tl;dr if you followed my 2015 setup — what’s changed since then:
- Beta 87A condenser microphone
- Zoom H5
- Acon Digital restoration
- Fastly to serve audio files
I have two different setup for recording online (Skype) or offline.
For Skype recording, the following setup is used and the recording is done in the closet, where clothes and package boxes help to reduce the room echo.
- Shure Beta 87A ($250)
- Mackie Onyx Blackjack ($150)
- DS7200B microphone stand ($13)
- Ultimate Ears UE200 noise isolating earphones ($18) I’ve been using Beta 57A dynamic microphone with Cloudlifter CL-1 for a long time, but recently have switched to BETA 87A upon Marco Arment’s podcasting mic guide, because I thought my voice with 57A was a bit muffled and the mid was too high with less clarity on the high. 87A is a super cardioid condenser mic, which doesn’t pick up a lot of ambient noise for a condenser.
It’s difficult to ask guests to spend as much money as I do, and to be honest even if they do it’s a bit hard to support the setup like mine, and all-in-one USB mics are much easier to use.
I recommend the guests to buy, or rather I buy and ship from Amazon, the following USB microphones:
- Audio Technica ATR2100 ($50)
- サンワサプライ USBヘッドセット (3400Y) Headsets make it easier to keep the distance between the microphone and mouth reasonable. The downsides are that it’s sensitive to the plosives. I often need to ask them to move the microphone above their mouth, almost near their noses.
Also microphone and headphone are often too close to each other and the mic is very sensitive to pick up the other side of the conversations leaking out from the headphones. There’s no good way to prevent this other than asking them to minimize the volume.
For offline recording, I use the following kit to record offline conversations and stream while recording it.
- ZOOM H5 ($260) + EXH-6
- BETA 57A x2
- BEHRINGER UCA202 I used to use a portable desktop mic stand such as this one but recently ask guests simply to hold a microphone near their mouth. This has a benefit of making it easier for them to keep it near mouth as they move their head around, but has a risk of them actually turning their head away from the mic when they don’t pay attention.
That’s why it is very important to let everyone monitor their own voice, so that they’ll notice it when they do not speak directly to the mics.
Zoom H5 has two line outputs, one for monitoring and one for line-out. I connect my own headphone to the monitoring port, and use Behringer UCA 202 for the line-out port. UCA 202 has its own monitoring port which you can use to connect other headphones. I have a headphone splitter in case I have multiple guests.
H5 has many options when it comes to recording the audio (USB I/F, or record directly to SD) but here’s what I come up with:
- Power the device with USB battery (like Anker)
- Record to SD directly
- UCA 202 to Mac for Skype input This way I can record individual tracks in WAV in a SD card, then use the combined track as a Skype input for streaming.
For more details, read an article I wrote back in May, not much has been changed since then.
- Skype Call Recorder Audacity is pretty lightweight and lets me see the input level in a wave form, which you cannot do with QuickTime player. Audacity has its own input device settings, separate from the system settings, so be sure to select the right input device when recording audio.
Streaming is an important part of my show, since it adds realtime feedback while recording the show, and gives the show more “live” feeling, so that the editing could be minimal.
I have a Mac mini box to stream conversations via Skype. This is done by making a bot running on Mac, piping the Skype output to Soundflower.
For streaming audio piped from Skype, I use Nicecast on Mac and Icecast 2 on the server (linode box in Tokyo).
- Logic Pro X
- Acon Digital Restoration Suite
- Acon Digital DeVerberate
- LAME MP3 Encoder Logic Pro X’s “Strip Silence” is an absolutely necessary feature when editing podcasts. I use the setting “4%, 1 sec, 200ms, 200ms” to find silences to strip.
For most recordings, the only effects I use are Acon Digital Denoise and compressor. For recordings done in a non-ideal room environment, I need to apply Acon Deverberate to reduce the room echo.
Auphonic does a phenomenal job in leveling the final audio, since most of the time, double-ender recordings have loud part and silent part. Compressor could reduce the dynamic range, but leveler is better suited if you have a part that is loud and then quiet elsewhere.
I purchased standalone license for both Auphonic and Auphonic Multitrack, the latter of which could be used for offline recordings to crossgate to remove room echo.
- Jekyll + Linode
- Heroku for dynamic content (supporter)
- Fastly for MP3 I switched from SoundCloud to Fastly for faster downloads and easier uploads and instant purging when necessary (Disclaimer: I work for Fastly).
Soundcloud’s podcasting package is fantastic if you have a mid to large audience for podcast audio since they provide unlimited streaming for a fixed monthly pricing. They also have a nice stats dedicated to the mp3 audio downloads. With Fastly, I use the log streaming feature to get the raw log files, then use my own perl scripts to build up similar stats from there.
I have some other websites and software to build my podcasting system:
- Google Calendar and Sunrise to schedule recordings
- Gumroad to sell transcripts and subscriptions
- OneSignal and Pushbullet for push notifications
- GitHub to prepare show notes
- Harvest and Square for invoicing
- Wufoo for sponsorship inquiry forms
In 2015, the change was minimal but I improved the offline recording a lot by eliminating the heavy Onyx Blackjack and introducing H5. I recorded about 50 episodes a year with this setup, and there was almost no recording that I didn’t like in terms of audio quality.