Transcoding to mp3 vbr questions

[ziddey]

I'm wanting to transcode my entire library. At first, I'll keep my old files available just in case my settings were too aggressive. I know the recommended approach is to rip directly to what you want instead of transcoding and enduring the additional loss.

In my collection, most of my older songs are 128kbps cbr. Progressing to 160, 192, 256, 320, as well as lots of vbr.

I'd like to use vbr, and cut down the footprint on my computer, as well as on my ipod (I currently do transcode beyond a certain bitrate for my ipod, but am looking to have everything slimmed down).

I did a lot of reading and found what "optimal" values would be to use, but figure I'd ask for a more personal opinion if one can be given.

Personally, I used to think 128kbps was all I'd ever need. Nowadays, I can tell 128cbr vs 160cbr for some songs. For some others, 128 is still good enough. But that said, I'm pretty much completely content with 160kbps cbr, so I'd like to work from there.

I understand that sometimes, I'd possibly want more than 160kbps for parts of certain songs (wouldn't be able to tell currently since many of my songs are >160). So that said, I'm imagining that setting V to 3 should be more than enough for me, and still allow it to drop low enough when not as needed? I did some tests at 2, and it pretty much hugged the max bitrate I chose. 3 seemed to go much more sharply lower.

And what about near silence parts? From the guides I've seen, the min bitrate is still set to be quite high (96 is usually the lowest I've seen recommended). Is there a way to allow silent parts to go down to say 32kbps (or even lower if it's still acceptable) or something like that without harm to where bitrate is needed? I assume this ties in closely with what I set V to?

I've been using winamp to do the transcodings using lame, but was wondering if there's something better to use. Also, if it could support multiple threads, or at least run simultaneous transcodes on the different cores, that'd be great (winamp only does one thread). Free would be nice too, and batch transcoding is a must.

So here's what I'm thinking:
lame vbr 32-192 V3

is there a problem with this?

Also, one thing I've found with winamp is that even if the source is only 128 or something low, after the transcode, it'd almost certainly be higher in bitrate than the original, and thus defeating the purpose. Is there a way to make it follow a different ruleset (maybe have it cap max bitrate at whatever the max/cbr of the song is?)

While I will be keeping the original mp3's for a while until I'm sure I'm satisfied with the results, I'd still like to get some advice and do it a way that isn't too horrible, since it will take quite a long time, even with two threads and 3.6ghz c2d.

Thanks!

edit: currently, I have songs going to the ipod transcoded to 96-224 v2, and am able to fit about 40% of my total collection. Albeit it's all the songs that I want on the ipod at the moment, my ipod is sitting at 28gb used (30gb ipod), so I'll be running into trouble real soon in the future. I'm sure 96-192 v3 would be totally acceptable and would win me over a lot more space, but like said, I'd rather do a main transcode on my pc and have everything the way I want instead of having to be constantly retranscoding to find what works.

edit2: also I've been using joint stereo. nothing wrong with that right? I understand it should be able to save some space without much sacrifice


edit3: while I'm at it, if a program could also normalize it as well. might as well do that as well so I don't have to be constantly changing the volume when on shuffle

 

[fellows]

Okay I understand your want to shrink your music collection. And given a 320kbps is almost 2x a vrb or 192kbps, I still don't think it's a good idea.

From my understanding Transcoding is never a good idea. You are taking an already lossy format, and putting it through another lossy codec.

Your idea of using 192kbps or VBR is fine. 192 is basically the set minimum that mp3's should be. There is a huge difference between 128 and lower vs standard 192. My understanding is that 3/5 people cannot tell the difference between 320kbps (the highest typical mp3 setting) and 192kbps, but I would never transcode my 320's down.

If you are starting from a lossless source like WAV, FLAC, etc then it's fine to make them 192 standard. Otherwise, try it out with a few albums and test if you can tell the difference but, I don't recommend it.

 

[phide]

I know the recommended approach is to rip directly to what you want instead of transcoding and enduring the additional loss.

And that's what I'd recommend doing instead, and keeping your archive in FLAC, TAK, wavPack or any similar lossless formats, but transcoding isn't going to be that big of an issue for portable usage.

In my collection, most of my older songs are 128kbps cbr. Progressing to 160, 192, 256, 320, as well as lots of vbr.

I'd suggest leaving everything at 128kbps as it is.

Nowadays, I can tell 128cbr vs 160cbr for some songs. For some others, 128 is still good enough.

Sometimes, being able to discern a difference between a lower bitrate and a higher bitrate doesn't mean that you should necessarily use the higher bitrate. Keep in mind that this is for portable use, so don't overemphasize the need to maintain absolute transparency. Try to make a compromise. LAME should offer acceptable results down to V8, but you might want to consider using ABR for bitrates that low. 96kbps should be passable, but Nero's AAC encoder may offer better results at that bitrate, and will be iTunes compatible so long as you change the extension to .m4a.

Is there a way to allow silent parts to go down to say 32kbps (or even lower if it's still acceptable) or something like that without harm to where bitrate is needed?

I don't know of any encoders designed to do that, no.

Free would be nice too, and batch transcoding is a must.

Try dBpoweramp Reference. Free trial, $28 to register. Should support everything you need.

also I've been using joint stereo. nothing wrong with that right?

Joint stereo isn't a bad thing with today's encoders despite some perpetuated myths. Take a look.

while I'm at it, if a program could also normalize it as well. might as well do that as well so I don't have to be constantly changing the volume when on shuffle

That's not what normalization would accomplish. That's what iPod's Sound Check and, alternatively, ReplayGain, are for. For shuffling, just use Sound Check, and turn it off when you're listening to whole albums (to preserve differences in track dynamics, as Sound Check calculates per-track only). You obviously need to have iTunes calculate Sound Check levels for your library first, which could take, well, a really long time.