Podcasting 101, please June 29, 2012 9:24 AM Subscribe
It's been a while since this has been asked, and I think more folks than me would benefit from it (hence me doing a MeTa rather than using the contact form): Could the podcast team talk about their current setup, hardware and software-wise?
I've launched a podcast at my job, and two episodes in I'm frustrated by what I'm getting in terms of sound quality. It's four of us on a Skype audio-only call, with good bandwidth all around. But I'm having a helluva time getting levels normalized throughout - they seem to spike and disappear all over the place - and my headset seems extraordinarily sensitive to background noise (air conditioner).
So, how do you guys get the quality that you get? Headset mikes, or table-based podcast mikes with separate headset? How do you record? (I'm using the $20 Call Recorder plugin for Skype.) Any other tips, or things you've learned from doing the podcast for so long?
I've launched a podcast at my job, and two episodes in I'm frustrated by what I'm getting in terms of sound quality. It's four of us on a Skype audio-only call, with good bandwidth all around. But I'm having a helluva time getting levels normalized throughout - they seem to spike and disappear all over the place - and my headset seems extraordinarily sensitive to background noise (air conditioner).
So, how do you guys get the quality that you get? Headset mikes, or table-based podcast mikes with separate headset? How do you record? (I'm using the $20 Call Recorder plugin for Skype.) Any other tips, or things you've learned from doing the podcast for so long?
"...and we all record our individual tracks using call recorder. cortex can speak to some of the post-processing."
Yeah, when I read what you wrote it suddenly occurred to me that I think that it doesn't occur to most people that you don't need to use the audio (that one person gets) that's xmitted across the Internet for the podcast — you can have the VOIP discussion but record each individual's mike locally at much higher quality and have the participants send their audio all to one location, to be put together.
Indeed, you're not even limited to using the mikes you're using for the call...although I guess if you can digitize with better equipment then you'd be using that for the call, too. But maybe not. Hell, you could just have each person record their speech simultaneously with higher-quality old analog equipment and then digitize it later and send it. You might do that if you don't have a good mike you can use with your PC but you do have other stuff and are able to digitize what it produces after-the-fact.
posted by Ivan Fyodorovich at 10:04 AM on June 29, 2012 [2 favorites]
Yeah, when I read what you wrote it suddenly occurred to me that I think that it doesn't occur to most people that you don't need to use the audio (that one person gets) that's xmitted across the Internet for the podcast — you can have the VOIP discussion but record each individual's mike locally at much higher quality and have the participants send their audio all to one location, to be put together.
Indeed, you're not even limited to using the mikes you're using for the call...although I guess if you can digitize with better equipment then you'd be using that for the call, too. But maybe not. Hell, you could just have each person record their speech simultaneously with higher-quality old analog equipment and then digitize it later and send it. You might do that if you don't have a good mike you can use with your PC but you do have other stuff and are able to digitize what it produces after-the-fact.
posted by Ivan Fyodorovich at 10:04 AM on June 29, 2012 [2 favorites]
I use a pair of closed headphones and Rode Podcaster that Matt lent me. Matt and Jess both use headsets.
I think you can't really beat a nice condenser mic on a mic stand like that for sound quality, but it's not necessarily a huge difference between that and a decent headset audio wise depending on what else is involved. I don't think the difference jumps out in the mefi podcast, for example, other than that Matt and Jess' headsets seems more likely to peak/distort when they get loud than my mic does.
And being disciplined about keeping your mouth and your mic lined up can be sort of a tricky thing in its own right depending on how you like to sit, what you're doing while you podcast (we do a lot of typing and checking out links while we're recording, so positioning a big mic in front of your face with a little desk-sized boom mic stand can be slightly tricky), how able your are to passively keep yourself in position while talking, etc. And the mic is gonna be a little more expensive than a headset. So it's a mixed bag, maybe not for everyone, clearly not really necessary if you don't want to bother.
Your perceived audio quality for any podcast that has multiple speakers is going to be about as good as the worst person's audio, is one thing to keep in mind. A listener can adjust to and totally put up with bad "long distance call" type audio and sort of forget that it sounds crappy if they're enjoying the content, but if you've got crappy audio on one side of the conversation and pristine audio on the other side that makes for a constant reminder to the listener that some of the audio is pretty crappy. So trying to make sure you've got a plan for upping the minimum quality if you've got rotating guests is a challenge to plan for. Even a shitty headset is better than someone shouting at their laptop's builtin mic.
