There is one point, I never really understood .... why do most (all?) mp3 encoders restrict the bitrate/frequency combinations ?
Example:
http://onlinehelp.avs4you.com/Appendix/Fil...mbinations.aspx

Why can't I encode 44100 Hz stereo at a bitrate of 96 kbps for example ??

Erm, you can.

Not according to the table I linked to above.
My lame converters don't let me either ...

This was just an example anyway. You can replace 96 kbps with 80 or 64 in my OP.

Here

works fine.

If you really should want to do this is another question.

The table that you linked to is highly misleading, if not outright wrong, and should probably be disregarded.

That restriction is there to protect users from their own ignorance. What sane use case would require what you are asking for?

Encoding mono (or nearly mono) material with joint stereo would be a valid use case. Of course, in addition to specifying the bitrate, you should also force a higher lowpass frequency, else the higher bitrate would be wasted.

Many thanks!!

I'm just a bit worried about the resampling parameter. If my source is already in 44.1 won't there be any resampling at all?
And how can I be sure about that?
I know, it's a very bad comparison, but just for the sake of argument: If you have a 128 kbps mp3 file, you can re-encode it a *second* time to 128 kpbs, so I'm a bit worried, that this forced resample parameter would force resampling even if the frequency matches the source.

Why so? I have seen *lots* of tables like this on the Internet. Hm ... I guess those tables reflect the predefined settings the encoder defaults to rather than a complete impossibility to combine those frequency/bitrate parameter combinations, right ?

I'm sorry to say this to a developper (who - as such - certainly has a lot of experience), but when you imply that users resorting to these settings are "ignorant" and "insane", you should be a little more tolerant in my view. And on top of that, audio quality is subjective.

I, for one, do find that 44.1/96 sounds better than 32/96.

Good idea, thank you. Would you, by any chance, know about a table specifying the lowpass frequencies lame defaults to at given bitrates (and/or other params) ?

Because of the code in the second quote I just posted.



Its an example meant to prove a point. Obviously using that with 44.1kHz audio makes no sense.



http://wiki.hydrogenaudio.org/index.php?ti...cal_information

If you specify --resample with the same sample rate as the source, lame simply takes that as a requirement to NOT change the sample rate. There is no resampling, and the file will only be encoded once.

Right!


Sure it does make sense, because without using it, it won't stay at 44.1 (in this case).


excellent, thank you!!

I'm not here to flash you the TOS#8, but am wondering if your end-goal is "quality as close as possible to the original source". It is quite easy to create files that sound better. It seems, then, that you prefer having some frequencies above 16 kHz with additional artifacting below it to having no frequencies above 16 kHz but with relatively less artifacting below it.

Very attentive and logical explanation, helping me a lot, thank you!

That's a very good analysis of yours.

But, to be honest, I'm ashamed to say that at the moment I'm in the middle of figuring out that there must have been some inadequacies in my encodings. Because I compared the results again and find them virtually identical (I don't have a good aural sense), and 44.1/96 certainly does not sound any better than 32/96 to me anymore.

I don't know what went wrong beforehand, but whenever I had to encode something to less than 44.1 kHz the sound was extremely dull/muffled (don't know the correct term in English), as if at least the upper 50% of the audible spectrum were cut off ...
That's why I wanted to avoid anything <44.1 kHz at all costs !!

I don't know where this error could have come from, as I was using lame (most recent versions) and never had any custom parameters ...


well, that's everyone's goal, I guess (if you leave aside the quality/filesize trade-off)
but other than that my goal was rather to avoid this horrible upper frequency loss at <44.1 kHz (which now seems to have been the result of some error)

I'm not sure why you would have picked such a low bitrate as 96 kbps, as you are pretty much guaranteed to end up with artefacts or loss of high frequencies or both, but if you really need to then you are better off with either VBR or ABR instead of CBR. Either one should give you noticeably better quality at that bitrate. I recommend that you do some listening tests for yourself.