The Song Fight Bulletin Board

I am consistently dissapointed in the quality of my stuff after I turn it into an MP3. I have what I think is a good sounding mix but, after I encode it, it always seems to come out much quieter and less dynamic. I know I have to lose something in the process but it just seems like I'm losing too much.

I currently use Sonar 4 and the attendant encoder.

Does anyone have any suggestions on how to improve this process so I get a better final product?

Thanks,

Steve

I'm not sure what options you get with your Sonar package, but in Cubase I can select the bitrate of the mp3. I noted that you use 128kbps, which is about where the difference in audio quality becomes significantly noticeable. Try upping it to 160kbps or 192kbps and no doubt you'll be far more satisfied.

Beware thought that those files, especially at 192, are significantly larger and require alot more d/l time. Not much of a problem for most folks who have broadband, but for those on dial-up your name will be mud.

Life's a series of trade-offs, isn't it?

yeah, as i was checking your archive for what bitrates you used, THC was typing the same thing i was thinking. 192 is good but higher than that is pointless for songfight.

also, it seems to me you like reverb. alot. i could be wrong. but it seems to really drown out your vox. if you had a better vox signal and less reverb, then doubled that, you might be surprised at the difference.

also, you like to use alot of brass in your arrangement, which really hogs alot of the low frequencies. maybe compressing the horns or maybe throwing some fuzz on the MIDI bass might help. i think if you could liven up the fake stuff, it would help breath life into your mixes.

that said, i like your stuff!

TheHipCola wrote:Beware thought that those files, especially at 192, are significantly larger and require alot more d/l time. Not much of a problem for most folks who have broadband, but for those on dial-up your name will be mud.

Life's a series of trade-offs, isn't it?

Side conversation, sorta related, branching here:

I try not to ship songs that are bigger than 3MB. Which is (as my songs are not so terribly complex that dinking about with the bit rate seems to make much difference) what the determining factor is for my final bit rate. Anyone else use size caps for determining their bit rate usage?

Am I just leafing out here, 'cause I'm bored at work and posting whatever comes to mind? Now, avatars, that's a whole different story, heh. This one comes down when the Residents sidefight is done (much to the relief of the dial-up people, I'm sure). Nice traffic-cam shot, btw, HipC.

It could also be that the encoder that Sonar is using isn't very good. I recommend LAME as your encoder. I use Sonar as well and never save it as an mp3 from the program. I always render a WAV and then convert it with a audio-converter program. This allows me to tag my song at the same time that I encode it.

Rabid Garfunkel wrote:
I try not to ship songs that are bigger than 3MB. Which is (as my songs are not so terribly complex that dinking about with the bit rate seems to make much difference)

All of my best work is done at frequencies only dogs and dolphins can hear - so many of my song files run into gigabytes


But I have found that when I used to use Audacity to convert my files - I was also often disappointed. Now I use Garageband and iTunes - I am disappointed less often (although "Pink Skirt" really pissed me off)

Caravan Ray wrote:All of my best work is done at frequencies only dogs and dolphins can hear

Ha! And I am handsome in infrared.

I've never met an encoder that did a bad job and I've used every cheap and available thing I could find. The old days of SF required 64kbps mono and folks made some nice-sounding stuff that way: nothing you'd want to spend $16 for a CD of, but nice-sounding.

Steve: Your "One Less" sounds great; I doubt very much the information lost in translation to MP3 is very useful. I think you're being far too picky about your production. Better to work on those songwriting skills, eh?

Mostess wrote:Steve: Your "One Less" sounds great; I doubt very much the information lost in translation to MP3 is very useful. I think you're being far too picky about your production. Better to work on those songwriting skills, eh?

Burn!

I noticed an enormous difference in quality when I switched from Acid's encoder to Lame. Mostly in high frequencies. Even keeping with identical bitrate settings, Lame just does a much better job.

At one point I had an A/B test file put together, but it was lost in the new computer setup. It really was drastic.

Get WinLame, then get RazorLame to be your friendly front-end.

Lame is great, I concur. I don't know if this is possible or if it's just my fetid imagination, but I find the sound quality of the mp3s are better through razorlame than when I use lame as a plug-in through my audio editor ... but it means that you can't edit ID tags (unless someone knows a trick that I am missing)

j$

I usually edit my tags in Winamp anyways. Just play the song in winamp, right click it and do 'view file info.'

Thank you all for the help.

I upped the bit rate to 160 for this weeks entry and I can definitely tell the difference.

I am also going to look into LAME.

Steve

Way late, and this is just differently-appropiated, but:

This is what I advise for VBR MP3 creation:

1. Download LAME 3.96.1 from here: http://www.rarewares.org/mp3.html (go to "Encoders/decoders built using LAME 3.96.1," then download the bundle)
2. Unzip that into a directory (say, C:\Program Files\LAME)
3. Download RazorLame at http://www.dors.de/razorlame/download.php
4. Unzip RazorLame into a directory (\LAME should be fine)
5. Open RazorLame. Go to Edit -> Options, and find lame.exe from step 2
6. Go to Edit -> LAME Options, then the Expert tab, then under "Custom options" enter the text "--alt-preset standard" verbatim as in the quotes. Check the box "Only use custom options." Hit OK.
7. To use RazorLame drag and drop, or use the Add button, to find .wavs (that is, export wavs from Audition and then drag them to here). Hit Encode and then it'll encode an mp3 to the same directory as the wav is from (though that's a configurable option, and you can also elect to have it delete the wav after processing the mp3).

the "--alt-preset standard" option tells LAME to encode with VBR. Most songs will fall in the 128-212kbps range, but in my experience it's usually 128-170kbps. Will sound fine. If you don't think it sounds fine, you can use "--alt-preset extreme" in step 6 instead, which will discard less audio information for the mp3 compression. APS (as it's called) is a good compromise, however, suitable, certainly, for SongFight. You'll have to edit the ID3 tags manually (and please do--just open it in Winamp or iTunes), but this will do the MP3 compression right.

However, if you don't want to do that, you can forget the --alt-preset and "only use custom options" shit. Instead just navigate about the tabs: under General, hit up 128 kbps, mode Joint Stereo. Under Advanced hit up optimization: quality. Under VBR, it's probably wisest to enable it, and then you can make the maximum bitrate 128 (if you want to use smaller block sizes only for stuff like silence) or 160, or 192, or any other value (and then it'll behave more or less like APS). Under "Quality" there's numbers from 0-9: 0 is the most kbps and 9 is the least. Default is 4 and that should be fine. You can use ABR instead of VBR and go for 128 kbps, and that might be better for what you're after. But definitely use ABR over CBR if you aren't going to use VBR. It'll be a negligible difference in file size, but the quality should only be superior.