How the AppleInsider podcast is recorded and edited
As the AppleInsider podcast approaches 160 episodes, one of the most common questions we still receive is how the show itself is produced. Our host Victor Marks offers a rundown of his process, including the hardware and software used to bring the weekly news program to life.
The official AppleInsider podcast regularly ranks among the top 10 technology-related podcasts available on iTunes, and has been a weekly production for over a year now.
If you're a regular listener, you know that the sound quality of the podcast hasn't always been as listenable as it is today. With that in mind, let's talk about what we've settled on for making a decent quality podcast.
At this point we're almost done, but sometimes there are unnatural pauses or things we would rather leave unsaid on the podcast.
I'll import the audio into ferrite on iOS. Ferrite is an audio editor that allows me to detect the silence, fine-tune how aggressively I want to detect the silence and then using a "tighten" option, will shrink the audio across the whole podcast. The "tighten" feature is going to appear in a future app release, and will only be available if you purchase one of the in-app upgrades.
I use iZotope RX5 for cleaning up audio (reverb, extra background noise) and intend to use RX6 soon, which adds a de-essing plugin among other changes. One of the great things about the RX software is that iZotope publishes a cookbook website to aid people in solving immediate problems, with examples addressing how to remove a siren, cat meow, or other noise. It's difficult, but sometimes the only place to meet people is at a caf, and you just have to live with the background noise. We do our best to try and minimize it, and RX is a huge help.
Then we edit in Audacity. I just haven't gotten comfortable using an iOS-only editor for adding in intro/outro music, or making precise edits. This is partly due to my using Ferrite on iPhone rather than a larger screen like iPad, and partly down to the maturity of the tools and my skill with them. I am focused on getting episodes out rather than being patient with touch input.
The last step to making the podcast ready for publication is to add chapter markers. Chapter markers work well in Marco Arment's Overcast app, and while we don't make chapter markers for every episode, it's something we've done for a few episodes by listener request.
The feedback you give us really does make a difference. We've tried using Audacity for chapter markers, and settled on Thomas Pritchard's Podcast Chapters for Mac as the best solution.
Zencastr is the project of Josh Nielsen, and it exists to solve both the audio quality of participants not local to the host, and solve the synchronization problem.
It's not always perfect, in that audio clocks are independent to the computer and not in perfect sync. If you record for long periods of times, the clocks can drift and audio can get out of sync. Zencastr handles this pretty well. They both record the audio of Skype or other voip apps, and also record over it's own VOIP service. Only one participant needs to have a login account for Zencastr, making it easy to use with guests. You login, create an episode, give the link to the guest or other participants, and as soon as everyone's got their mic inputs sorted out (built-in versus external mic), start recording.
One benefit of Zencastr is that even if the VoIP connection is shaky, as long as the connection holds up at all, the local recordings will be good.
There have been times where I've been traveling, and I haven't been able to actually hear the other people on the call and have had to guess at what they've said, but because they kept talking, the local recording cache that Zencastr uses has saved the show. When you stop recording, it uploads everyone's audio to your dropbox, and then allows you to kick off post-production. The post-production used is powered by Auphonic.
Auphonic, made by Georg Holzmann, is many things. It's a website: Auphonic.com. There, you can buy credits for minutes of audio to be processed. You can process a single audio file, or you can upload multiple files of the exact same length (thanks, Zencastr!) and process them into a single file - it will adjust the levels of each, and remove noise from them, making the gain reasonable for our listeners who like to hear us while running, for example.
But Auphonic is also a pair of desktop apps. I've used Auphonic Multitrack when I wasn't aware that Zencastr used it behind the scenes, and before we used Zencastr. The desktop app doesn't have all the features of the web app, but it works offline, and once purchased, it doesn't require the reloading of credits for processing.
The Auphonic web app can take a text file of chapter markers generated by an audio editor like Audacity to make podcast chapters. The AppleInsider podcast used this feature once, but Audacity is the linux of audio editors, and isn't always the easiest thing to work with. It's good, but it's got some rough spots. Kudos to Auphonic for being compatible.
All tracks have to start and stop at the same time. Use one person per audio track, and put music in separate tracks. It processes the tracks individually as well as combined, to result in the fastest way to get to a good mix. The desktop app is barebones - you drag tracks in, you choose the target loudness, and select whether or not to use the adaptive leveler and crossgate options, and process the audio. There's very little to go wrong.
Being able to process audio when I'm mid-flight is also a great benefit.
Georg is prolific, and Auphonic doesn't stop there. It's also an app for iOS, and acts as a recorder. This app is great, but it doesn't work as well for our podcast needs because we have difficulty getting VoIP audio to use a microphone external to the iOS device. Some of this is because we were using lightning connections, and some of it is because even after getting a headphone jack compatible audio interface, we ran out of patience for trying additional microphones.
