Ok.... so here's the deal: I finally got my feet wet with digital audio on my iPAQ. Moved to Vorbis at a friend's recommendation for (generally) smaller files at (generally) better quality than the MP3 I started with.... I'm not unhappy but it's time to explore the possibilities. Currently using OGG Vorbis compression at around 96Kbps on my iPAQ. I've read a lot here, and know I need to think in terms of -q settings rather than bitrates, so I'll translate to -q2.5.

I want files that sound good on my portable and on my desktop as well. "FM Radio quality" is acceptable to me, but as low as I'm willing to go; "CD Quality" is the ideal. If I can also use the same files on my (near future) home entertainment center's digital audio player with good quality, so much the better. My desktop player can handle pretty much any format and frankly is still in flux as to what I'll use long term.... the iPAQ can currently handle WAV, WMA, MP3, and Vorbis. File Size is a premium on the iPAQ, not as much so on the desktop. I also know that the encoder is where the quality is at..... but I'll worry about that after I find the starting points.

There's tons of info on this site..... but I'm getting myself confused and need some starting points.

1. As I said in the title...... am I dreaming about the goal of having a single song-file for my iPAQ and my Desktop with acceptable quality?? is it a reasonable hope..... or did someone slip something into my drink?
2. Is Vorbis my best answer? Or MP3? AAC (which my iPAQ player doesn't support... yet)? something else?
3. Quality is in the ear of the beholder...... but I need a suggestion starting point for file size vs. quality. Where should I start with on quality setting (and why if you'd care to add that)??

Thanks for all your help.
D.

yes for most people (here) a bitrate of 96 is just not even near to transparent quality, more like up from 180. BUT, this does not have to be valid for you. as you said, everyones hearing abilities are different from others, so the best solution would be if you ABX the different bitrates in vorbis. that way, you can be sure you get the quality that is best suited for you.
you might almost certainly want different bitrates (or just one transparent) for your home pc and your portable, as you prob. won't notice much artifacts in a noisier surrounding, but prob. will at home.

as for the coise of codec, there are two things to consider in your case: quality (which is partly subjective), and hardware support.
so, for your portable you are (atm) bound for mp3, wma and vorbis. it would be a good idea to ABX the codecs against each other, with a few samples of your favorite music. stick to the one which has the best (or least worse) sound to you.
if that is too much work for you, you might just want to use mp3 or vorbis at a bitrate of 96 to 128 (vbr).
for your home pc, it's another thing. there, you are (I assume) in a rel quiet surrounding and thus might want to have better quality, you are also more free in the choise of codec, so that adds mpc, aac and of course lossles (but what you said so far, I woudn't recommend lossles in your case).
again, ABXing the codecs would be the best solution.

so if this is your main interest, you might want to use aac (generaly supposed to have better 'quality' than mp3 at the same bitrate) up from a vbr-rate of ~130.

I agree with Digga's recommendations, and I'd also like to add something since I've been in the same boat recently considering a Vorbis quality setting which was the best compromise between small filesize for portable capacity limitations, while offering close-to-transparent sound quality.

Performing ABX tests, as Digga suggests, is the surest way to know the lowest bitrate that sounds transparent to your ears. Generally, if your ears are somewhat artifact-trained, then it may take Vorbis -q 5 (160kbps nominal) or -q 6 (192kbps nom) to give you transparency (or perhaps using the GT3b1 encoder...see below). If you've never practiced hearing specific, discreet artifacts, then -q4 (128kbps nom) may sound just fine to you. As a good compromise for PC and portable playback, and for the best quality/filesize ratio, -q 4 would be my first recommendation. If you are able to ABX that setting in more than a small handful of tracks, then try -q 5. At any of these bitrates, Vorbis is one of the best encoding formats you could choose.

Also, take a look at the two different Vorbis encoders. GT3b1 is tuned for higher sound quality at higher -q settings (-q 5 and up), but generates higher nominal bitrates accordingly (20kbps higher than Vorbis 1.0 and 1.0.1 at equal -q settings of 5 and up). Vorbis 1.0.1 is the latest Xiph build of Vorbis, and would be the better option for quality settings of 4 and lower, but also will work for higher quality settings if (because of limited decoder overhead of some portable players) you need tighter control over bitrate spiking than the more aggresive GT3b1.

When considering different formats...in a recent 128kbps listening test, Vorbis, MPC, QT-AAC and WMA-Pro were tied for sound quality. You say your iPaq supports WAV, WMA (Standard), MP3 and Vorbis. WAV is too big, WMA-Standard doesn't compare to WMA-Pro (or these other formats) in sound quality, and Vorbis beats MP3 for sound quality at this bitrate range. Hence, Vorbis would be your best option among compatible formats. Also note that it offers equivalent sound quality as QT-AAC in this range as well (since you mentioned this format in addition to the others).

I'm not going to say they're trained, but I do tend to hear things, particularly anomalous things, others don't. I remember at the beginning of my career, the old CDC "dishwasher" hard-disk units (300 mb on a removable disk-pack with 9, 14" platters) when I'd hear a small strange sound through a closed door to the computer room, all the fans and airconditioning on both sides, penplotters running, a radio on our side of the door, etc... and yell to everyone to save their work about 20 seconds before the heads crashed.

You both mention ABX tests.... I'll have to look that up as I don't think I've come across that term on these forums yet.



I've also read the term "Peeling" with regards to Vorbis, now Transcoding.... are these possibly solutions? I could keep a higher quality file on the desktop and transcode or peel it to the portable?? I do tend to rotate the audio on my iPAQ so it might be a solution.

Thanks again for the help....
D.

A viable process for Vorbis bit peeling does not exist.

the differnet programs with further explanations are linked in the faq:
http://www.hydrogenaudio.org/forums/index....7516#entry74066

basicly, abxing in the case of audio compression is to blind test different codecs/settings/bitrates against an original (wav file). one file (x, as you don't know which one) out of two is played, and you have to decide which one it is (a or b ). this process delivers objective results, bypasing the famous placebo effect.

If I might piggyback on the newbie train....

I have EAC and LAME up and running with good results on alt preset standard setting. I've read the FAQ about LAME setup. I want to compress the same thing at a variety of bitrates to judge the MP3 quality vs. file size for myself. My application is limited memory mp3 player for use on airplaine. FAQ recommends -Y and then goes on to list ABR rates. This is where my questions come.

1. Why not VBR at lower rates?

2. Is alt-preset <bitrate> the way to go for my initial tests?

3. Any sources on tuning LAME settings between 64K and 128K?

Thanks,
Alan

1. new LAME builds (see recommended compiles) support the "medium" preset but you can create a command line of your own (the alt-preset optimizations won't work then) that covers your favourite bitrate range ...

2. alt-preset <bitrate> will give you ABR files ... I had good results with --alt-preset 135 for my 128 MB USB MP3 memory stick

3. I would not use any mp3 settings in the range (64->128) you mentioned ... better use a more optimized codec like vorbis/AAC in that range if you are not stuck to mp3 ...

Is there any codec that is 'transparent' at 96 kbps?

Generally speaking, no.