Software: we use Call Recorder in Skype, each of us making a separate recording (and so each of us owning a copy of that plugin) and then everybody sends me their separated just-them track from that after splitting the Call Recorder track up with some (free, I think?) Split Sides of Conversation utility Matt pointed us to. And then I load each separate vocal track into Garageband and do a little bit of work.
This is kind of huge; if you have more than two people on a call and only have one Call Recorder session, you're stuck with either everybody on one track or splitting it up to your voice on one track and everybody else on the other. That's a huge handicap vs. being able to mix all the different tracks independently: you can't easily adjust relative volume levels, you can't compress/EQ/filter individual tracks, you can't get rid of cross talk with a quick snip of one or another person's chatter. So if you haven't already figured out a way to do this, it's something to look into.
But so, yes: I load the tracks into Garageband and mostly don't do much other than sync them up in time and apply a little bit of filtering. My standard adjustments:
1. pan me a skosh left, Matt a skosh right, leave Jess dead center. It's enough that someone listening in stereo gets a little bit of sense of "space" in the recording where we're positioned like three chairs on the same side of a table, and since Matt and I have somewhat deeper voices than Jess the separation works with that to create a sort of "low, high, low" frequency distribution across that stereo image as well. If you're listening in mono (single speaker or earbud, or you're deaf in one ear, or whatever) it should be subtle enough that everyone is still plenty audible.
- tweak the global volume for each track a little if one of us is overly quiet for whatever reason that time.
- apply some compression to each vocal track to help keep the overall volume of our talking more on the loud side even when we get a bit quieter or mumbly. This is super important! Because people aren't great about being steady in the volume when they're talking, and so you're going to have those quiet bits that are hard to hear under even ideal circumstances, and people often don't listen in ideal circumstances. I keep tweaking this periodically; reading podcast threads will make it clear that getting our vocal audio regularly on the loud side so that it's listenable in a car or whatever is one of the ongoing challenges of mixing the podcast. (Your spiking-and-disappearing problem is possibly people turning their heads away from their mic, partly people being naturally a little inconsistent in their volume level, partly maybe your folks being absentmindedly mumbly/shouty, and partly not compensating for all that with vocal compression.)
And that's usually it. Sometimes if someone has a particularly noisy track that time I'll use a noise gate and maybe a touch of EQ to try and reduce the amount of distraction, but that can be sort of hit or miss.
After, I render down the tracks and send 'em to Matt, and he does the actual editing job on the track contents. He does that in Garageband as well; it's entirely a distribution of labor thing, one of us could (and occasionally does) do the whole schmear.
Any other tips, or things you've learned from doing the podcast for so long?
My main thing is source audio quality. Make it as good as you can when you record, because that makes everything easier later. So, like:
Raise your recording quality to some acceptable floor. Halfway decent mics or headsets. Have people record in quiet environments; make sure people aren't swallowing their mics and causing nasty blown-level distortions; turn off computer speakers so you don't pick up random noises; encourage your co-casters to speak at a reasonable and steady volume and avoid going mumbly. Get separate recordings for each speaker so you can mix well.
If you are able to survive and function while doing so, turn off the A/C for the duration of your recording. Persistent low-level background noise in a recording is just shooting yourself in the foot.
posted by cortex (staff) at 10:07 AM on June 29, 2012 [14 favorites]
I think you can't really beat a nice condenser mic on a mic stand like that for sound quality, but it's not necessarily a huge difference between that and a decent headset audio wise depending on what else is involved. I don't think the difference jumps out in the mefi podcast, for example, other than that Matt and Jess' headsets seems more likely to peak/distort when they get loud than my mic does.
And being disciplined about keeping your mouth and your mic lined up can be sort of a tricky thing in its own right depending on how you like to sit, what you're doing while you podcast (we do a lot of typing and checking out links while we're recording, so positioning a big mic in front of your face with a little desk-sized boom mic stand can be slightly tricky), how able your are to passively keep yourself in position while talking, etc. And the mic is gonna be a little more expensive than a headset. So it's a mixed bag, maybe not for everyone, clearly not really necessary if you don't want to bother.
Your perceived audio quality for any podcast that has multiple speakers is going to be about as good as the worst person's audio, is one thing to keep in mind. A listener can adjust to and totally put up with bad "long distance call" type audio and sort of forget that it sounds crappy if they're enjoying the content, but if you've got crappy audio on one side of the conversation and pristine audio on the other side that makes for a constant reminder to the listener that some of the audio is pretty crappy. So trying to make sure you've got a plan for upping the minimum quality if you've got rotating guests is a challenge to plan for. Even a shitty headset is better than someone shouting at their laptop's builtin mic.
