LAME is still the best MP3 encoder, but further tests and conversations reveal that it just may not be the best for podcasting.
Variable bitrate (VBR) MP3s and LibSyn
Bob Jordan wrote in to point me to LibSyn's PDF on their Podcast Companion App, which reveals an incompatibility with VBR MP3s.
Avoid using VBR (Variable Bit Rate Encoding) as it is not compatible with our system.
I exchanged several emails with top people at LibSyn and they confirmed that VBR files are absolutely not compatible with the iOS app they offer as part of their larger hosting packages. Extensive study revealed a few portable digital audio players that have trouble with VBR, and even some minor issues in iTunes.
LAME versus Fraunhofer
As I explained in more detail in episode 6, LAME is the best MP3 codec for quality, that is, when you use it with variable bitrate (VBR). LAME is terrible at constant bitrate (CBR). However, Fraunhofer, the creator of MP3 compression, has an MP3 encoder that is the best at CBR.
Fraunhofer's MP3 encoder is not free—it's usually included in expensive audio-editors like Adobe Audition. However, Fraunhofer IIS's own site tells how you can get their MP3 encoder for free.
If you are an end user and would like to use the Fraunhofer mp3 encoder or decoder, please use Apple iTunes or Windows Media which integrate the Fraunhofer mp3 software. Please note, that although mp3 was developed at Fraunhofer IIS, we do not sell any mp3 products to end users and do not provide end user support for mp3 devices and software. [Emphasis added]
I've known that Fraunhofer was credited in iTunes, but didn't realize until reading this that iTunes does use his actual MP3 encoder, not just parts of his technology. So if you don't already have iTunes, download it now for free.
Encoding MP3s: convenience and potential incompatibilities vs. speed and hassle
The method I'm about to show you for encoding Audacity projects into MP3 via iTunes is more of a hassle, with several additional steps and some cleanup. But when you have everything set right, it may be a whole lot faster.
45-minute stereo podcast in Audacity on a Core i7 MacBook Pro:
- Export from Audacity as LAME, VBR quality 5, joint stereo: 8 minutes, 10 seconds
- Export from Audacity to uncompressed WAV, use iTunes to convert to 128 kbps CBR, joint stereo: 2 minutes
Although it's a little more of a hassle to use iTunes, the above test revealed a 75% savings in time and produces a file guaranteed to work everywhere.
How to make an MP3 from Audacity and iTunes
1. Export as WAV
In Audacity, go to the File menu and click Export, name your file, change your format to “WAV (Microsoft) signed 16-bit PCM,” and then click Save. There are no options for WAVs
2. Drag into iTunes
I suggest for creating a playlist or smart playlist to hold your new files. But however you drag the WAV into iTunes is fine.
3. Configure iTunes for MP3s
Go to the Edit menu (Windows) or iTunes menu (OS X) and click Preferences, click Import Settings… under General, set Import Using to “MP3 encoder,” Setting to “Good quality (128 kbps),” then click OK and OK.
4. Convert
Right-click your imported WAV and click Create MP3 Version.
5. Find the MP3
Search your iTunes library for the new MP3 file, drag it out of iTunes back into your folder, and now it's ready to be tagged and uploaded. You can also delete the uncompressed WAV and delete the files from iTunes if you want.
Although this is many more steps and leaves a little cleanup behind, it still took only a quarter of the time to use iTunes than to encode directly from Audacity. It's still free, and this guarantees compatibility and high quality with everything.
This is where my friend, Cliff Ravenscraft of Podcast Answer Man, can say, “I told you so,” and ignore our conversation about all of this in his episode 125 in which I defended and explained LAME.
Taking a live-show break after August 25
I'm getting married in a short time! Although I will have prerecorded episodes still released on a schedule, Wednesday, August 25, will be the last live recording for a while. But you can still catch that live recording at 8:00 p.m. (ET) on our Live page. After I'm married, I may rearrange the Ramen Noodle™ and The Audacity to Podcast™ to be live on the same night.
Need personalized podcasting help?
