« Simulated Annealing

The amateur's guide to making a podcast

My first time making it out of the German forest alive

The German forest 🌲

In Jessica Abel’s excellent illustrated guide Out on the Wire, radio producers discuss various aspects of audio storytelling. One of the best stories is told by Radiolab host Jad Abrumad. At the beginning of his career, he was tasked to work on an hour-long segment of Wagner’s Ring Cycle. It’s an awe-inspiring and terrifying piece for even professional musicians to play, “an eighteen-hour cycle of operas that tries to encompass the totality of European art in one work.”

Abrumad, blissfully ignorant of the immensity of the project, quickly learned his mistake when everything started falling apart:

Fast-forward a couple months, I had missed four deadlines, I’m on the verge of getting fired, and I haven’t slept for four days. I had the pressure of ideas that I just couldn’t reach, I had the pressure of being a newbie and talking to people who were very sophisticated. And I had the pressure of deadlines that were going “splat!” left, right, and center.

This feeling of insecurity and overwhelming anxiety is so common in the storytelling process that the Radiolab team has a name for it: the German forest.

german_forest An evergreen forest in Morsbach, Germany.

But making it through this forest is actually an essential part of the creative process, as Abrumad explains:

When I head the Wagner thing on the radio later, I was like, “Whoa, somewhere in the middle of that trauma, I think I found my voice. There’s a real correlation between time spent in the German forest and these moments of emergence.

And to be clear, the German forest changes. That sense of, the work is just too big to put my head around this, how am I gonna do this, that never changes. But what does change is that the terror gets reframed for you, because now, you’ve made it out a few times. You can see over the treetops, and into the future, to where, there you are, you’re still alive there, you’re still alive. So you begin to recognize the German forest for what it is. It’s actually a tool. It’s the place you have to go to hear the next version of yourself.

I can remember clearly how lost I felt when I started the podcast. Not only were there many logistics I had to figure out, like what equipment and editing software to use, but also I knew my storytelling style and voice were nonexistent. I wanted to make a podcast because I love podcasts. I knew the kind of podcast I’d like to make but had no clue on how to get there. I was starting to take my first few steps into the German forest and, let me tell you, it did not feel good.

Now, a year and a half later, I’m officially retiring the San Fransokyo Podcast. It has fulfilled its original mission: documenting our year in Japan and keeping in touch with friends and family while living abroad. I’ve finished fourteen episodes of the podcast and learned so much about podcasting and storytelling. I have successfully made it out of that cursed German forest alive.

Granted, it's only the small, sparsely-wooded forest of a hobby podcast.

With the ending of the podcast, I thought it was a good time to do some reflection. I’ve also talked to quite a few friends who are interested in starting their own podcasts. For one reason or another, they haven’t started on these projects yet. So I thought I could take all the information I’ve collected and turn it into a resource for them. I’m also planning on making another podcast someday, so this will be record and reminder for my future self as well.

Setting up the podcast 🤔💬

Research and planning ✍🏻

Here are some of the resources I gathered over the course of the podcast. In terms of concrete guides, I looked for getting started guides and articles about podcast basics. Here are two I referenced:

I loved learning how experts construct compelling audio stories. I can’t stress enough how highly I recommend Out on the Wire: The Storytelling Secrets of the New Masters of Radio. I also encourage you to read This American Life’s collection of learning resources, Make Radio.

And of course, podcasts themselves are also the perfect medium to hear from the masters directly. I listened to the following episodes for helpful advice and inspiration:

In all honesty, I probably could have done much more prepwork. I was too impatient to produce tangible output and jumped straight to working on the first episode after some cursory Google searching. I stumbled on most of these resources accidentally rather than intentionally seeking them out.

Still, I think it’s better to be impatient than overcautious. It’s easier to dive straight into the German forest and learn through trial and error, rather than trying to get it perfect from the beginning.

Equipment 🎙

All my choices for equipment and software were motivated by convenience, rather than optimizing for the best-in-class tooling. This was a great approach to get me past the initial setup phase, where it can be easy to get stuck due to analysis paralysis.

Here is the breakdown of my podcast equipment:

While this seems like a lot of equipment, I didn’t end up using all of it. For instance, I originally purchased the portable recorder with the intention to record a lot of b-roll background audio. I kept forgetting to bring the recorder along or I would end up not using the audio I did record. I would recommend skipping the portable recorder unless you find yourself really needing it.

