open source libraries.) While it does not require any effort on my part except to pay the energy bill, the conversion from a lossless format to MP3 VBR, with the rest of the dBpoweramp transformation I do to hopefully reduce CID memory usage is highly compute intensive. For my 6,708 ALAC tracks, once I have a copy extracted from my iTunes Library, it takes a whopping 53 hours for my 4.0GHz quad core to accomplish the task.​

- - - - - - - - - -

BERT'S dBpoweramp MEDIA PLAYER 8.1 WORKAROUND 09-17-2017

It's been nearly a year since I posted the last version of this nearly 1,000 posts upthread. For those that didn't catch that and care about an overview of the transformations I make on my master music library files for use in my Tesla MS, and the basic steps I use with dBpoweramp -- either the Mac or Windows version -- to automate most of the effort, here you go. I have no affiliation with the products themselves.

My settings accomplish the following:

• Use a temporary directory containing copies of the source audio files I want to have converted for use in my MS
• Puts the contents of ALBUMARTIST into TRACKARTIST on every track
• Changes all dBpoweramp-supported Album Art to JPG, and reduces it’s size to a maximum 300x300
• Changes TRACKTITLE to “DISCNUMBER-TRACKNUMBER TRACKTITLE” on every track
• Performs Volume Normalize analysis against each track, physically changing the contents of the track copy so the Tesla driver does not have to fiddle as much with the volume level as soft/loud parts of a track, or across tracks of disparate albums, are played.
  • This performs a methodical analysis of each track’s volume, looking across 6000ms windows of time and adjusting the volume up/down just as you would trying to keep relative playback volume closer to the same during very soft and loud parts of the track. This seems to most closely emulate what I know as ASL or Automatic Sound Leveling options that are done real-time in other vehicles I’ve owned, except here it’s physically preprocessed into a copy of every track for playback later in my Tesla. I'm a pretty critical listener, and do not find this processing takes away from my content enjoyment or introduces some unnatural annoyance. The side benefit is the relative volume level, as you switch from USB Media Source to e.g. FM, is pretty close to being the same, which in my experience isn't what happens otherwise.
  • THE BIG DRAWBACK is this takes an extraordinary amount of hands-off compute time to accomplish. It increases the conversion time by 5-16X. If you don't want to use it, just delete the Volume Normalize line from the DSP Effects once you load my settings.

• Converts every source audio file of varying encoding types to MP3 VBR -V 0 Quality
• Places the newly created version of the MP3 files into a 2-level directory structure: One folder off of the root for each album with related tracks inside, correctly handling track sequencing and multi-disc albums (assuming you have them tagged correctly in your master library to being with).

The process

• Extract files you want to convert from your master library to a temporary directory on your Mac or PC. (I use "Export for iTunes" available on the Mac App Store.)
• Start dBpoweramp Music Converter
  • Point it at that temporary directory you just made and filled with source tracks
  • Click the Convert icon
  • On the conversion menu:
    • Encoding: MP3 (Lame)
      • Target: Quality (VBR)
      • Move the slider all the way to the right (high quality / larger file)
      • Encoding: Normal
    • Output To: Select "Edit Dynamic Naming" in the dropdown
      • Base Location: Specify the folder you want to put the converted tracks into
      • Dynamic Naming: [album] - [album artist]/[disc]-[track] [title]
        • If you have a real need and think through the implications, change that directory structure as you desire for use in Folder View, noting pros and cons of too shallow and too deep folder structures discussed elsewhere in this thread, and implications that may have on Tesla's other Views if you choose to use them
    • Unzip, then Load the attached DSP Effects / Actions. (Don't be concerned the filename says "FLAC". Your encoding selection above is what counts.)
    • Click Convert and you’re on your way. Let your computer do it’s thing and go do something more productive.

I suggest you give the process a try with a small number of tracks from various albums to see how it works in your MS. Make changes to fit your own needs and desire, then let the Converter rip against your larger set of tracks. Oh, and if you change the settings, remember to save the DSP Effects so you don't have to go through the trial-and-error again one day.​

Enjoy!

Great job Bert, really appreciate the detail.

 

[f-stop]

Tracks are now correctly restarting where they left off, every time I reenter my MS.
I've had one week of zero rescans or loading errors -- which has not happened in a very long time.
...and it's not because of a Tesla firmware change, but because of a choice I decided to make.
[...]

