On using Audacity for podcasting

[JeffB]

I occasionally record college basketball games & /or the interviews before/after the game & convert them to .mp3s. It saves space for my small collection & I'm guessing doesn't affect audio quality much in talk radio format as opposed to shooting for high quality music. I occasionally share it with friends who may have missed the game or saw it in person & would like to hear the replays or the pre & post-game commentary & interviews.

I've always used Lame to convert it to the .mp3 format. Audacity does the tagging for me as well.

But I just came across a website devoted to podcasting that has an article:
9 Things Audacity Doesn’t or Can’t Do
http://theaudacitytopodcast.com/tap056- ... -container

What Audacity can’t do

Like any program, Audacity has its limits that can’t be broken. Each of the following would be best done in a different program.

1. Audacity can’t effectively tag MP3s

Look at Audacity’s tag editor and you’ll notice some missing fields, especially the cover art field. Use a different program like MP3Tag (free for Windows) or ID3 Editor ($15 for Windows and OS X) for tagging your audio files.

2. Audacity can’t make good podcast MP3s

LAME is the best MP3 encoder! But it’s the best for music, and only when you use variable bitrate (VBR). This isn’t very compatible for podcasts, 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.

I'm wondering if the above is actually true (still?). I've used Audacity to make those .mp3s without any problems as far as I could tell. Perhaps my computer plays them fine but they won't work in an iPod or other .mp3 player, though.

I did try to use the above recommendations but it was a bit of a pain and then I couldn't tell where iTunes saved the .mp3 file it made. I also have to have extra disk space to store the exported .wav file before eventually deleting it.

Any thoughts, info or recommendations would be appreciated.

[billw58]

Opinions are like noses - everybody has one.

Yes, if you want cover art embedded in your MP3 Audacity can't do it.

The second point seems contradictory - LAME is the best MP3 encoder but iTunes does it better. That said, I always export as AIF and let iTunes convert to MP3, but only because it does it faster. Yes, finding the MP3 that iTunes has produced requires some digging, so probably not the best solution for you. As for disk space, you can trash the WAV file after you've converted to MP3.

-- Bill

PS: Please don't double-post. It just wastes everyone's time. I've deleted your duplicate post.

[JeffB]

Thanks for the reply, billw58.

In the podcast that accompanies that page, the author, Daniel J. Lewis, goes into a bit more detail than the above summary I cut & pasted. He claims that music encoded as variable (VBR?) is better, but for podcasting a constant (CBR) format is better for some reason. He says Audacity uses the former method, whereas iTunes uses the latter.

I'm also not sure about the claim that the tags from Audacity don't work (well?). They seem to work for me when I play the .mp3 file on iTunes on my computer. Maybe it doesn't work on portable devices, however. Or maybe Audacity has improved its tagging capabilities in recent releases, though that article is pretty recent and is discussing the Beta version of Audacity.

One other claim made here or elsewhere is that the conversion to .mp3 should be done in stereo as it can be incompatible with some players otherwise. It was claimed that it didn't take up any more room than a mono recording, which surprised me somewhat. In any event, I've always recorded these in mono, assuming the original AM broadcast was in mono and that it would take up less disk space. It is also a little easier to navigate in Audacity as I often end up with quite a few tracks as I usually stop the recording for commercial breaks. I also edit out some of those commercial breaks if I'm too slow to stop it in time or am out of the room etc. It would entail a bit more scrolling up & down if they were in stereo, though that probably isn't such a big deal.

I believe they said that you can convert it in stereo mode even if the original is in mono, however. I just tried it & noticed that there is an options button that lets you select either constant or variable bit rate & also the channel as "stereo" or "joint stereo". That would seem to negate his contention that Audacity / Lame wouldn't do as good a job of conversions to .mp3 for podcasting if I understand it correctly. But I usually get a pop-up in Audacity saying "Warning. Your tracks will be mixed down to a single mono channel in the exported file". I wonder why that is, if the options for .mp3 export has "stereo" ticked?

What is the difference between "stereo" & "joint stereo"? I assume that's not some sort of hippie thing.

[kozikowski]

The encoder stops for a doobie between the left and right channels.

Stereo is discrete left and right channels. You could, if you wish, have completely separate shows on left and right and the system wouldn't care.

Joint Stereo cares. Joint Stereo is two channels, but they're not left and right. They're sum and difference. To torture the arithmetic a little, sum just adds up left and right. Done. Difference is magic. There is only a difference if left and right didn't match.

So if you started out with a stereo show where the majority of the work is in "the middle," basically mono, then joint stereo is far more efficient than just stereo. Of course, the player has to know what that is.

Stereo FM is Joint Stereo. Mono receivers listen to the sum signal and ignore everything else. Stereo receivers use both signals to push the violins to the left and the trumpets to the right.


Variable Bit Rate is a little magic, too. You can call that Constant Quality.

If you tried to compress a full-on, complex orchestra, there might not be a lot of difference between VBR and CBR -- Constant Bit Rate. But if you needed to compress a tuning fork -- mostly pure, simple sine wave or tone, then VBR would reduce the number of bits needed and the work necessary and the size of the output file. CBR would assign this trivial work to the office assistant and the rest of the bits would stand around drinking coffee and talking about last night's game -- but they wouldn't go home. The size of the file would not change.

