Prep

I don’t prep. Cold reading gets easier the more you do it, and I routinely record in just 1 fandom, so don’t need to research character voices. I will search on Youglish for pronunciation of words I’m unsure about as I go along (‘glowering’, I am looking at you). Typically, I will only record from authors who have a blanket permission statement, often found via this browser extension by Brickgrass, powered by Rindle's fpslist.org.

Record

Don't drink right before recording, since drinking doesn’t immediately hydrate the vocal cords (the windpipe is a different pipe from your gullet) and drinking can make mouth noises worse. I pause if I can hear a motorbike in the distance, an ambulance whining, or a neighbour driving in my street/slamming car doors. My mic is very good and the general rule of thumb is “If you can hear it, so can the mic”.

This is a great 10 minute warmup with the National Theatre.

I definitely recommend watching this video on breathing.

Something that has improved my skill in narration itself, is watching coaching videos, such as this one with Barbara Rosenblat about not overdoing it and letting the words speak for themselves, this one about taking your time, Nancy Wolfson on relaxing your voice (I think of her lying on the floor whenever I start!!!), and this one as well as this one on ‘active hush’ which I was doing already.

I also benefitted from this lesson on performance, and if you’ve got the appetite for more, this one, this one, and this one also helped me a lot.

I podficced for 2 years with a second-hand Snowball ICE, and now I record with a RØDE NT1‑A (cardioid condenser) with a Scarlett 2i2 audio interface.

In brief, mics like Snowball ICE, Yeti, Bee, and ‘podcaster’ mics are USB-powered. XLR microphones are more expensive, are higher quality, and require their own power via an audio interface.

I’ve pasted together a few phrases from both microphones to show you the difference in audio quality:

Until recently, I recorded in front of a bookcase, in a small room full of stuff, which resulted in no room echo. This was GOOD and you do not need a booth, providing you stop for outside noises like seagulls etc. I am now in a DIY booth (walls of rockwool for sound proofing, acoustic foam for sound dampening). It's important to know the difference: sound proofing - stops outside noises coming in, sound dampening - stops you sounding echoey/sound bouncing off flat surfaces. The screenshots of example noises in this post are from before my booth.

Audacity takes up the lower half of my laptop screen, and my script is in the window behind it. This improves my posture, by preventing me staring down at my phone. Typically, my script is in Word, because sometimes I stumble 6 times over each typo, so I need an amended script. I prefer to read in Comic Sans, but any dyslexia-friendly font can help reduce stumbling.

For time-saving reasons, I record using punch and roll.

I record a test sentence and listen back, mainly to check that yes I am using my mic and not the inbuilt laptop mic.

I often record live on Discord. I can record for about 60 mins before I lose focus, get tired/uncomfortable, and this results in 30-40 mins of almost-ready audio.

To reduce lip noise, I recommend lip balm, and not shutting your mouth at the end of sentences.

Edit

GarageBand vs Audacity. (vs Reaper vs Audition vs Studio 1 vs Ocenaudio)

I used GarageBand for a year before switching to Audacity. Simply put, GB’s editing tools are minimal and Audacity is free. I have never looked back to GarageBand since changing to Audacity. (EDIT Feb2024, I am now recording in Reaper and editing in Audacity due to rare but infuriating faults with Audacity. Reaper is $60 one-off fee. Please email me if you are stuck with Reaper)

Audacity vs the rest

Though they cost money, the other DAWs are ‘non-destructive’ editors, so after saving a file you can rewind your editing to get the raw back. To compensate for this, always export a raw, unedited WAV. Learn from my mistakes, I'm begging you. If you're new ish to editing, I highly recommend you duplicate the track, mute and collapse it, as shown below. This gives you an immediately accessible backup.

I don’t do noise reduction because my mic is very quiet, but for most people, this will be the first step. In plain English, -60dB of background noise, means that it's 1000 times quieter than your voice. When I used the Dog Clicker method, I used irrationalpie’s advice on highlighting the peaks.