Informative post @BertL, worthy of a close re-read and some experimentation.

just a quick thought... Shouldn't resetting the CID clear memory? I can 100% reproduce the loading errors on a small newly formatted USB drive containing fewer than a dozen AAC/M4A files. I thought I did a reset before each of my previous tests, but I'm away and without access to my MS and can't recheck to confirm right now.

 

[BertL]

Informative post @BertL, worthy of a close re-read and some experimentation.

just a quick thought... Shouldn't resetting the CID clear memory? I can 100% reproduce the loading errors on a small newly formatted USB drive containing fewer than a dozen AAC/M4A files. I thought I did a reset before each of my previous tests, but I'm away and without access to my MS and can't recheck to confirm right now.

Thx.

Doing a full reboot when I first took delivery used to clear CID memory -- at least everything related to USB music. I don't believe it probably does any longer. I don't remember which firmware release (have to search upthread or in another before this one started for a post by me), but I reported, as did some others confirm, that not everything started being cleared at one point... e.g. when I first noticed the change, ODDLY, Album Art began sticking around after the full reboot -- to the point you could change the physical art inside a track, remove the device, reboot, reinsert it, but if that track filename and directory structure remained 100% the same, MP would assume what it had in memory was still the right version and not look at the actual tag inside the track to rebuild its version -- just like it was previously doing on what I call a "rescan". Things may have changed (again), but without any real MP USB documentation as to what is supported in the first place, let alone what is changed release-to-release, or one of us doing extensive testing again, it's anyone's guess.

My personal guess after this latest Ah-Ha, is Tesla likely designs and tests MP with MP3 files. We know that their Infotainment testing isn't the best compared to what Tesla likely does for Safety and Autopilot systems, but I suspect what happens with non-MP3 filetypes (codecs), is catch-as-catch-can based on how the Tesla-specific code and generic libraries supplying the other codecs and generally-available MP with USB handling code that Tesla may not be spending enough time testing. I wouldn't doubt at all that an owner who inserts the same tracks that are perhaps mostly FLAC with a few MP3 or something else, may act differently than if the same tracks were all MP3. It's why I made the specific effort after reading your post to recode the exact same source tracks I have been using for months in my MS as FLAC, into MP3 VBR to reduce a variable.

The additional problem remains, making even my latest test inconclusive, is that as owners, we never know what memory is really available to MP USB since Tesla has never documented that for owners, or seems to have set maximums in what they allow (e.g. too many tracks with too much tag data for your specific situation, and you'll find no error message, just a USB scan process that infinitely slows down and never completes, with perhaps other parts of the CId slowing down and maybe an automatic reboot -- I've had all the variations)... How the CID is managing your specific Nav History, if one is using line mode or hybrid views, how many Nav tiles and what level of detail are being maintained, number of Superchargers on your possible route, number of phone directory entries you have and the length of the related data in each one that Infotainment maintains, number of MP favorites, to what degree AP is downloading and uploading variable amounts of info to the mother ship, staging of new firmware releases, etc -- all eat up part of the 2GB CID memory before MP USB begins to pull in what it needs to map against any USB device, and then there are the working areas needed to just keep everything running... All of that is something we as owners just can't understand in sufficient detail to accurately reproduce some of these problems even on our own vehicles Hour-to-Hour, and IMHO why some owners see issues and others don't.

It comes down to Tesla just needs to tighten-up this whole area of their code to improve stability -- but Elon does not see it as important because not enough owners are screaming, reporting it via normal Tesla methods, or perhaps hitting him at the right time on Twitter when he's in a receptive mood. For more than a year, I've suggested the fastest way for Tesla to start improving this is to just document what MP USB is designed to support: Codecs, Bit rates, File naming and max directory depth, supported tags and related detail, AND a max number of tracks on a single USB a device. Some owners that happen to be getting away with more today (one owner reported over 15K tracks), will be sad to see a limitation imposed upon them. Others like me that have spent man-weeks trying to figure out what seems to work across multiple releases, and where my own MS sweet spot is before MP failures become much more pronounced (<7K tracks with minimized tagging in my MS), will find it refreshing to know what they can work within -- just as I have with every other auto mfgr for years that documents what their MP is designed to support as a couple pages in their owner's manual. (IIRC, generically, my former Lexus, BMW, and MBZ seemed to have a max of 8K tracks, with a narrow spectrum of supported codecs, bitrates, and tagging, with a max of 8 levels to the directory structure; 1 or 2K Phone contacts with limitation on the length of each field; and perhaps 10 favorite Nav locations and a similar number of past destinations.). I would take reasonable limitations any day, opposed to the unexplainable environment I continue to have.

