Audio File Compression - What's your favorite

[Version3]

The fact that iTunes compression was pointed out to be fairly lousy on another thread, made me curious as to what method others use to compress their audio?

I get even worse results than iTunes when I compress straight out of Logic Express, so I use Sorenson Squeeze 4 for MP3 compression. I've been compressing at 64/stereo/22.5 but am making the 'Light' show 128/stereo/22.5 because I like the sound better. I thought about trying both shows at 96 for a spell to see how they turn out.

What's your software/setting formula?

[podcastrant.com]

I use Audacity's compression which I've read is worse than ITunes. I have compliments on my sound so I don't know. I listen to shitty lo-fi indie music so my ears are damaged beyond "technical esthetic" repair.

I would also be interested in what other more audio 'techie' podcasters think.

[Big Mike]

I compress at 192 stereo on iTunes. I've only tried a few other things. I like iTunes fine, for what it does.

[The Truth]

I compress at 192 stereo on iTunes. I've only tried a few other things. I like iTunes fine, for what it does.

On a TALK show?

Dumb.

[Version3]

Please let's keep the posts helpful. You've stated your opinion on the max requirements for the talk format, but the question of the thread is for what podcast producers like to finish their shows at, so I'd expect that podcast producer comments will be the most useful here.

[Craig]

I'm a geek at heart, so I compress with a G5-optimized version of LAME on the command line, using parameters I picked up from searching the LAME forums.

Craig

[Version3]

I've never used, or dealt with LAME, so I know nothing about that. But I AM a Mac user. :wink:

[jeffoest]

I've used itunes, command-line LAME, Audition and Cool Edit Pro. I have found Audition generally better for talk material and LAME to be better for music material through my rather rigorous testing.

I used to go with 64/22 and have worked through intermediate settings including VBR and now just stick to 128/44.1 CBR using Audition. It sounds the best without any artificasts (which I really hate but most people probably don't care as much about) and has the high frequencies over 11Khz that I think make the voice sound more professional (if using good mics). I conciously made a decision to optimize my sound for high-bandwidth situations.

[Version3]

Well, we fight with voice sound anyway because none of the three of us know what it takes to make it sound 'just right'. We use three dynamic mics that are much better suited for vocal performances than voice work... but they don't sound like poop.
I've used iTunes the most for quite some time (until recently) but a few shows back it just starting sounding extremely bad. I don't know if it's the compressor in iTunes, or something else changed.

We do such a long show that encoding at the level you use Jeff would probably cripple some of our subscribers... but it DOES sound much better.

[kneelw]

Since iTunes 5 came out, mp3s encoded with it at lower bitrates sound like crap. Thanks to Craig's suggestion on a previous thread (http://www.podcastalley.com/forum/viewtopic.php?t=4751), I use the Itunes-LAME plug-in for iTunes at 64/stereo/22.5 and it's better than anything iTunes produced. It does take a little longer to encode, but I'm not too fussed about it.

- Neil

[Version3]

Thanks for that Neil... that's about when I noticed iTunes was turning our mediocre audio files in to full-fledged crap. I'll check out that LAME plugin for iTunes. :D
Extended thanks to Craig for bringing it up as well. :)

[kickasspodcast]

I am happy with the encoding functions of Soundforge.


Can't complain, sounds good to me.


Jack

[jeffoest]

Since iTunes 5 came out, mp3s encoded with it at lower bitrates sound like crap. Thanks to Craig's suggestion on a previous thread (http://www.podcastalley.com/forum/viewtopic.php?t=4751), I use the Itunes-LAME plug-in for iTunes at 64/stereo/22.5 and it's better than anything iTunes produced. It does take a little longer to encode, but I'm not too fussed about it.

- Neil

I'll be darned. I didn't know that either. Previously I WAS using iTunes for the encoding - I thought I was just getting pickier!! LOL Weird that a piece of software would actually go backwards on some functionality with a version upgrade.

[SPThom]

I am happy with the encoding functions of Soundforge.
Similarly, I've used Sony Vegas without any apparent problems. Mono 64kbps MP3. There's a bit of hiss/white noise in the background, but I think that's actually my iTunes... I've played it elsewhere and it sounds clear.

(Yes, I use Vegas... I'm sure I should be using Acid or something, but I'm used to Vegas from cutting video. I do some fine-tuning in Soundforge.)

[SFEley]

Here's a typical command line for me:

lame -q 0 -b 96 -m j --tl "Escape Pod" --tt "EP024: The Death Trap of Dr. Nefario" --ta "Benjamin Rosenbaum" --ty "2005" EP024_DrNefario.aiff EP024_DrNefario.mp3

For those who aren't familiar with the options:

-q 0: Quality level 0, or highest quality, and slowest encoding. This is probably unnecessary, since my own ears can't tell a difference and it takes 1.5 eternities on my Mac Mini, but I'm paranoid. When it's too late at night and I'm in a hurry, I use -q 1 instead. It's about 5 times faster and sounds the same.