The biggest credit of getting our levels right, and right from speaker to speaker, keeping the noise minimal, and cross-gating so we don't speak over each other as much, goes to Auphonic.
We had a few requests for chapters in Overcast. You people who listen are the best, most handsome, and in possession of a taste for the finer things in life that I could ever hope to meet.
I was happy to attempt to put chapters on the podcast, and Thomas Pritchard of Podcast Chapters helped. Pritchard makes the one of the finest applications for reviewing at high speed and adding chapter markers suitable for use with Overcast.
The process is really simple. Import the MP3, being playing it, and Command-N to add a chapter at the appropriate point. Give the chapter a title, optionally a URL, and that's it. Export it, and upload the resulting file to the Web server. It's simple, without a lot of features, but it doesn't need that many - it just does the one thing well. I do like that it allows for playback at higher speeds, so I can listen through it to find the places in the audio that should be chapters that much faster.
Another method of adding chapters is Forecast. I have yet to use it, but I have it installed and am looking into it. The benefits I see here are, long silence warnings (if for some reason I've forgotten to use Ferrite to strip them out), and the ability to mark and export sponsor air checks (short MP3s of just the advertising reads, so that sponsors can hear that their ad was read the way they hoped.) It lacks instructions, and is still very young, but if I can more easily add it to my workflow with few steps, it makes good sense.
Comments
I know podcasting apps are very personal but this is a shout out to Downcast. It also does chapters and I have been using it for years. I listen every week! Keep up the good work!
John
Please continue your awesome work!
one thing I have noticed in the last couple of weeks or so, it is a voice overlapping for several seconds, it is just me or somebody else noticed that?
This just shows you this technology is still too hard to use. This is far more complicated than it needs to be. Back in my youngers years before college I use to do video and audio production and that was far simpler than what is being done for this show.
I like listening to what everyone has to say but, there has been time that the audio quality has been all over the place. Even last week show, for some reason had long pauses between each person conversation.
I used to do radio shows years ago, and we'd do call-in. Telephone networks are 600ohm and have very poor audio quality for trying to actually record a show. This workflow is convoluted but generally gives good results. The last week's show pauses were because I was trying to ease off being as aggressive with the 'tighten' settings in ferrite, so as to avoid our-sentences-beginning-just-as-another-person's-speech-ends. Fine tuning, in everything.
Skype->Zencastr->Auphonic->Ferrite->Podcast Chapters->publish to server->write article to accompany podcast->publish.
If there's demand, I could record a screencast of the steps with all the mouse clicks and taps.
Maybe if you employ one of the many noise reduction tools available equally across each episode then the sound would be uniformly cleaner.
Thats the only criticism I have at this point. Otherwise it's a brilliant podcast.
I understand Jason Snell (of Macworld and The Incomparable podcasts fame) likes Ferrite also, but i recall he uses Logic Pro X to combine tracks and perform noise reduction. Nice to know there are now so many high quality yielding options for podcast production!
Another podcast I listen to is MacBreak Weekly, with which I have a love/hate relationship. They have a fun group, but one or two too many. Way too much crosstalk and Leo has the annoying habit of cutting off his co-hosts just when they are about to say something I want to hear! If it wasn’t for René and Alex I would quit listening, because Andy and Leo are more into Android than Apple. A damn shame because Andy used to be an Apple journalist. But the breadth of knowledge of this team is pretty great, and it’s always lively.
Last is Clockwise which I appreciate for its humor and brevity. MacBreak has gotten too long.
Thanks, and keep up the good work.
(That said, I haven't gotten to first episode yet of actual multi-source podcast creation, though I've experimented with it as proof of concept... so far I've only used it for video and screen content I've published out into the wild. I really need a swift kick in the you-know-what to actually get my podcast launched.)
This is one off my longest-running professional rants. I've even brought it up on this forum before.
Noise reduction can do more harm than good, depending on how it's applied.
After only a few seconds of listening, humans tend to stop noticing noise that is consistent. It's not desirable, but it's less intrusive than constant CHANGES in the background noise, which our brains zero in on with laser-sharp focus. Aggressive noise reduction will cause the listener to notice the background noise (or, more accurately, the change between the "natural" environment and the unnatural quiet) instead of it being filtered out of conscious awareness by the brain's natural processes.
In my broadcast work I used Izotope to reduce the background noise somewhat, but chose to leave some in to achieve a somewhat nosier but much more natural sounding end product.
(I haven't, but I only listen with ear-buds or in the car.... which is probably the case for the vast majority. I only mention that as some audio issues are only obvious listening on some studio monitors or high-end headphones, etc.)