The audio mixer was actually the opposite case. I originally purchased one microphone (Blue Yeti USB Microphone) and thought I could use it for all types of recordings. It ended up being suboptimal at capturing high quality audio of two people and I struggled with editing together the audio. Since the podcast ultimately centered on conversations between me and Dru by the third episode, I needed a good two-person setup. Using two USB microphones often results in audio drift, so I finally decided to commit to XLR-mixer setup.

In general, I was surprised by the high price of entry-level podcasting equipment. I think spending a few hundred dollars is probably necessary for decent quality audio, though my setup is still quite sensitive to ambient environmental noise. I would definitely consider upgrading the equipment and investing in soundproofing my recording area for the next podcast.

Editing software 💻🎧

In terms of the editing process, my setup was dead simple:

I learned how to use Audacity a long time ago so it was the easiest editing software to start with quickly. It fulfilled the bare minimum I needed for editing so I didn’t bother learning more professional software like Pro Tools or Logic Pro. I suspect that I could have cut down a lot of my editing time with more advanced software features, although I haven’t done enough research to confirm if this is the case. I definitely plan on investing more time in learning upgraded software the next time around.

audacity The interface for Audacity hasn’t changed since high school.

Music was a pretty defining part of my podcast style, which I’ll go into more detail in a following section. I was able to find great artists and tracks for free, a fabulous outcome considering I was not generating any revenue from podcasting. I am definitely be interested in purchasing licensed professional music for my next podcast, such as the Sono Sanctus music library, which is used to great effect in the Imaginary Worlds podcast.

Hosting and content services 📈🤳🏻

Last but not least, I had to figure out how to host the podcast RSS feed and other miscellaneous content. The services I used were:

I thought Squarespace was a slam-dunk choice, considering that it could take care of both my website and podcast hosting needs at the same time. Squarespace takes care of a lot of the annoying website setup, making it easier for me to get the whole production up and running faster. Plus, I figured that Squarespace had good podcast support since they advertise all the time on podcast shows.

squarespace Taking a deeper look at Squarespace reveals pretty big gaps in their podcasting support.

Unfortunately, Squarespace was not the ideal podcast hosting service. The podcast features are sorely lacking, missing analytics around critical metrics like podcast downloads or listener demographics. Squarespace is primarily a website creation service, so the data they provide is focused on the most useful information to monitor your website growth.

Squarespace does provide RSS subscriber stats, but this number is severely overinflated in the case of podcasting. Starting with episode 5, I added a separate analytics service (Podtrac) which gave me the actual listener data I needed. I wish I had realized this earlier since I don’t have accurate information for those earliest episodes.

rss_subscribers I definitely don’t have ~4k listeners of the podcast.

In general, I don’t think Squarespace is a particular good fit for my needs. My podcast was a finite project, so the $12 per month cost is pretty steep to simply keep up the RSS feed now that I’m not generating any more new content. Plus, as an experienced software engineer, I can build and host a website at almost no cost using free, open-source tools like Jekyll and Github Pages. I’m planning on migrating off of Squarespace and see no benefit to using it for future podcasting projects.

I continue to struggle with the marketing aspects for all of my projects. I tried to copy the social media channels of my favorite podcasts, to little success. I found that the podcast’s Instagram account was the easiest to consistently maintain. I quickly abandoned the Twitter account because I am bad at Twitter in general. I definitely want to invest more energy learning about effective marketing for my future projects, podcast or not.

Closing the gap 🙏🏻

THE GAP by Ira Glass.

Nobody tells people who are beginners, […] all of us who do creative work, we get into it because we have good taste. […] There’s a gap – for the first couple years you’re making stuff, what you’re making isn’t so good. It’s trying to be good, it has ambition to be good, but it’s not quite that good. But your taste, the thing that got you into the game, your taste is still killer. Your taste is good enough that you can tell that what you’re making is kind of a disappointment to you. You can tell it’s still sort of crappy.

A lot of people never get past that phase. A lot of people, at that point, they quit. The thing I would say to you, with all my heart, is that most everybody I know who does interesting creative work, they went through a phase of years where they had really good taste, they could tell what they were making wasn’t as good as they wanted it to be. They knew it fell short. […] The most important possible thing you could do, is do a lot of work.

– Ira Glass, “On good taste”

The original interview with Ira Glass.

Finding my voice 🗣

Establishing a distinct style is the most difficult part of podcasting. Before I started the podcast, I had a long list of podcast shows that I loved but no idea how to synthesize them into something that suited my voice and content.