I no longer offer one-on-one consulting outside of Podcasters' Society, but request a consultant here and I'll connect you with someone I trust to help you launch or improve your podcast.
Ask your questions or share your feedback
- Comment on the shownotes
- Leave a voicemail at (903) 231-2221
- Email feedback@TheAudacitytoPodcast.com (audio files welcome)
Connect with me
- Subscribe to The Audacity to Podcast on Apple Podcasts or on Android.
- Join the Facebook Page and watch live podcasting Q&A on Mondays at 2pm (ET)
- Subscribe on YouTube for video reviews, Q&A, and more
- Follow @theDanielJLewis
Disclosure
This post may contain links to products or services with which I have an affiliate relationship and may receive compensation from your actions through such links. However, I don't let that corrupt my perspective and I don't recommend only affiliates.
Related Content
For a while I've been using iTunes to encode the mp3 files for the Airplane Geeks podcast, and I'm very happy with that. Previously I used LAME in Audacity.
My switch came about after I started thinking about bit rates when I noticed the podcasts I subscribed to were coming through at a variety of bit rates. I even sent Cliff a question on that which he discussed on a Podcast Answerman episode. His response left me with a few questions, so I did some research and ran across the LAME vs. Fraunhofer debate. (If you search for the two you'll see that there are very strong feelings on either side.)
One article demonstrated the difference between the two encoders at a variety of bitrates and showed that Fraunfofer had higher audio quality at lower bit rates, and there wasn't much difference at higher bit rates. That single study didn't silence the controversy, but since my (mostly speech) podcast is single channel mono at 64kps, I decided to try Fraunhofer. Besides, that's the guy who invented mp3 so I figured he ought to know how to encode the stuff!
Like you, at the Fraunfofer website I discovered you can't buy the encoder, but you can use it through iTunes as you described in TAP010. (I passed that on to Cliff, who was unaware at the time.) Anyway, the workflow using iTunes to encode the mp3 is simple enough and it works well for me.
Thanks for the great podcast, and congratulations on your marriage!
Max
Great show as always! In the beginning of our podcasting network we used Ubercaster to create our MP3's which utilizes the Lame encoder. The files were created in VBR until we eventually transitioned to Adobe Audition and began using CBR. We ran into the same situation with our iPhone app where older episodes would not work properly; however, in our case there wasn't a big impact since they were older shows. It is unfortunate these limitations exist across platforms since it takes away the control, design and decision making away from the content creator. I appreciate the plug for Waves of Tech but most importantly if anyone requires a "Subscribe and follow us here" widget, Daniels design is functional, eloquent and his work is above reproach.
(20 weeks later, but I didn't want to leave this unreplied.)
You can expect that almost any free program will use LAME, since it's leading (and best) free MP3 encoder. Adobe can afford the licensing for Fraunhofer because Audition is so stinkin' expensive. I have sometimes considered re-encoding my old episodes, too. But no one has complained yet.
(20 weeks later, but I didn't want to leave this unreplied.)
Thank you, Gene! This does remind me to do an MP3 tagging episode, unless you feel I've covered that enough within an episode since August.
Daniel, I got here via a tweets between you and @nigelrunner and I can tell you that the time to export an .mp3 via Soundtrack Pro (which also uses LAME, shockingly) also suffers from very long encode times. I've always used iTunes for my encoding but thought you might find it interesting that it's the same elsewhere. Very well done episode on a sticky topic;) With Apple's use of LAME in a pro app like STP, you're point about it being best for music is validated even more. Cause otherwise it would certainly be bizarre for them to be using LAME in STP when you get if free with iTunes. But on the flip side, you have to wonder why they use FH to encode at all since the encoding .mp3's almost certainly wasn't designed to encode podcasts specifically. Interesting stuff.
Fraunhofer, LAME, AAC, and Ogg Vorbis have been in quality-to-size wars for a while. WMA is like the annoying little brother who asks to play, too, but only wants to play Candy Land.