-b 96: Bitrate of 96 kbps. I used to use 64, but Vox Monitor complained that it sounded too compressed. >8-> And it does make the intro and outro music sound noticeably better.

-m j: Joint stereo mode. This is fairly recent; until a few weeks ago I was encoding in mono, which made sense because it's really just story narration. I changed a few weeks ago because: A.) there's no reason my bed music can't be in stereo, and B.) I'd heard that a few older MP3 players have trouble with mono files. That second one was really the clincher.

...Anyway, after it's done encoding, I drag it into iTunes to finish tagging. I set the genre to "Podcast," attach my copyright notice to the comments, and add my logo image. Then I drag it back out of iTunes, fix the silly renaming that iTunes does, and upload it. Done.

[rookiee]

I'm not sure if I did something wrong, but when I compressed one of my mp3s at 22.5khz, some media players on the web played it ALL wrong, and I had to go back and upsample to 44.1khz. Anyone else have this problem?

[Version3]

I actually love SoundForge... but I'm a Mac user now, so bye-bye to that. Peak has it's similarities, but they are really just as different as Mac OSX and Windows.

[Big Mike]

I compress at 192 stereo on iTunes. I've only tried a few other things. I like iTunes fine, for what it does.

On a TALK show?

Dumb.

er....I have a music show.

[Hittman]

Since I’m only doing voice (with some minimal music beds) I use 64/44. I’ve done an A/B comparison between 128 and 64, and there is a very small difference, but not much, and not enough to justify double the file size.

I’m using Audition, which does a good job of compression.

[justSue]

I compress at 192 stereo on iTunes. I've only tried a few other things. I like iTunes fine, for what it does.

On a TALK show?

Dumb.

er....I have a music show.

now THAT's a snap!

[guscave]

I'm using the plugin that comes with Cubase SX. I've been compressing at 128 kps @ 44.1khz, but since my show is about 30 min long with continuous music playing all the time, the files were still to large (over 20 mb). so now I'm compressing at 96 @ 44.1. Not a big difference in sound quality but I'm now storing at about 13mb.

[mpeacock]

I use Audition at 64/22.5 for what is primarily a talk show -- some music for transitions and beds, but that's not the focus of the show -- and have only received compliments on the production values (which I assume implies acceptable sound quality). What was interesting to me was the drop in file size when I moved from recording on the iRiver internal mic to an external Radio Shack condenser mic. Even after using Audition's noise reduction filter with the iRiver mic, the reduction in background noise with the change in mics really cut down the file size.

[SFEley]

I use Audition at 64/22.5 for what is primarily a talk show -- some music for transitions and beds, but that's not the focus of the show -- and have only received compliments on the production values (which I assume implies acceptable sound quality). What was interesting to me was the drop in file size when I moved from recording on the iRiver internal mic to an external Radio Shack condenser mic. Even after using Audition's noise reduction filter with the iRiver mic, the reduction in background noise with the change in mics really cut down the file size.
Unless you're using variable bitrate encoding (which isn't recommended, since it won't play on all players), that nature of the sound shouldn't make a difference at all to the file size. A sound file of a given length encoded at a given bitrate will be the same size as any other sound file of the same length.

[Version3]

Is that accurate? Are you sure? I do not encode using VBR, and my file sizes vary based on the content. The data rate is set, but the data is still a variable (the sounds) and will affect the finished file... is this true?

[jeffoest]

Yep - SFEley is correct.

[SPThom]

Seconded.

Hence why it's 64kbps... It's literally storing 64 kilobits of information per second of playtime.

[Version3]

heh, learn somethin' every day.

[SFEley]

Is that accurate? Are you sure? I do not encode using VBR, and my file sizes vary based on the content. The data rate is set, but the data is still a variable (the sounds) and will affect the finished file... is this true?
Think about it, Version3. You know what "kbps" stands for in 64 kbps? "Kilobits per second." It's the number of bits that represent one second of audio. So a file that's ten seconds long will be composed of 640,000 bits, which comes to 80,000 bytes. This is true no matter what sound is in those ten seconds. Ten seconds of a silence or ten seconds of a Metallica concert will both be 80,000 bytes. The silence, however, will be more accurately represented, because the samples are easier to compress.