When I look back, it’s pretty clear how I was course-correcting as each episode came out. I recently listened to the first episode again to understand how far the podcast developed. When juxtaposed against the final episode, it’s clear how polished and tightly-edited the show grew over time. You can clearly hear these changes, among many others:

  • Less stilted and awkward narration
  • Smoother transitions between speaking and music
  • More varied and energetic speaking cadence
  • Increased jokes and humorous tones

Translating these outcomes into some actionable takeaways:

  1. Solo narration is hard: After the first episode, I quickly abandonded the “one person talking”-style of podcasting. Some can do it really well, but I can’t seem to recreate a casual, conversational tone when I’m locked in a closet by myself with a microphone. It was easier to have rapport with another person there, for both unprompted and prompted conversation, and I think the show becomes much more fun to listen to because of this change.

  2. Mold the recorded audio however you see fit: At the beginning, I was loathe to cut or rearrange any recorded audio. It felt like I needed to be true to the original recording process and the natural flow of conversation. Slowly, I realized the final audio sounded much better if I was willing to construct episodes wholecloth, taking pieces of recordings from all over the discussion and mercilessly cutting out uninteresting dead weight. Oftentimes, our discussions would meander and backtrack. Repeatedly circling back to the same topics was fun as a discussion participant but confusing for a listener. I worked on cultivating a mindset that focused on the audience, even during the recording phase:

Never stop thinking about pacing. If an answer seems boring, politely move things along. Charm. Cajole. React with amazement when they say something amazing. Laugh if they’re funny. Don’t forget that YOU are part of the interview.

– Jessica Abel, Out on the Wire

  1. Steal liberally from your favorite podcasts: I was initially worried that it would be obvious which podcasts I was copying, so I avoided it for a long time. But after trying it once, I realized you could hardly tell. So much of a podcast's unique style depends on the voice of the host and the original context. A different person saying the same lines feels completely different. Plus, when I released those experimental intros, no one noticed or cared.

  2. Listen to podcasts you don't like: I honestly found it more helpful to refine my style by listening to "bad" shows versus good ones. I could easily hear things I wanted to avoid in my podcast and adjusted accordingly. I won't namedrop these shows, since to each his own in terms of podcasting style.

In the end, the most effective strategy for developing my voice was just to do a lot of work. Just as Ira Glass said, my output significantly improved by simply finishing and releasing many podcast episodes. I experimented with many different ways of recording, layering in sounds, and cutting together the audio. At every step, I tried to improve the podcast voice just a little bit more, and slowly the show inched closer to a final sound that I could be happy with.

I’m nowhere near done learning about my storytelling style—after all, I’ve only been through the German forest once!—but I’m leaps and bounds better than before.

Musical framing 🎶🖼

Music was the most rewarding aspect of the editing process for me. I personally feel that the right music can elevate an episode from a decent one to something truly great. Getting the music perfect was also important to the show’s content — Japan actually has a very distinct soundscape, and I learned that adding music at the right time made our words more interesting and evocative.

It took me a while to settle on a mood and consistent tone for the music. Eventually the show’s musical identity coalesced around bright, synth-pop. The music also frequently included the use of pentatonic scales and vaguely Asian themes. I was lucky to find an amazing artist Bad Snacks that I used many times, both for transitions and background music.

The single most-used song in the podcast was Mizuki, incorporated in four different episodes.

I also recommend experimenting with Foley sound effects to make the final audio more engaging. Since our episodes were all conversation-based, I used sound effects primarily to delineate when we were recounting a story. This auditory “bookmarking” made it easier to distinguish storytelling sections from the other parts of the episode, such as segments where we debated back-and-forth or cracked jokes.

Here is a shortlist of the five best clips from the podcast which showcase the impact of adding sound effects and background music. I am particularly proud of how perfectly synced the rhythm of our speech is with the beat of the music. You also will notice the editing quality improving across these clips:

I kept returning to the same few artists since it was the easiest way to ensure a consistent sound. These following groups contributed the most to the show’s final musical sound:

Artist # times used
Bad Snacks 19
Aaron Kenny 7
Kevin MacLeod 3
Doug Maxwell 2
Jeremy Blake 2

My favorite compliment about the show was from a friend who said that the final episode had very “professional-sounding music.” It felt great because of how much time and care I spent choosing and layering in the music. Quoting again from the unbeatable wisdom of Abel’s Out on the Wire:

Music is the frame around the picture. It makes it more real than real. More than just two guys talking.

