linux podcasting with lame
In my research for trying to find the best settings for making my low bitrate podcasts to sund as good as they possibly can get I have found the following....
lame -q 0 --abr 24 --lowpass 8 --lowpass-width 0 --resample 22.050
produced a very good sounding mp3 for podcasting for human voice that was processed with noise reduction in audacity and applied a NAB equilization curve to the whole audio clip before writing out to a 44.1 16 bit MONO PCM wav file.
Lame -q0 -b 64
this was my baseline of acceptable audio quality.. the 6.5 second audio sample has a file size of 52,452 bytes encoded at that sample rate and quality setting. (Note setting to anything other than a Q value of 0 which is highest quality is silly with today's processor speeds)
encoding with the first setting produced a file that was 19,500 bytes in size and had slightly more “twinkle” artifacts in the sound as well as sounding more like a Low end FM radio transmission with the high frequencies filtered out. Increasing the –lowpass 8 to a –lowpass 12 resulted in adding a large amount of “twinkle” into the audio making it more difficult to enjoy listening to. I also tested with adjusting the value up to only 9 and it also added a significant amount of twinkle that makes the file harder to listen to you might get away by increasing it up to 8.2 but I did not find any advantages and i think it might have increased the "twinkle" sound.
So with my calcualtions and the reccomended settings a 30 minute podcast will take up 5.4 meg of space. Certianly near the size of a standard song mp3 encoded at 160 and not a large wait for the download with the bonus of sounding a large amount better than most of the podcasts out there right now. This can be tweaked further... By lowering the –abr setting to 22 I had a very slight increase in “twinkle” but a file size reduction to 4.99meg for a 30 minute podcast. Saving a half a meg every 30 minutes might be important when you start looking at longer podcasts. For example a 2 hour podcast will be 2 meg smaller at this lower bitrate setting.
All of this will create a 64kbps podcast that has an average bitrate of what you set in –abr setting...
Using 100% free tools and a low end pci SB-16 soundcard in a computer that most people would consider unuseable, I was able to produce a low bitrate podcast that in quality and size starts to put everyone else's podcasts to shame in terms of audio quality, hopefully other podcasters will reap the benefits of my research and upgrade the sound quality of their podcasts by using these great free tools and a little bit of learning the command line.
Now to help the person not familiar with lame and too busy to type lame –longhelp and learn the other settings here's a quick tip for you....
--tt 'title of the mp3' this will set the title in the id3 tag
--ta 'artist of the mp3' this set's the artist field of the id3 tag
--tl 'album of the mp3' This set's the album in the id3 tag
--ty '2005' sets the year
--tg 'death metal' set's the genre field of the id3 tag
--tc ' comments go in here' will set the comment field of the id3 tag
--add-id3v2 forces the addition of the v2 id3 tag
-c mark as copyrighted
-o mark as non-origional
you are limited to 30 characters in each tag field. V2id3 tag will allow more but it is incompatable with many mp3 players.
-a flag will downmix from stereo to mono. This is HIGHLY suggested as mono files at low bitrates sound a whole lot better than a stereo one at the same bitrate... Think of it as this... a stereo mp3 is really 2 mono mp3's so your stereo podcast set at –abr 32 will sound like a mono podcast set at –abr 16 ... in other words.... like crap compared to how it would sound in mono. Podcasts for talk shows need to be mono, do not get suckered into making it stereo.
I do not own an ipod, personally I think they are over-hyped and that the Aireo has more feature with it's wifi and fm transmitter built in. But I really like the idea of podcasting. It is another example of how to avoid the mess that is professional broadcasting. with podcasting ANYONE can become a broadcaster with no cost other than the hosting of the podcast files. Granted the tips above are assuming you have a perfect wav file for input into lame. if your recording of your podcast sucks to begin with, the encoding for low bitrate podcasting will only make it sound even worse. Go to radioshack, get a small mixer, a ok microphone AND windscreen for the microphone as well as cables needed to connect it all to the line-in of your soundcard.... to make things even better go to a local guitar center and get a better mixer, bett microphone windscreen AND a p-popper screen as well as a boom mic stand and if you need to spend some cash an isolation mount for the microphone on the stand.
spending less than $300.00 in audio gear can get you sounding like a $20,000,000.00 radio studio on your podcasts..... now you need talent and a fine radio voice... something that seems to be really missing with most podcasts :-)
here's a trick, when you are recording your podcast, talk lower in tone than you normally do.... just a little bit. and be sure you are almost eating the microphone when you talk, 1 centemeter from the screen or windscreen foam is perfect. Now you need to talk a bit louder than you normally do, use the mixer to adjust the level's down to about 95% of maximum when recording. NEVER pause for more than 1 second and never EVER say "ummmm" or "uhhh" or the other stammering and other bad habits that are typical.
Finally, a simple side word on professional sounding broadcasts... profanity usually equals stupidity to most people. Profanity can be used effectively to put extreme emphasis on a strong point but spewing the filth casually only makes you sound dumber than you are and is not impressive to anyone but maybe a 8 year old kid. If you really feel you must say the f word alot to show how cool you are then WARN listeners of the fact that you are intentionally crude and use profanity. Yes, you havethe right to podcast 30 minutes of profanity, but you also have the responsibility to let potential listeners know that fact.
lame -q 0 --abr 24 --lowpass 8 --lowpass-width 0 --resample 22.050
produced a very good sounding mp3 for podcasting for human voice that was processed with noise reduction in audacity and applied a NAB equilization curve to the whole audio clip before writing out to a 44.1 16 bit MONO PCM wav file.