Software: we use Call Recorder in Skype, each of us making a separate recording (and so each of us owning a copy of that plugin) and then everybody sends me their separated just-them track from that after splitting the Call Recorder track up with some (free, I think?) Split Sides of Conversation utility Matt pointed us to. And then I load each separate vocal track into Garageband and do a little bit of work.
This is kind of huge; if you have more than two people on a call and only have one Call Recorder session, you're stuck with either everybody on one track or splitting it up to your voice on one track and everybody else on the other. That's a huge handicap vs. being able to mix all the different tracks independently: you can't easily adjust relative volume levels, you can't compress/EQ/filter individual tracks, you can't get rid of cross talk with a quick snip of one or another person's chatter. So if you haven't already figured out a way to do this, it's something to look into.
But so, yes: I load the tracks into Garageband and mostly don't do much other than sync them up in time and apply a little bit of filtering. My standard adjustments:
1. pan me a skosh left, Matt a skosh right, leave Jess dead center. It's enough that someone listening in stereo gets a little bit of sense of "space" in the recording where we're positioned like three chairs on the same side of a table, and since Matt and I have somewhat deeper voices than Jess the separation works with that to create a sort of "low, high, low" frequency distribution across that stereo image as well. If you're listening in mono (single speaker or earbud, or you're deaf in one ear, or whatever) it should be subtle enough that everyone is still plenty audible.
- tweak the global volume for each track a little if one of us is overly quiet for whatever reason that time.
- apply some compression to each vocal track to help keep the overall volume of our talking more on the loud side even when we get a bit quieter or mumbly. This is super important! Because people aren't great about being steady in the volume when they're talking, and so you're going to have those quiet bits that are hard to hear under even ideal circumstances, and people often don't listen in ideal circumstances. I keep tweaking this periodically; reading podcast threads will make it clear that getting our vocal audio regularly on the loud side so that it's listenable in a car or whatever is one of the ongoing challenges of mixing the podcast. (Your spiking-and-disappearing problem is possibly people turning their heads away from their mic, partly people being naturally a little inconsistent in their volume level, partly maybe your folks being absentmindedly mumbly/shouty, and partly not compensating for all that with vocal compression.)
And that's usually it. Sometimes if someone has a particularly noisy track that time I'll use a noise gate and maybe a touch of EQ to try and reduce the amount of distraction, but that can be sort of hit or miss.
After, I render down the tracks and send 'em to Matt, and he does the actual editing job on the track contents. He does that in Garageband as well; it's entirely a distribution of labor thing, one of us could (and occasionally does) do the whole schmear.
Any other tips, or things you've learned from doing the podcast for so long?
My main thing is source audio quality. Make it as good as you can when you record, because that makes everything easier later. So, like:
Raise your recording quality to some acceptable floor. Halfway decent mics or headsets. Have people record in quiet environments; make sure people aren't swallowing their mics and causing nasty blown-level distortions; turn off computer speakers so you don't pick up random noises; encourage your co-casters to speak at a reasonable and steady volume and avoid going mumbly. Get separate recordings for each speaker so you can mix well.
If you are able to survive and function while doing so, turn off the A/C for the duration of your recording. Persistent low-level background noise in a recording is just shooting yourself in the foot.
posted by cortex (staff) at 10:07 AM on June 29, 2012 [14 favorites]
you don't need to use the audio (that one person gets) that's xmitted across the Internet for the podcast — you can have the VOIP discussion but record each individual's mike locally at much higher quality
Exactly, yes. That's part of why we do the Call Recorder thing at each end, split off the individual tracks, and re-assemble locally. Your recording of your audio will be more or less pristine; my recording of your audio will be a lot less so.
Hell, you could just have each person record their speech simultaneously with higher-quality old analog equipment and then digitize it later and send it.
Questionable approach, if only because slight drive-speed issues will be amplified by long running times to the point where two parts of a conversation recorded on two different analog decks could end up out of sync with each other at the far end of an hour-long conversation. If Alice is even two seconds ahead of Bob after an hour, the conversation will sound weirdly disjointed at best with Alice constantly biting off Bob's prompts and Bob taking forever to respond to Alice's.
posted by cortex (staff) at 10:13 AM on June 29, 2012 [1 favorite]
Exactly, yes. That's part of why we do the Call Recorder thing at each end, split off the individual tracks, and re-assemble locally. Your recording of your audio will be more or less pristine; my recording of your audio will be a lot less so.
