With the problems some mp3 portable players had with VBR files mostly a thing of the past, does anyone out there no any playback problems with VBR files on MP3-CDs?

The company I work for currently producers Audiobooks on MP3-CD, and it would be great if we could keep the overall quality of sound, and squeeze a book onto one disc instead of spread over two, using VBR.

I've done some tests using the "strictly enforce iso" line with VBR files and this keeps the portable players happy, but what of the MP3CD players in cars and DVD players etc?

As the discs will be mass produced we obviously need compatibility with all common players.

Any feedback appreciated.

Since you work for a company why not rent cars with built-in MP3 playback capability for testing. Or work out some deal with a local dealer to test the MP3 playback function of cars.

Are you certain that CBR is ruled out? What's the (reasonable) max length of a book?

I don't think it is ruled out. My concern was that as VBR files should have in theory (or promise of manufactures specs) played fine on portable players, in practice some encodings wouldn't.

We currently test our MP3CDs on various players, hardware and software, but with many different products with many different features we can't check everything. We have had to change maximum file time since we began for instance. Initially we kept one chapter one file, until a customer told us his batteries ran out on his MP3CD walkman before a chapter which was 1 and 1/2 hours long would finish. With no bookmarking on his player he would have to start over again, and hold fast-forward for minutes on end...we now split chapters into smaller sections.

Our current CBR settings give about 16 hrs per CD, but for a book 18hrs long that means the cost of another CD being made. The blockbuster epic titles can be 20-24 hrs, which is huge in manufacturing cost on Audio CD, and makes it cost prohibitive to customers. MP3CDs are the way to go here.

I have VBR files working ok in tests, but was wondering if any negative experiences have been had by others, or any technical limitations are known.

Is CBR 64 kbps Mono not high enough quality for an audio book?
(I believe that'd allow about 24 hours per CD)

To keep a distinction between MP3 downloads, we do 96kbps on the CDs.
You will hear the difference if you put the MP3CD through your DVD player on a good system. And people feel they are paying more than a download but getting quality.

This is why we want to change to VBR, keep a fuller sounding file still with a good spread of frequencies, and get more hours on the disc.

I had an older mp3-CD car deck that supported VBR playback and almost all portable devices support VBR mp3 playback these days so I don't think you need to worry about VBR compatibility.

I'm not entirely sure I'm following you on that reasoning about 96 kbit MP3s distinguishing from downloaded files, and their perception to the listeners.

But are those 96 kbit/s files mono? Because if they're stereo, at that bitrate (especially CBR) it's not surprisable you can distinguish them from the original recording on good equipment.

I suppose stereo is indeed irrelevant for audiobooks, so LAME -V6 -mm, or a mid-bitrate mono ABR preset, such as --preset voice (= --abr 56 -mm) should provide fine results, saving quite big on those current 96 kbit/s encodings of yours. Play and ABX around a bit with these presets, and let us know the (your) outcome.

If any (buggy) hardware players are still hiccupping over VBR files (including ABR), just stick with mono CBR 64 or 80 kbit/s, which won't exactly make your stomach turn while listening either.

Edit: added some stuff.

I don't think they want to offer the MP3 CDs as just the same quality of files with a different delivery method, but rather as an upscaled product that certain people will appreciate and gladly pay more for.


I probably wouldn't notice much of a difference between 64 and 96 kb/s spoken-word media. But not everyone is equal, or a potential customer. Some people don't even need to hear a difference in order to appreciate the higher quality.

From personal experience, my iriver flash DAP (IFP-799) is the only device I've had a problem with. It plays VBR files, but doesn't correctly read the length.

The main problem I'm aware of with VBR files in standalone players is the seeking showing the wrong time display (edit: and the correct length as mentioned by Cosmo). You may or may not wish to consider this, and could choose ABR or CBR to improve matters.