Again, a technique for greater efficiency, but the music player has to know what's going on.


Everybody invokes the "lame" package, but if you use lame in its native form and read the instructions, you'd be stunned at the pages and pages of different options and settings. Nobody implements all the tools available. You only get them all if you use the stand-alone package, or know how to program.

Code: Select all

.........

--signed
    Instructs LAME that the samples from the input are signed (the default for 16, 
    24 and 32 bits raw pcm data).
    Required only for raw PCM input files. 

--unsigned
    Instructs LAME that the samples from the input are unsigned 
    (the default for 8 bits raw pcm data, where 0x80 is zero).
    Required only for raw PCM input files and only available at bitwidth 8. 

--little-endian
    Instructs LAME that the samples from the input are in little-endian form.
    Required only for raw PCM input files. 

--big-endian
    Instructs LAME that the samples from the input are in big-endian form.
    Required only for raw PCM input files. 

--mp2input
    Assume the input file is a MPEG Layer II (ie MP2) file.
    If the filename ends in ".mp2" LAME will assume it is a MPEG Layer II file. 
    For stdin or Layer II files which do not end in .mp2 you need to use this switch. 

Koz

[billw58]

The encoder stops for a doobie between the left and right channels.

Stereo is discrete left and right channels. You could, if you wish, have completely separate shows on left and right and the system wouldn't care.

Joint Stereo cares. Joint Stereo is two channels, but they're not left and right. They're sum and difference. To torture the arithmetic a little, sum just adds up left and right. Done. Difference is magic. There is only a difference if left and right didn't match.

Not quite. There are two flavours of joint stereo. http://en.wikipedia.org/wiki/Joint_(aud ... int_stereo

-- Bill

[JeffB]

The encoder stops for a doobie between the left and right channels.

Koz

So that's where we get the term "high fidelity"?

Thanks for the reply. I'm not sure I quite understand it all, but then again I probably don't have to.

I guess for me the bottom line would be whether or not some .mp3 players might have trouble playing back &/or reading the tags from an Audacity/Lame conversion to .mp3.

I was also interested, perhaps, in some day trying to do a Podiobook series http://www.podiobooks.com/index.php and that site, at least, seems to have rather strict rules on how podiocast encoding & ID3 tagging:

Encoding
Bit Rate: 128 kbps
Sample Rate: 44.1 kHz
Stereo Mode/Channel: Joint Stereo

ID3 Tagging
It’s easiest if we provide an example of completed tags.

NAME (also called TITLE): Nocturnal 09
ARTIST: Scott Sigler
ALBUM: Nocturnal
GROUPING: Podiobooks.com
TRACK: 9
YEAR: 2008
GENRE: Podcast
ARTWORK: (included on every episode)

Should Audacity/Lame be able to handle that by exporting a file to .mp3 format? Or is the author in the original post correct that that is out of Audacity's league and it should be exported to a .wav file, then encoded with yet another encoding program?

[steve]

I guess for me the bottom line would be whether or not some .mp3 players might have trouble playing back &/or reading the tags from an Audacity/Lame conversion to .mp3.

Virtually all MP3 players these days can play joint stereo and VBR, though some players still have a problem with displaying the play time correctly for VBR encoded MP3s. For Podcasts the user is quite likely to want to know if they are 30 minutes from the end or only 20 minutes, so for this reason it is probably better to use CBR for Podcasts, even though VBR will usually give better overall sound quality for the same file size. Also, if the Podcast has a very large audience then you can bet that someone will have a crappy old MP3 player that can't handle VBR.

Stereo or Joint Stereo is generally favoured for music, but for a speech only Podcast it is rather a waste. When encoding with CBR (Constant Bit Rate), the number of bits per second is constant regardless of what the sound is. 1 minute of audio at 128 kbps CBR will be a little less than 1 MB regardless of whether it is mono or stereo, but if it is encoded as true stereo then there are only 64 kbps for each channel. If 64kbps gives adequate sound quality, then it would be better to make a mono show and encode it at 64 kbps, which will then produce a file size of less than 0.5 MB (faster to download and cheaper if you have to pay for your bandwidth).

Everybody invokes the "lame" package, but if you use lame in its native form and read the instructions, you'd be stunned at the pages and pages of different options and settings. Nobody implements all the tools available. You only get them all if you use the stand-alone package, or know how to program.

Audacity can use command-line arguments if you select "external program" as the export format. http://manual.audacityteam.org/index.ph ... al_Program
As an example, for a very good quality, mono, highly compatible speech only Podcast you may want something like this:
--cbr (constant bit rate - for compatibility)
-b 64 (64 kbps - good quality for mono)
-a (mono)
--resample 44100 (resample to 44.1 kHz sample rate - for compatibility - not required if the tracks in your Audacity project are already 44.1 kHz)
-h (high quality - qval=2)
-c (mark as copyright)
--lowpass 8 (low-pass filter at 8 kHz)
--strictly-enforce-ISO (comply as much as possible to ISO MPEG specification)

putting this all together:

Code: Select all

lame - --cbr -b 64 -a --resample 44100 -h -c --lowpass 8 --strictly-enforce-ISO "%f"

These values may be tweaked to suit your needs.
For substantially smaller file size (with slightly reduced sound quality) try changing the "-b" option to a lower value such as "-b 32" and drop the lowpass filter to 7 kHz.