[…] and using LAME any other way produces poor quality MP3s. That’s why I recommend that you export as WAV and create your MP3s with iTunes.3. Audacity can’t make musicAudacity can import MIDI files, but it won’t be a friendly […]
CONGRATS!
I'm guessing this was for my wedding? I've now been married a year, so thank you! 🙂
[…] the preferred method for music, but may produce incompatible podcast files. Instead, I recommend exporting to WAV and using iTunes to create you podcast MP3s.Please support our contentWe have expenses for our podcasts. Please look at these options for how […]
“LAME is terrible at constant bitrate (CBR)” – That is just plain false! It is not terrible. Get a decent, fast (preferably x64) build from rarewares.org and it works just fine! Sounds great, too!
For clarification, terrible at the low bitrates recommended for podcasting, like 128 kbps stereo or 64 kbps mono. Fraunhofer outperforms LAME in quality.
But the process you describe is also more complicated for podcasters.
How do you know? Did you encode several MP3’s of each and do a blind ABX test? Where is the actual hard data on this? I still disagree; LAME produces mp3’s just fine at low bitrates, mono, whatever. How could it possibly be MORE complicated than using Audacity’s built-in LAME compatibility, as opposed to opening a whole other piece of software just to do one step of encoding? You said yourself that this method takes extra time, so how is it more complicated to use an updated LAME build? Bah.
I can see this is going to be a slow conversation and may take another year for a reply. 😛
No, I didn’t do an ABX test, but others have so I’ll let you find and reference their data. I did human listening tests: LAME CBR compared to Fraunhofer CBR. LAME didn’t sound as good. However, LAME excelled when I switched it to VBR. But LAME VBR from Audacity actually took more time than exporting from Audacity to WAV and using iTunes to encode to Fraunhofer CBR. Besides this, VBR is not good for podcasting as it has many compatibility problems with software and hardware players.
What makes your suggestion more complicated is that you’re suggesting someone go out and find the right LAME build and manually plug that into Audacity. That’s far more complicated than the standard LAME for Audacity installations, and more complicated that using iTunes, which most podcasters should already have.
Ok so I was not ‘flattening’ to mono before I exported to wav and used iTunes custom settings to change it to mono. In powepress I now get this error Warning, Media URL https://traffic.libsyn.com/digitalscrapbooking/SIP014_Kelleigh_Ratzlaff_Papercraft_Designer.mp3: Sample Rate 16Khz may cause playback issues, we recommend 22Khz or 44Khz for maximum player compatibility. Should I worry!?!
Hi, Melissa!
44.1 KHz is the standard for audio podcasts. Anything other than 22 or 44.1 will sound like chipmunks in some players.
You can change your project rate in the lower-left corner of Audacity, or change it in whatever other program you’re using to record.
Thanks Daniel – I’ll check it out and try to fix it for next episode! 🙂
[…] iTunes is free software for Windows and OS X and it uses the Fraunhofer software for making MP3s. Fraunhofer is better at CBR, so we’ll use iTunes to make our podcast MP3. […]
I just want to say thanks! Albeit I don’t comment here often, I keep coming back to it to check your tutorial and tips! They are simply great! Thanks again!
Thank you for coming back!
Audacity offers several choices for MP3 encoding. VBR, CBR, and ABR : “Average bit rate”.
I read posts indicating Fraunhofer fixed bitrates are closer to ABR than pure CBR, which could explain how differences compared to Lame could be so strong.
Did you try Audacity with that Average Bit Rate option ? That could be a lot easier than going through itunes, and cheaper than buying Audition.
LAME (used by Audacity) produces a better quality file with VBR. But CBR is better for podcasting as it’s more compatible. ABR and VBR would both introduce compatibility problems.
So we have to look at this strictly sticking with CBR. With that requirement, Fraunhofer performs better than LAME.
Faithfully,
Daniel J. Lewis
I tried that, but for me, LAME is better with spoken word. I get sharp ess and tinny sound with the itunes conversion.