The other issue to consider is whether using 32 kHz or 22.05 kHz sampling rate may be a problem with any decoders or give audible issues with DACs that aren't critically filtered. I've learned French with Michel Thomas (J'ai appris francais avec Michel Thomas) recently and just to ensure I avoided any problems encoded his CDs to LAME -V6 --vbr-new using LAME 3.97 (I didn't even set Mono, just let LAME smartly encode with M/S stereo when it's mono). With foundation and advanced courses plus the review CDs for both, 18h 31m of speech encoded to 521e+6 bytes. That's an overall bitrate of 62.5 kbps (8 * 520702.357 kbit / 66659.426 s)

I dare say you could get an awful lot of your talking books in -V5 --vbr-new onto a single CD. Given that for music, -V5 is typically reckoned to average 130 kbps and -V6 is about 115 kbps (click top link to Wiki then LAME), I'd estimate that if -V6 speech is about 62.5 kbps (based only on one test!), then -V5 speech might be around 70.6 kbps, and a standard '74-minute' CD-R with about 650MB formatted data capacity would amount to 8*650e+3 kbit / 70.64 kbps ~= 73600 s = 1230 m = 20.4 hours (estimate only). If a CD is actually 650 MiB, then you might get about 6% more than my estimate, or about 21.7 hours.

The reason -V5 is attractive is that in a multiformat ~128kbps listening test on varied music (not speech or killer samples, I hasten to add!) indicated that lame -V5 --vbr-new came out very well with an average score representing perceptible but not annoying differences from the original, and of course with almost universal compatibility. Speech may be a different story, of course.

For shorter audiobooks, -V2 or -V3 could be very attractive for a 'premium product', being almost indistinguishable from CD in the vast majority of cases (or largely "transparent"). So long as you use --vbr-new, the wiki says that neither of them will enforce a 128kbps minimum, allowing the bitrate to vary freely, thus suiting monophonic speech encoding rather well. However, if people would like to transfer the MP3s from your CD-ROM to their DAP (iPod, generic personal MP3-player, MP3 phone or whatever) they might still appreciate the more modest bitrate of -V5.

More severe problems with VBR did exist in the past. I had first-hand experience with a player, which wouldnt play any VBR files at all! Though, that was 4 years ago. I dont have enough knowledge about how the current situation is like.

I divide CD players in cars into two categories:

1. Aftermarket radios which are name brand installed units. (Pioneer, Sony, Kenwood) The the vast majority of aftermarket mp3cd players have no problems with any mp3 files you throw at them.

2. Factory Installed. i.e. the ones that come with your vehicle. These units can be a different story. Manufacturers are known to cut corners and save money when designing these units. I have experienced some units not being able to play my VBR files and such. While others have had problems with specific bitrates and frequencies. It's really a mixed bag.


This is a tricky question as there is no universal standard for standalone mp3 players. And there are plenty of older generation ones still out there. So unless you're talking about a specific brand and model, there is no 100% guarantee that your CD will work in every player out there. Keeping this in mind you want to be shooting for what will give you the most compatibility. In this sense CBR is the obvious choice.

Your best bet if you go the VBR route is have a back up plan in place. For instance if you have customers complain their CD doesn't work, give them a easy way to exchange them for a CBR copy. Personally I would stick with the most commonly used bitrates and frequencies. For books on tape: 64kbps CBR at 44.1hz or 22 KHZ.

It often feels strange trying to optimise for new technologies, when working also with the mono format used on wax cylinders and 78's!!
I should have mentioned we work CBR 96 kbps in glorious Mono, and stick to 44.1 for the odd player with D to A converters than doesn't like anything else, (as Dynamic pointed out).

Thanks for all the comments here, its a great help.

Lyx and spockep confirm what I suspected, that rogue players are hidden out there and hard to find.
What was the brand of player you had trouble with Lyx? As for car units, I have found Merc and Hyundai factory issue problem free, but they were recent models....

As for the best encoding for Audiobooks, most people if they don't get to hear a choice of A or B can be happy with crunchy whistling low rates. When I did tests though with rates at those used by some Audiobook download stores, and played it with one coming in at average 80kbps VBR at 32khz to other staff here, they would prefer the 80 everytime.

A lot of people copy to their iPods from MP3CDs, so we can't just fill the disc space with audio as they all want to save storage space, so we will stick to one size for all discs.

So, maybe I'll release one title as VBR and see what customer feedback is like before going further.