Suggestions for reducing MP3 size but keeping so-so quality?

[j_hallgren]

New to this stuff...have some MP3 files that I created from ripping CD's of mine...I'm not that picky on audio quality so used what I understand is "128bit, 44Khz, Stereo" format...the largest is 4864KB for a 5:10 song...well, I wanted to put some of these on my older 128MB thumb drive but ran into space issue...so I need to try to shink the files a bit...

Thus looking for suggestions on what format to use...I'm used to listening to talk show on radio on bad reception station so my desired playback quality is not that critical...I'll be using a FM modulator to play the songs back thru my car stereo, which is factory unit so not great quality...looking for something that might resemble the sound I'd hear if song was heard on a regular FM station...

Also, what freeware (that I may possibly have forgotten that I've got already) would you suggest for doing this?

Thanks!

[FulciLives]

Many people feel that 128kbps is the absolute lowest bitrate one should use for music when it comes to the MP3 format.

In fact many never go lower than 192kbps for quality reasons while others will only use 320kbps (the highest bitrate possible with the MP3 format) again for quality concerns.

Asking to shrink below 128kbps is just not a reasonable thing to do if you want the barest amount of "quality" at all.

You need bigger hardware with more breathing room i.e., something with more space that your 128MB thumb drive offers.

Also I suggest you start encoding at 192kbps at the very least for "reasonable quality" as 128kbps is really not enough bitrate for most types of music. Maybe OK for like a radio talk show or something that is just voice but definitely not music.

- John "FulciLives" Coleman

P.S.
Most people would argue that 128kbps MP3 is already at if not well below quality FM radio audio quality.

[Alex_ander]

Use 'joint stereo' instead of true stereo, this reduces file size at a selected bitrate, don't use other encoders than Lame (best builds are 3.90.3 and the latest 3.97) since they may give extra distortions in joint stereo mode.

[AlanHK]

Read the manual for your drive and see if it supports any other formats than MP3.
MP3 isn't the most compact, but it is the most widely used.

Just try ripping tracks at lower rates and load them. The size of the file is basically directly proportional to the rate. 128 k is about 1 minute/MB. Thus 64 k would be 2 min /MB. But that would be tolerable for spoken word, but pretty rough for music. More like shortwave than FM.

Also, if at all possible, go back to the original CD to make new files, don't use your MP3s as the source.

[j_hallgren]

Thanks for the replies! I found that I had Audacity on my sys...had not used it for months and didn't have any shortcuts to it...it had lame 3.96...tried converting a few tracks to 128k/44KHz...one track went from 4256KB to 2832KB...in listening to it via my PC, I couldn't tell the difference.

I don't have the original CD's here at my summer place so re-ripping is not possible...

Also tried going to 112k/32KHz...size went to 2480KB...again, my ears aren't that picky or good so couldn't really tell a difference...

What impact does the reduced MHz rate have in quality? I presume that it means that some higher freq sounds are trimmed, in this case? And how does that affect file size?

Also, I noticed that output went to stereo from joint stereo...would this add to file size again, and offset gains from reductions elsewhere?

[smitbret]

- John "FulciLives" Coleman

P.S.
Most people would argue that 128kbps MP3 is already at if not well below quality FM radio audio quality.

Actually, FM is somewhere closer to something between 70-80kbps. If you do a direct comparison, it's obviously a considerable amount lower than 128kbps.
-Brett

[celtic_druid]

Joint stereo doesn't decrease the size at a set bitrate. At a set bitrate the filesize is determined by the bitrate and only the bitrate. Joint stereo is more efficient and works fine in lame.

In various listening tests 128k has been shown with recent lame builds to be generally transparent. Talking VBR here though. ~192k VBR would be transparent for pretty much anything except for samples where even 320k CBR isn't enough. Only reason to encode to CBR is if you have a broken decoder that can't handle VBR.

I'd suggest lame with -V 5. See if it sounds ok.

Re-encoding lossy to lossy is never a good idea. Especially if your source is 128k CBR.

[Cornucopia]

Couple of suggestions (although I strongly recommend NOT re-encoding if you want to avoid compounding the artifacts):

1. Apply Dynamic Range compression and/or Peak Limiting (mildly). This will keep most of signal is small window of volumes.

2. Apply NoiseGating/Expansion to make areas that are "almost silent" become "silent". If they're all 0000's, they'll compress better. If it's spoken word, you can get more extreme with this and actually replace areas between the words with true silence. This allows the rest of the signal to make more use of the remaining bits. (which brings us to next option...)

3. If your player(s) will support it, use VBR encoding.

4. Apply HiCut and LowCut filters (aka BandPass) to signal. This has an even greater affect on bitrate savings than does #1 and #2.

5. (***This one's experimental -- just worked it out): 1/2 wave rectification (aka only positive-going part of signal is retained). Can try converting to mu-law to get approximation of this (if you can't get rectification plugin/algorithm).

6. Big one: go mono if you can. This will save 1/2 the bitrate.

HTH,
Scott

[edDV]

Couple of suggestions (although I strongly recommend NOT re-encoding if you want to avoid compounding the artifacts):

I was going to say something similar.

If you are going for a world record for the worst audio you can stand, first ditch stereo but from then on always encode from highest quality source you have. Never re-encode.

[j_hallgren]

So could someone now point me to which free software I should use to try to create one using HiCut/LowCut (BandPass)? I may be overlooking the obvious, but I don't see this option in either Audacity or BeLight, which are the two that I'd installed long ago...

Also, when using BeSweet, at the end of a conversion, I get a soft abort msg from it...

And is there a guide to doing batch converts using BeSweet? I don't quite catch on to how..

Part of my problem is I get overwhelmed with all the terminology...

[AlanHK]

And is there a guide to doing batch converts using BeSweet? I don't quite catch on to how..

I don't know about the abort, or the more complex filtering; I think you'd want a more interactive sound editor app.

But to do batch converts, install the "Besweet Gui". Then choose "Wizard" mode.
That in turn has a "batch mode". You can now open a Windows Explorer window and grab a bunch of media files and drop them into the file list.

Then select the output options and then "Go" and it'll output them in the same folder with a timestamp in the name.