Hell, you could just have each person record their speech simultaneously with higher-quality old analog equipment and then digitize it later and send it.
Questionable approach, if only because slight drive-speed issues will be amplified by long running times to the point where two parts of a conversation recorded on two different analog decks could end up out of sync with each other at the far end of an hour-long conversation. If Alice is even two seconds ahead of Bob after an hour, the conversation will sound weirdly disjointed at best with Alice constantly biting off Bob's prompts and Bob taking forever to respond to Alice's.
posted by cortex (staff) at 10:13 AM on June 29, 2012 [1 favorite]
jessamyn: "we all record our individual tracks using call recorder"
Hey, can we get access these individual tracks so we can remix them?
posted by Plutor at 10:46 AM on June 29, 2012 [5 favorites]
Hey, can we get access these individual tracks so we can remix them?
posted by Plutor at 10:46 AM on June 29, 2012 [5 favorites]
"Questionable approach, if only because slight drive-speed issues will be amplified by long running times to the point where two parts of a conversation recorded on two different analog decks could end up out of sync with each other at the far end of an hour-long conversation."
Is that actually the case with even consumer analog recording equipment after, say, 1980? Maybe a cheap cassette player, I guess. But I was thinking more about people who might have an analog 4-track and a good XLR mike that they can't use directly with the computer.
posted by Ivan Fyodorovich at 11:08 AM on June 29, 2012
Is that actually the case with even consumer analog recording equipment after, say, 1980? Maybe a cheap cassette player, I guess. But I was thinking more about people who might have an analog 4-track and a good XLR mike that they can't use directly with the computer.
posted by Ivan Fyodorovich at 11:08 AM on June 29, 2012
But I'm inclined to believe you, now that I think about it. Your example of error is one part in 1,800 which is pretty darn small for a motor-drive without separate circuity to maintain a precise speed via a feedback mechanism Which sort of reminds me of the whole self-correcting clock thing, which famously flopped because people didn't want to buy clocks that the manufacturer admitted outright were inaccurate. Now we have NTP.
posted by Ivan Fyodorovich at 11:12 AM on June 29, 2012
posted by Ivan Fyodorovich at 11:12 AM on June 29, 2012
I actually use Callburner when I join the podcast because I record on my gaming box (which is not a mac.) It does the split-track thing natively. I have a kind of cheapo headset but it mostly seems to work ok.
I agree that mic discipline is a key thing. I learned it by raiding in WoW for a couple of years, where my guildies would bitch if I was eating the mic or breathing on it or whatever. It's really just a matter of practice and feedback.
posted by restless_nomad (staff) at 12:12 PM on June 29, 2012 [1 favorite]
I agree that mic discipline is a key thing. I learned it by raiding in WoW for a couple of years, where my guildies would bitch if I was eating the mic or breathing on it or whatever. It's really just a matter of practice and feedback.
posted by restless_nomad (staff) at 12:12 PM on June 29, 2012 [1 favorite]
Thanks for the input so far, all. The "turn off the AC" advice is the immediate takeaway for me so far.
In my particular situation, I've got a pool of about 10 people, any three of which will be joining in on any given chat. I'm not averse to spending $200 to get each of them a copy of Call Recorder, but the bigger issue is that not all of these folks are techies (and not all of them are on Macs), so I'm not sure I want to go down the road of getting them to record their portion locally. Any other advice when the situation is such that, I've got to capture the Skype audio and use it? Any way to maximize the quality of it?
posted by jbickers at 12:34 PM on June 29, 2012
In my particular situation, I've got a pool of about 10 people, any three of which will be joining in on any given chat. I'm not averse to spending $200 to get each of them a copy of Call Recorder, but the bigger issue is that not all of these folks are techies (and not all of them are on Macs), so I'm not sure I want to go down the road of getting them to record their portion locally. Any other advice when the situation is such that, I've got to capture the Skype audio and use it? Any way to maximize the quality of it?
posted by jbickers at 12:34 PM on June 29, 2012
Alice constantly biting off Bob's prompts and Bob taking forever to respond to Alice's.
A lot of my conversations are like this.
posted by box at 1:38 PM on June 29, 2012 [2 favorites]
A lot of my conversations are like this.
posted by box at 1:38 PM on June 29, 2012 [2 favorites]
Also I hear the pros just put a pillow over the spittoon when the mic is live. You don't need a special locking lid with a foam cover.
posted by fleacircus at 2:07 PM on June 29, 2012
posted by fleacircus at 2:07 PM on June 29, 2012
Any other advice when the situation is such that, I've got to capture the Skype audio and use it? Any way to maximize the quality of it?