[SPThom]

Think about it, Version3. You know what "kbps" stands for in 64 kbps? "Kilobits per second." It's the number of bits that represent one second of audio. So a file that's ten seconds long will be composed of 640,000 bits, which comes to 80,000 bytes. This is true no matter what sound is in those ten seconds. Ten seconds of a silence or ten seconds of a Metallica concert will both be 80,000 bytes. The silence, however, will be more accurately represented, because the samples are easier to compress.
Haha... Jinx. :P

[anotherquizshow]

I use MusicMatch Jukebox. It seems to work ok.

Michael

[audio2u]

For those using Lame, you should be aware that the developers have dropped the -alt preset system in favour of the new -V0 ~ -V9 scale, and that the currently recommended version is 3.97beta.
This can be downloaded from Rarewares, and used with a frontend like All2Lame.
Personally, I use the abovementioned version and encode my podcasts as -V5 --vbr new which generally gives a bitrate of around 120-140kbits.

[SFEley]

Personally, I use the abovementioned version and encode my podcasts as -V5 --vbr new which generally gives a bitrate of around 120-140kbits.
FYI, VBR encoded files won't play on all MP3 players. You've probably got great quality, but you may risk losing listeners with old devices.

[Craig]

Just out of curiosity, has anyone here who's using VBR ever received a complaint from someone who can't play the file? I'm not sure the "VBR doesn't work on all MP3 players" argument is valid anymore.

Craig

[SFEley]

Just out of curiosity, has anyone here who's using VBR ever received a complaint from someone who can't play the file? I'm not sure the "VBR doesn't work on all MP3 players" argument is valid anymore.
I don't know, since I've never encoded in VBR. >8-> I heard it from Doug Kaye of IT Conversations, who is as close to an absolute technical authority as podcasting has, so I just took his word on it.

I did, however, get an e-mail after my very first episode, from someone who claimed that my initial MP3 file -- encoded with a 24kHz sample rate -- wouldn't play on his Palm PDA. I resampled it to to 44.1 and he was fine. If some players are finicky about the sample rate, I can easily believe that they'd be finicky about VBR as well.

[audio2u]

No, I've never received a complaint from a listenere about VBR not playing on their device.
I think you're right in thatthis argument is becoming less and less of an issue these days.
It would only affect people with REALLY old players (like >3 years old, and who keeps a hardware device that long, these days?) or really old software.

[revupreview]

VBR-encoded files may sound better, but I've had problems with the pause function on both an iPod and an iRiver when playing these. Neither player seems able to resume from exactly the same place it was paused, sometimes being out by several minutes. :(

[jeffoest]

Yea, I was the big VBR guy if you remember. But I got tired of the emails from people saying that they had various problems with it. I went back to CBR.

[gozar]

VBR-encoded files may sound better, but I've had problems with the pause function on both an iPod and an iRiver when playing these. Neither player seems able to resume from exactly the same place it was paused, sometimes being out by several minutes. :(

I have an first generation iPod, and there is a podcast I listen to that is VBR and pause does not work right. Whenever I pause, it jumps a head a minute or two, making it almost impossible to listen (if I have to pause it). I e-mailed the podcaster, and he's going to change for his next podcast.

I'm pretty sure that Apple isn't going to fix the firmware on this old of an iPod. :-)

[PaulofCthulhu]

We've always used:

Adobe Audition
MP3Pro setting
32kbps@22(44Khz)

Talks shows with some songs.

Best bang for the buck, or we'd be blowing out our bandwidth every month.

Paul

[PaulofCthulhu]

Mind you, we had a crack German audio engineer go through the podcast and he came back with a detailed six page report on what was wrong with it (graphs 'n all) - frightening.

Still, effective for the bandwidth (and we keep 64KBPS masters anyway).

Paul

[defiradio]

Mine is a music show and I've found that 128, 160, or 192kbps at 44.1kHz Sampling is the best for music. I use Adobe Audtion sometimes and othersI use Audacity which uses LAME. If you set Audacity up to use LAME there really is not much of a difference between Auditions 160kbps/44.1 and Audacity's.

[womengrow]

The fact that iTunes compression was pointed out to be fairly lousy on another thread.
I actually think iTunes does a decent job with a fair amount of control over the MP3 compression. And it's free. If you need more options, try Quicktime Pro.

[SFEley]

The fact that iTunes compression was pointed out to be fairly lousy on another thread.
I actually think iTunes does a decent job with a fair amount of control over the MP3 compression. And it's free. If you need more options, try Quicktime Pro.
LAME is also free and a better MP3 compressor than iTunes.

[audio2u]

LAME is also free and a better MP3 compressor than iTunes.

Hear hear!
And for low bitrates (<128kbit), LAME performs better (read:less artifacts) than the Fraunhofer CODEC.