Good luck with your further exploration!

 

[tomas]

2448 cfc software. Tested shuffle. Started from same song 3 times and got different sequence thereafter. Is this fixed? Do we have real random shuffle back?

 

[BertL]

^^^ I admit to being pessimistic (sorry!), but suggest owners on this later firmware release let MP USB play through at least 10-12 tracks without touching any MP controls, doing a track advance, or turning off your MS, to ensure there are no repeats before we consider there has been perhaps a significant change. If there has, that would be GREAT!!! Then, the next step is to test how random/shuffle operates on a longer set of tracks to understand if there are any repeats in the sequence before all songs in the “list” have been played. IMO Tesla has never documented if they are providing a random or shuffle ( meaning no repeats in a list until each track is played one time), so that is the other part of the equation we all still need to understand, since it’s not clarified in our Owners Manual.

I say all that having watched a Model 3 new delivery walkthrough video last week, where MP had only Slacker and inTune options in its interface — not even AM/FM (which also surprised the Tesla delivery specialist from his tone), let alone having any semblance of USB enabled yet in the interface. It just showed me that Elon does indeed puts Infotainment way down in the list of what’s important for initial delivery, with USB functions lagging way behind streaming... so, if we got a USB random/shuffle improvement in our “old” MS’ now, I’d be totally astounded with the unusual and unpredictable priorities Elon and his exec team are giving the engineering folk. . Personally, I suspect as I’ve said before that Elon is off on the new latest thing — being Model 3 and now AP 2.1 — so maybe we’ll get a retrofit of what Infotainment improvements are being designed for the latest Tesla Generation back to our classic vehicles one day, despite a promise of new UI and such to have been delivered last year.

 

[aesculus]

Yesterday out of the blue my car (fw 2017.34 2448cfc) decided to resume play on the last track of the last album I had listened to a number of trips before. I am pretty sure I had not listened to anything for at least three shorter trips so this was a bit of a surprise. And the really weird thing that I don't think it has ever done before, was to actually start the track I was listening to at the same place I shut the car down. I am not sure it has ever done that before. Prior to this I was always caught in Groundhog Day whenever it resumed playback at the beginning of the last track listened to.

 

[tomas]

^^^ I admit to being pessimistic (sorry!), but suggest owners on this later firmware release let MP USB play through at least 10-12 tracks without touching any MP controls, doing a track advance, or turning off your MS, to ensure there are no repeats before we consider there has been perhaps a significant change. If there has, that would be GREAT!!! Then, the next step is to test how random/shuffle operates on a longer set of tracks to understand if there are any repeats in the sequence before all songs in the “list” have been played. IMO Tesla has never documented if they are providing a random or shuffle ( meaning no repeats in a list until each track is played one time), so that is the other part of the equation we all still need to understand, since it’s not clarified in our Owners Manual.

I say all that having watched a Model 3 new delivery walkthrough video last week, where MP had only Slacker and inTune options in its interface — not even AM/FM (which also surprised the Tesla delivery specialist from his tone), let alone having any semblance of USB enabled yet in the interface. It just showed me that Elon does indeed puts Infotainment way down in the list of what’s important for initial delivery, with USB functions lagging way behind streaming... so, if we got a USB random/shuffle improvement in our “old” MS’ now, I’d be totally astounded with the unusual and unpredictable priorities Elon and his exec team are giving the engineering folk. . Personally, I suspect as I’ve said before that Elon is off on the new latest thing — being Model 3 and now AP 2.1 — so maybe we’ll get a retrofit of what Infotainment improvements are being designed for the latest Tesla Generation back to our classic vehicles one day, despite a promise of new UI and such to have been delivered last year.

yes, you are pessimistic. We all know these are easy fixes, so it is always possible they finally got around to some. Before 8.0, shuffle worked. It was random. Since 8.0, my shuffle has been a predictable sequence until this test. I wish you or others on this software would try to either corroborate or disprove.

 

[BertL]

yes, you are pessimistic. We all know these are easy fixes, so it is always possible they finally got around to some. Before 8.0, shuffle worked. It was random. Since 8.0, my shuffle has been a predictable sequence until this test. I wish you or others on this software would try to either corroborate or disprove.

Agree that additional collaboration will be useful as others receive 2248, and the fix should be simple as we've discussed many times here. It will be GREAT when the issue is resolved at last, and I appreciate your POV for something for us to be on the look out for!