Original is a 96 kbps stereo mp3 file with very good sound (though I’m not doing the recording, so I don’t know details) converted to mono in audacity and exported for itunes as wav or aiff (would that matter, lossless is lossless?)
Conversion via iTunes 11.1.5 with import settings at 96 kbps stereo/48 mono
vs. Export from Audacity 2.0.5 as 48 kbps mp3 constant bit rate.
Macbook pro with 10.8.5
weird.
ps LAME 3.98.2
That really surprises me because it’s at these lower encoding rates that the differences come out the most.
How is your original file only 96 kbps stereo? I suggest increasing your original recording quality to as high as you can go.
Reading how everybody recommends to not go below 128 mono, I don’t get it either, why the quality of the original is as good as it is. Maybe the good mic.
(commonly recommended 128 kbps stereo not mono)
Yes. The higher quality the microphone and audio chain, the more destructive low compression rates are.
For example, if you record a telephone call, it’s already extremely low quality. So you won’t notice much (if any) difference between 32 kbps mono and 12 kbps mono.
But 64 kbps mono is a great standard for voice-podcasters. It sounds great with any microphone. Sure, it’s not as good as 96 kbps mono, but the differences are small enough that few will notice.
[…] The Audacity To Podcast #10 […]
[…] https://theaudacitytopodcast.com/tap010-audacity-and-itunes%E2%80%94making-not-lame-mp3s/ […]
Hallo Daniel!
These days many people claim that the Fraunhofer CBR encoder sounds much better than LAME CBR.
However, I did not find any files yet, where I could really hear that one or the other is “better”.
Do you maybe know about any comparison or did you try to create some files yourself? I would be very interested in hearing a difference somewhere …
The only comparison I could find is from the following blog post:
http://www.richardfarrar.com/which-is-the-best-mp3-encoder-for-podcasts/
There it is really possible to hear a difference between the LAME and Fraunhofer versions (with headphones):
both LAME versions (CODEC-1.mp3 and CODEC-2.mp3) include noticeable noise and distortions during speech periods.
However, I don’t know what exact encoder version and settings where used there, but if one encodes the same example with latest LAME 3.99.5 CBR (64kbps, mono), it sounds IMHO much better than the examples in the blog post:
https://auphonic.com/media/audio-examples/CODEC-4.mp3
(exact command: lame -h –cbr -b 64 CODEC.wav CODEC-4.mp3)
One could also say that it sounds even “better” than the Fraunhofer example:
“was used to encode this file” (4sec in CODEC-3.mp3) contains quite some musical noise in the Fraunhofter version, which is not present in the LAME version.
Also in your MP3 version of this podcast (tap010.mp3), it is possible to hear quite a lot of musical noise.
(Disclaimer: there will always be some kind of artifacts at such low bitrates and it also depends very much on the content)
Please let me know if you have any further references, examples or thoughts!
I would be happy if I could hear the difference myself – so that I can use the best possible encoder 😉
Thanks for your great work,
LG
Georg
When I did these tests, I think Audacity was using LAME 3.98. I think you’re right that it’s time to retest the assumption. Back then, I could hear differences when I used the Heil PR40. I’ll retest this and post some results soon.
Thanks Daniel, I would be very interested in such files!
Please also post the original source WAV files, so that the test keeps reproducible.
Please note: in some versions of iTunes you will need file | convert in the main menu instead of right-clicking the actual file, in order to convert to MP3.
Otherwise this is not a hassle at all, in comparison with everything else that needs to be done to get the audio files ready! Thanks for that, I had no idea about the Lame and VBR issues!
Sacha Horowitz
The What Is Being Human podcast
One more perfect tip. My exported Audacity files are getting noisy at first, do you know how to fix the problem?
And again, thanks for the tip … That’s why I’m always here!
Interesting stuff, I have to say that I love to check your tutorial and tips, very informative.
I just want to say thanks for sharing!
[…] For more about WAV, MP3s, and the LAME encoder read this in-depth article and podcast by Daniel J. L… […]
very good your content I liked it quite worth !!
http://comprasseguras.com.br/