Podcast by the numbers 📶

Last but not least, a summary of the podcast by the numbers. I have already analyzed some podcast data in my 2019 recap post, but I wanted to reexamine the data since that post didn’t include all the episodes.

Editing time ✂️

I was most curious to see how my editing efficiency changed over time since editing was particularly laborious and time-consuming for me. I carefully time-tracked the time spent making the podcast to see if there was any improvement.

The first episode was 17 minutes in length and took 11.5 hours to produce. The time to produce combines the audio recording and editing phases, as well as writing the blog post for the website.

time_spent Time spent per episode. The length of the final episode (blue) is included for comparison.

Editing time always took the most amount of time for every episode. Writing the blog post, including the time spent uploading the audio and updating the podcast RSS feed, always took approximately an hour. Recording time was very inconsistent since we often would record several hours of material, which I would then cut into a few smaller episodes. For instance, we recorded 3.4 hours of audio which was then split into three separate episodes (8-10). The two bonus episodes (B1 & B2) also reused audio from previous recording sessions that didn’t make it into the final cuts for previous episodes.

My editing efficiency did slowly improve over time. The below graph shows my “edit ratio” per episode which shows a clear trend downward. For the final episode, my edit ratio was 17.5—17.5 minutes of editing time for each minute of final recorded audio.

edit_ratio Edit ratio per episode, where edit ratio is defined as editing time divided by final episode duration.

A lot of my increases in time efficiency came with simple skill improvement and knowledge. For example, I spent so much time when I was making the first few episodes just listening to final cuts of the episode over and over. I would listen to the episode inside, outside, in quiet environments, and noisy environments. Listened while I was biking, on the subway, when noisy cars were passing by. That was the only way I could hear all the changes I needed to make to confidently refine the pacing and audio levels.

Slowly over time, it became unnecessary to listen to the final cut so many times. I could adjust the audio properly in the first few passes while I was editing individual segments. By the last episode, I was good enough that I only had to listen to the final audio cut once, just to confirm that I was happy with episode’s overall sound.

biking Biking is a great time to listen to the final draft of an episode.

I also was able to spend less time in the editing phase once I had established the musical style and voice of the podcast. I didn’t have to spend additional time exploring new artists and songs since I already had defined a core set of musical groups that I could reuse for each episode. I didn’t need to listen to other podcasts for inspiration as part of my editing process.

But despite all the personal growth, I continue to be disappointed in the amount of time I spend editing. According to The Audacity to Podcast, a meta-podcast about podcasting, audio podcasting should have a 4:1 work ratio, with four minutes of work for every one minute of content. That implies that I’m at least 5x slower than other podcasters.

I learned that I am a perfectionist about my podcast, willing to sink more time into the process to achieve a higher quality of content. While this obsessiveness has had a positive impact on the resulting sound of this show, I think it’s unsustainable to spend this much time on individual episodes. The sheer amount of time it takes for me to produce the show is holding me back, the most immediate impact being how inconsistent the release cadence was for the show. It’s also hard to commit this much time to a project which I’m not working on full-time, alongside other projects and obligations. Speeding up the editing phase makes it easier to keep podcasting as a long-running hobby.

I recently talked to a coworker who majored in broadcasting for her college degree. She said said that an 18x ratio of edit time to final audio time was actually pretty good. So I might be being a bit hard on myself.

Still, I would like to improve my editing ability. I want to learn new editing software and research more techniques for speeding up the editing process. I also think upgrading my equipment could cut even more time by providing better base audio to work with, removing the need to clean the audio manually.

Number of listeners 👂🏻🙋🏻‍♀️

Starting off, I had low expectations for the number of listeners for the podcast. I would have considered it a wild success if we got more than 20 listeners for an episode. The podcast was primarily a vehicle for us to keep in touch and share our day-to-day experience of living in Japan with close friends and family. It’s honestly more like an audio diary than professional podcast, so I was not expecting it to become popular with thousands of listeners. And spoiler: it didn’t.

But a lovely surprise was discovering there were podcast listeners coming from countries where we didn’t know anyone, like Turkey and South Africa. My favorite type of feedback was always when I was told someone I didn’t know, like a friend of a friend, really enjoyed the show.

audience Distribution of listeners by country.

Country # downloads
US 506
AU 37
JP 23
UK 9
DK 3
TR 2
IT 1
FR 1
ZA 1

In terms of audience numbers, the show started off with consistently around 15-20 listeners. We started gaining more listeners around episode 9 and peaked (146 unique downloads) in the second-to-last full episode: Episode 11: Slow March Into the Grave. This was perhaps the most interesting one since it was our last recorded episode while we were still in Japan, reflecting on the last year of living abroad.

listeners Growth (and decrease) in listeners over time by episode.