I don't yet have 2248 to test myself (still on 2017.32 9ea02cb), but as I tried to constructively suggest based on your view, unless others get to it first, I will likely do additional tests against more tracks while not touching anything in MP (no pause, forward, reverse or power-off), as well as then exploring if we still have a "repeat" function or if it has perhaps been enhanced to a real "shuffle" mode once the new firmware becomes available to me. I suggested the additional tests off the top of my head to rule out things I did in my own past 8.1 testing giving me false positives in some of the interim releases, and to help rule-out where some other past first impression posts ended-up not being as conclusive -- nothing more. ...it's just my very old software testing self coming back out trying to be thorough. Sorry if my post ended-up seeming like some sort of attack -- that's never my intent! My best always.

 

[tomas]

NP, you've gone above and beyond in testing this stuff and developing workarounds. I know there have been false alarms about fixes, so let's see what our members find. I don't trust my own findings on this until they are corroborated.

 

[RichardL]

As a datapoint, my X has 2448cfc and I still have the same single sequence of songs on shuffle, so no fix for me. Also, and I know the X seems to behave differently than the S, but I still get rescans all the time (usually after a short stop, like at the store) and it never, ever restarts playing the USB at all - it always defaults to Bluetooth (which I have never ever actually selected) and I have to reselect USB and choose the starting song after each stop, regardless.

So for me, its just as bad as its ever been

 

[BertL]

...but I still get rescans all the time (usually after a short stop, like at the store) and it never, ever restarts playing the USB at all - it always defaults to Bluetooth (which I have never ever actually selected) and I have to reselect USB and choose the starting song after each stop, regardless.

So for me, its just as bad as its ever been

What codec (filetype) are your USB tracks -- do you have anything other than MP3? How many tracks do you have on your USB device, and are your tracks highly curated with extensive tag data?

I ask because (knock on wood) I'm approaching 3 weeks with zero rescans, 100% of the time my tracks pick up where they left off, and I've had no odd CID reboots since I migrated to 100% MP3 VBR files with minimized tag data as described upthread. I'm waiting to hit the month mark to declare this as the best workaround combination I've come up with for my use, and to proclaim how much lower my MS frustration level has become. Since the 7.x firmware days when I used FLAC exclusively, and my more recent memory reduction efforts that decreased failure rates, I've never gone even a week without a CID or Infotainment failure of some sort, and used to sometimes have it happen multiple times/day that really got me going. ...now, if random/shuffle would be truly random, I'd be a much happier owner until one day when I can stop using my MP workarounds all-together.

 

[DoNotBurnCarbon]

May '17 S100D v8.1 (2017.32.6 ca28227 installed on September 1, 2017)

Until approximately a week ago, the USB drive would always rescan (Loading nn% ...) after driving and leaving the Model S for greater than a few minutes. For the last week or so, that rescan has not happened. We can come back to the Model S, and USB music is immediately available.

As background, the USB is an approximately 2-TB Seagate USB with a few MP3 files and mostly FLAC files, totaling over 45,000 tracks in approximately 1.1-TB. Therefore substantial time is required when rescanned. That rescan is fine when iit s needed, but not fine when merely leaving the car and returning. The current function is welcome, yet I do not understand how or why the change happened.

Is it possible that Tesla performed some over-the-air modification without request?

 

[RichardL]

My tracks are nearly all (95+%) FLAC. When i had my S it never rescanned unless I physically removed the stick. When I got the X, I simply moved the same stick, with the same songs (and both cars were on the same software version), and immediately got rescans after every stop (which was behavior documented by other X owners). The situation is slightly better now, but rescans are still common. i did try a stick of just MP3s, with no difference in behavior. All my songs have consistent tags, small artwork etc. per your previous ideas and suggestions.

I just was out for a while so I did some experiments. I started out, and the car did a scan. I then chose a starting song, and clicked the shuffle. The second, third, fourth songs were the familiar set that always play on shuffle.

I then returned to radio, then went back to USB, and chose a new song, clicked shuffle and each time, I did indeed get new songs in the sequence. I went through this three times, selecting the same starting song, then selecting shuffle and each time i got different songs to follow.

I stopped for 20 minutes, it did not rescan, so I was able to shuffle again. Then i stopped for 5 minutes to get a sandwich, it decided to rescan. When it finally finished, I selected a start song, selected shuffle and songs 2, 3 4 etc. were the usual sequence of exactly the same songs.

So it seems that it is improved provided it does not rescan - whereupon it goes back to the exact same sequence.