(note, I've not included any id3 tag options, but I would recommend using a dedicated MP3 tagger such as the excellent free MP3Tag, or use your favourite media player to ensure that the tags are fully compatible with your player)

The full list of Lame options can be found here: http://lame.cvs.sourceforge.net/viewvc/lame/lame/USAGE

LAME is the best MP3 encoder! But it’s the best for music, and only when you use variable bitrate (VBR). This isn’t very compatible for podcasts, and using LAME any other way produces poor quality MP3s.

Rubbish.
The "Presets" are optimised for music, but the settings in Lame have the flexibility to suit virtually any type of audio encoding. Audacity offers a wide range of settings (as you have seen) including VBR, ABR, CBR, joint stereo, stereo and a very wide range of bit rates. There is however one useful feature missing from Audacity's Lame interface, which is the ability to encode a stereo project into a mono MP3. If you want a mono MP3 then you need to ensure that your Audacity Project is mono before you export (unless you use the "external program" export option with the "-a" switch).

[JeffB]

Thanks for all the help, billw58, kozikowski, & steve.

I'm learning stuff I never even knew existed, which is a good thing I guess. I never realized my recordings might not work in some .mp3 players, though fortunately, that sounds like it's probably pretty rare.

I don't see why the podiobook site wants people to record in joint stereo. It seems like overkill and as steve points out it would save disk space & bandwidth, their protestations to the contrary notwithstanding.

I'm still a little confused, however. When I click on "export selection" option from the file menu, a window pops up asking where I want to save the new .mp3 file. When I click on the "Options..." button it pops up another box with options for selecting Bit Rate Mode, Quality (ie 128 kbps), & Channel Mode (ie. Joint Stereo).
(see attached screenshot):

audacity-mp3-options-screen.gif (100.34 KiB) Viewed 2702 times

Note that it doesn't have an option for "Mono".

But after I click OK & continue on with the export, another window pops up just before it's saved that says it will be "mixed down to a single mono channel". (see 2nd attached file).

mp3.export.gif (10.07 KiB) Viewed 2702 times

Is that because I recorded it in mono? If so, why would I need a warning that it will be "mixed down" to mono during the export process. Why doesn't it list "mono" as an option?

If I record it in stereo could it be exported as a stereo .mp3? It hasn't really been an issue to me thus far as I was intending it to be mono anyway. But if I tried to do the podiobook thing it sounds like they want it in joint stereo. Not a big problem, mostly just curious at this point.

[steve]

As I wrote in my previous post:

There is however one useful feature missing from Audacity's Lame interface, which is the ability to encode a stereo project into a mono MP3. If you want a mono MP3 then you need to ensure that your Audacity Project is mono before you export (unless you use the "external program" export option with the "-a" switch).

How it works is:

When you export a mono show to make an audio file (any format) the exported file will be mono.
When you export a stereo show to make an audio file (any format), the exported file will be stereo.

The only exceptions to this are:

1) If you have changed "Edit > Preferences > Import / Export > When exporting tracks to an audio file" from the default setting of "Always mix all tracks down to stereo or mono channel(s):" to "Use custom mix:" http://manual.audacityteam.org/man/Impo ... ces#export
In this case you will be presented with a choice of channel mappings before each export.

2) If you are using an external encoder and that encoder is set to convert stereo output from Audacity to mono. In this case, Audacity is still exporting a stereo show, but the external encoder can then remix the stereo to mono. (see example in my previous post)

How does the "joint stereo / stereo" options in the MP3 options fit into this?
Those settings only have any effect if the exported show is stereo.
If you export a stereo show, those settings determine which kind of stereo encoding will be used.
If you export a mono show, those settings are ignored because they are only relevant to stereo.

why would I need a warning that it will be "mixed down" to mono during the export process.

When you export audio, all audio tracks are "mixed down".
What this means is that:

• If there are multiple tracks, they will be mixed together,
• Each track will be converted to the "Project Rate" (the sample rate that is displayed in the lower left corner of the main Audacity screen), http://manual.audacityteam.org/man/Selection_Toolbar
• Any "Envelopes" will be applied, http://manual.audacityteam.org/man/Envelope_Tool
• Track Gain and Pan settings will be applied. http://manual.audacityteam.org/man/Audio_Tracks#panel

Assuming that Edit > Preferences > Import / Export > When exporting tracks to an audio file" is set to "Always mix all tracks down to stereo or mono channel(s):", the mix down will be to stereo for stereo projects and to mono for mono projects.

[JeffB]

Thanks for the detailed explanation again, Steve. Much appreciated.

Merry Christmas to all, and a blessed New Year!