Why the sharp dip in listeners after episode 11? I suspect this is partly because a significant portion of our traffic came from social channels, primarily Facebook and Instagram posts. Most of these 150 listeners were probably extended family and acquaintances who weren’t regular listeners of the show, so they didn’t stick around after checking out episode 11.

audience Social media drove 50% of website visitors on months when new episodes were released.

audience Overall distribution of visits to the website.

Release cadence and episode length ⏰📏

Another contributing factor to the decrease in overall listeners likely had to do with the release cadence. I was sticking to monthly-ish episode releases for most of the podcast. But after episode 11, we were so busy with moving back to the US and I didn’t release another episode for four months. I’m not shocked that the number of listeners declined accordingly.

@eriktorenberg, who runs the Venture Stories Podcast, has a great Twitter thread summarizing lessons learned from publishing more than 500 podcast episodes. One of his top pieces of advice is to have consistent release cadence and episode length.

Graphing episode length and release cadence together, you can see both metrics were pretty inconsistent. In terms of episode length, I was originally aiming for around 20-30 minute episodes. As the show progressed and we had more and more interesting content to discuss, the length of episodes ballooned to almost an hour.

episode_length Episode frequency was semi-regular until September, when we started heading back to the US. Episode length also trended upwards.

It was hard for me to cut down these final episodes because I liked how all the segments revolved around central themes. I think the long episodes are very enjoyable as standalone pieces. However, taken together, I can see how annoying high variance in episode length could be as a listener. I definitely plan on sticking to a more consistent schedule for future podcasting projects.

To be fair, while I could have been more disciplined, the irregularities couldn’t be totally avoided. Our time in Japan was filled with many other activities. Aside from other projects I was working on, I was going to Japanese language school five days a week. We hosted various groups of friends and family for 7 months out of the 13 months in Japan. We capitalized on living in Tokyo and travelled as much as we could. Every single holiday (Japan has 50% more national holidays compared to the U.S.) we tried to plan a trip to a different area of Japan, either by ourselves or with our international visitors. In total we travelled 72 out of 396 days, or about 18% of the time we lived in Japan.

days_travelled We travelled outside Tokyo at least a few days every month except in October, when my parents came to visit, and August, our last month in Japan.

Final reflections 🤔📻

If you’ve gotten this far in the article, you might be (justifiably) overwhelmed by the amount of information I’ve regurgitated. I would encourage you to ignore all of it and get started, to just do a lot of work. I started out knowing nothing and made a lovely podcast despite that fact.

I think that everyone thinking about a big life change should consider making audio recordings about the process – it was really fascinating to go back and listen to our mental states at different stages of our time abroad. I’m excited to go back to listen to it again in one year, five years, and dozens of years from now.

You don’t have to turn those recordings into a podcast, but I’d recommend that part too. Even though some days I wanted to scream in frustration at the slowness of editing, I found that the podcast format adds a surprising amount of depth and richness to what would just be “two guys talking” in a room. Adding background, atmospheric music and refining messy discussion into a streamlined story changed the listening experience for the better. It also felt good to learn a concrete skill, making it easier to create a podcast again in the future.

To conclude, I’ll leave you with the words of Jonathan Goldstein, host of Heavyweight. He captures the magic and difficulty of audio storytelling:

There is an excellent way that a hard swallow in someone’s voice, mixed with some swell of just the right music that can be very much like a throat punch to the heart. I think the right combination of music and story can be that powerful. I wasn’t going to say this, but I will, and I’m not even going to say it in quotes: The right music can free the heart to love. What I also learnt at [This American Life] is that everything you do as a producer is in the interest of getting meaning across. The beauty of this is that you decide, at every turn, what that most important meaning in a story is. The hard thing is that every little thing that’s there that isn’t about that meaning is just a distraction. So each breath, choice in level, word, silence and bit of music works towards the refinement of your chosen meaning. Producing a radio story is like writing a book, reading it to an audience and conducting a symphony all at the same time. By this I mean to say that radio is for control freaks.

This article was last updated on 05/06/2020. v1 is 5,071 words and took 12.75 hours to write and edit.

Simulated Annealing

A newsletter on software, creativity, and good books. Get new posts sent to your inbox.

powered by TinyLetter