Good internet connections and a little bit of luck is all I know about maximizing the raw audio quality Skype will be receiving. It's not impossible that there's some more robust call-sides-tracking solution out there that will do more than Call Recorder can, but I haven't gone looking.
Other than that, yeah, try and train your callers to stay on-mic and avoid mumbling. A more even signal going in means less fiddling around later.
And even if you can't juice the tracks separately, you can still use some compression and some hand tweaking of volume levels to reduce the amount of mumble-LOUD-mumble dynamics going on; going into Garageband and just nudging up little bits of conversation where someone gets way too quiet is tedious if you have to do a lot of it but will help significantly; applying compression/limiting to the whole several-people-talking track will at least help some as well, though if you've got crosstalk with one person quiet and the other loud you won't really be able to fix those bits. But if nothing else compressing the whole thing will help improve the overall apparent volume level some.
If you do audio inserts (music tracks, etc) as part of the final editing process on the podcast, do your filtering and such to the vocal tracks first and then add that stuff later. Generally you're not going to want to apply any of your EQ or compression stuff to an already-nicely-mastered song track when you apply that stuff to your newly recorded vocals.
posted by cortex (staff) at 2:29 PM on June 29, 2012 [1 favorite]
Good internet connections and a little bit of luck is all I know about maximizing the raw audio quality Skype will be receiving. It's not impossible that there's some more robust call-sides-tracking solution out there that will do more than Call Recorder can, but I haven't gone looking.
Other than that, yeah, try and train your callers to stay on-mic and avoid mumbling. A more even signal going in means less fiddling around later.
And even if you can't juice the tracks separately, you can still use some compression and some hand tweaking of volume levels to reduce the amount of mumble-LOUD-mumble dynamics going on; going into Garageband and just nudging up little bits of conversation where someone gets way too quiet is tedious if you have to do a lot of it but will help significantly; applying compression/limiting to the whole several-people-talking track will at least help some as well, though if you've got crosstalk with one person quiet and the other loud you won't really be able to fix those bits. But if nothing else compressing the whole thing will help improve the overall apparent volume level some.
If you do audio inserts (music tracks, etc) as part of the final editing process on the podcast, do your filtering and such to the vocal tracks first and then add that stuff later. Generally you're not going to want to apply any of your EQ or compression stuff to an already-nicely-mastered song track when you apply that stuff to your newly recorded vocals.
posted by cortex (staff) at 2:29 PM on June 29, 2012 [1 favorite]
Offtopic:
Any chance of releasing future podcasts at a lower bitrate in addition to the higher quality one? Not sure if there is much demand for it, but the files are pretty large.
posted by EsotericAlgorithm at 4:06 PM on June 29, 2012 [1 favorite]
Any chance of releasing future podcasts at a lower bitrate in addition to the higher quality one? Not sure if there is much demand for it, but the files are pretty large.
posted by EsotericAlgorithm at 4:06 PM on June 29, 2012 [1 favorite]
Heh.
I skipped the [more inside] and the [title] and the [everything else, really], except the deck and was expecting
posted by notyou at 4:07 PM on June 29, 2012
I skipped the [more inside] and the [title] and the [everything else, really], except the deck and was expecting
Well for me here at the home office it's just a laptop and a wireless keyboard sitting on a desk I made out of an old door and some breeze blocks. When PB gets here he'll tell you about the tinker toy Turing machine and the sled dogs that do the real work...Headsets and mics are cool, too.
posted by notyou at 4:07 PM on June 29, 2012
Using something like Mumble rather than Skype should give better audio quality and be painless to set up even for not techs.
posted by onya at 1:26 AM on June 30, 2012
posted by onya at 1:26 AM on June 30, 2012
Just looking around for the recording software I'd used a few months ago, I found the "Podcast Capture" application under Apple's standard Utilities folder. Hallelujah!
posted by nataliepo at 3:00 PM on July 13, 2012
posted by nataliepo at 3:00 PM on July 13, 2012
On second thought, that doesn't work really, really well. Go with what everybody else said.
posted by nataliepo at 7:59 AM on July 18, 2012
posted by nataliepo at 7:59 AM on July 18, 2012
You are not logged in, either login or create an account to post comments
posted by jessamyn (staff) at 9:27 AM on June 29, 2012 [2 favorites]