Lame -q0 -b 64
this was my baseline of acceptable audio quality.. the 6.5 second audio sample has a file size of 52,452 bytes encoded at that sample rate and quality setting. (Note setting to anything other than a Q value of 0 which is highest quality is silly with today's processor speeds)
encoding with the first setting produced a file that was 19,500 bytes in size and had slightly more “twinkle” artifacts in the sound as well as sounding more like a Low end FM radio transmission with the high frequencies filtered out. Increasing the –lowpass 8 to a –lowpass 12 resulted in adding a large amount of “twinkle” into the audio making it more difficult to enjoy listening to. I also tested with adjusting the value up to only 9 and it also added a significant amount of twinkle that makes the file harder to listen to you might get away by increasing it up to 8.2 but I did not find any advantages and i think it might have increased the "twinkle" sound.
So with my calcualtions and the reccomended settings a 30 minute podcast will take up 5.4 meg of space. Certianly near the size of a standard song mp3 encoded at 160 and not a large wait for the download with the bonus of sounding a large amount better than most of the podcasts out there right now. This can be tweaked further... By lowering the –abr setting to 22 I had a very slight increase in “twinkle” but a file size reduction to 4.99meg for a 30 minute podcast. Saving a half a meg every 30 minutes might be important when you start looking at longer podcasts. For example a 2 hour podcast will be 2 meg smaller at this lower bitrate setting.
All of this will create a 64kbps podcast that has an average bitrate of what you set in –abr setting...
Using 100% free tools and a low end pci SB-16 soundcard in a computer that most people would consider unuseable, I was able to produce a low bitrate podcast that in quality and size starts to put everyone else's podcasts to shame in terms of audio quality, hopefully other podcasters will reap the benefits of my research and upgrade the sound quality of their podcasts by using these great free tools and a little bit of learning the command line.
Now to help the person not familiar with lame and too busy to type lame –longhelp and learn the other settings here's a quick tip for you....
--tt 'title of the mp3' this will set the title in the id3 tag
--ta 'artist of the mp3' this set's the artist field of the id3 tag
--tl 'album of the mp3' This set's the album in the id3 tag
--ty '2005' sets the year
--tg 'death metal' set's the genre field of the id3 tag
--tc ' comments go in here' will set the comment field of the id3 tag
--add-id3v2 forces the addition of the v2 id3 tag
-c mark as copyrighted
-o mark as non-origional
you are limited to 30 characters in each tag field. V2id3 tag will allow more but it is incompatable with many mp3 players.
-a flag will downmix from stereo to mono. This is HIGHLY suggested as mono files at low bitrates sound a whole lot better than a stereo one at the same bitrate... Think of it as this... a stereo mp3 is really 2 mono mp3's so your stereo podcast set at –abr 32 will sound like a mono podcast set at –abr 16 ... in other words.... like crap compared to how it would sound in mono. Podcasts for talk shows need to be mono, do not get suckered into making it stereo.
I do not own an ipod, personally I think they are over-hyped and that the Aireo has more feature with it's wifi and fm transmitter built in. But I really like the idea of podcasting. It is another example of how to avoid the mess that is professional broadcasting. with podcasting ANYONE can become a broadcaster with no cost other than the hosting of the podcast files. Granted the tips above are assuming you have a perfect wav file for input into lame. if your recording of your podcast sucks to begin with, the encoding for low bitrate podcasting will only make it sound even worse. Go to radioshack, get a small mixer, a ok microphone AND windscreen for the microphone as well as cables needed to connect it all to the line-in of your soundcard.... to make things even better go to a local guitar center and get a better mixer, bett microphone windscreen AND a p-popper screen as well as a boom mic stand and if you need to spend some cash an isolation mount for the microphone on the stand.
spending less than $300.00 in audio gear can get you sounding like a $20,000,000.00 radio studio on your podcasts..... now you need talent and a fine radio voice... something that seems to be really missing with most podcasts :-)
here's a trick, when you are recording your podcast, talk lower in tone than you normally do.... just a little bit. and be sure you are almost eating the microphone when you talk, 1 centemeter from the screen or windscreen foam is perfect. Now you need to talk a bit louder than you normally do, use the mixer to adjust the level's down to about 95% of maximum when recording. NEVER pause for more than 1 second and never EVER say "ummmm" or "uhhh" or the other stammering and other bad habits that are typical.
Finally, a simple side word on professional sounding broadcasts... profanity usually equals stupidity to most people. Profanity can be used effectively to put extreme emphasis on a strong point but spewing the filth casually only makes you sound dumber than you are and is not impressive to anyone but maybe a 8 year old kid. If you really feel you must say the f word alot to show how cool you are then WARN listeners of the fact that you are intentionally crude and use profanity. Yes, you havethe right to podcast 30 minutes of profanity, but you also have the responsibility to let potential listeners know that fact.
Comments
Post a Comment