First I master the audio. I use a macro with the following 4 steps:

1. High-pass filter – 80Hz, 12db roll-off for speech. This removes the inaudible low frequency rumbles, very low pitch sounds, such as microphone errors, thunder, rumble as much as possibe, and so allows you to delete sections of your audio without creating an audible editing ‘click’.
2. Then Filter curve EQ, with the preset “low rolloff for speech” this reduces the 80-100Hz frequencies. This helps sort out popping, and reduces extreme proximity effect. It’s another high-pass filter and might not even be necessary.

 

3. Then I do “RMS Normalize” which is a Nyquist plugin (I.e., needs to be downloaded, put in the Audacity plugins folder on your computer, then added to your effects menu, these instructions might help), with the setting of -20dB.

- Normalise = adds the same amount of volume to the entire track
- Root Mean Square Normalise = you analyse the entire waveform, all peak values are averaged to create a reference, this then acts as the anchor for the normalising value

Since this puts my podfic in ranges acceptable to audiobook distributors, I know it’ll make loudness comfortable for listeners. If you do the “Dog Clicker” method, you’ll want to avoid using this macro until you’ve replaced the clicks with labels, as you don’t want these peaks messing up the RMS calculation, since it uses averages from the entire file.

4. Finally, the last step in the macro is “Limiter”. The settings are “soft limit” 0, 0, -3.5, 10.

Thus, my macro looks like this:

I then use the iZotope Mouth Declicker plugin. The free Audacity declicker add-on is quite good but takes ages. Feel free to send me your WAV files and I will use the iZotope declicker for you, as it's £1000 otherwise. I really don't mind.

Below is a session of unedited raw punch and roll audio. Note how small in height the blue waveform is. This is too quiet. Nothing worse than trying to listen to a quiet podfic on your walk home and pausing/rewinding every time a car drives past.

After running the aforementioned mastering macro (1 click which takes about 30 seconds) you can see the volume difference in the height of the blue waveform:

To fit all your audio into your screen, press Cmd+A and then Cmd+E.

Here is the same step, but more zoomed in:

Note how before, the very bottom is a thick pink-purple line (low frequency) and after, the bottom is pale purple. Thanks to the high-pass filter, I can now delete sections with no audible editing click sounds.

To view the spectrogram, click this tiny triangle and select “multi view”:

Why the “Click Removal” plugin might not work for you

This is designed for decrackling vinyl. Not clicks caused by spit, the contraction of your soft palate, the sound of your lips touching, etc.

Here is a before and after on the Mouth Declick:

The plugin will never catch every click. If it does, it might start deleting the Ts and Ps. Don’t expect ‘perfect’ settings, just a plugin that greatly reduces the clicks.

Adjusting clip boundaries

This line at 25:52, between “Audio Track 39” and “Audio Track 40” is what’s known as a ‘clip boundary’. This is where I punched in when doing punch and roll. I haven’t started speaking at exactly the right second, so I am selecting the gap before I spoke, dragging up to the clip boundary, and pressing backspace to delete. This’ll leave my pacing the way I intended.

I then click on the clip boundary to merge the 2 clips. This is how that error now looks:

I sit very still at the end:

And I shall select 0.5 seconds of the quietest part, and copy it, to paste over 0.5 seconds of strange noises throughout. This is what I call “visual editing” because I’m not listening. Essentially there should either be quiet room tone:

Breathing, or speech:

Not blips like this, which isn’t a word, a breath or a quiet room. Probably the declicker missed it:

Or strange clucking sounds:

Any remaining dark pink lines at the bottom are cars driving, motorbikes droning, or aeroplanes:

To delete a dodgy noise, drag your mouse to draw a rectangle around the offending sound:

Then select Effect > Spectral Edit (I have set up a keyboard shortcut to “Y”, so I select and press Y for speed)

Alternatively, you can drag sideways a smidge and hit the delete key. This won’t mess up pacing because it’s a tenth of a second long:

This all needs to go, I think this is my husband leaning on the squeaky bannister:

This high-pitched whining is I think my bluetooth doorbell:

This marvellous sound is a bird:

Let's talk about an editing tip that has changed my life. It's called the punchcopypaste macro. This is my macro setting. and I have assigned it the keyboard shortcut 'P' for Punch (via Audacity > Preferences > Keyboard). Put 3 seconds of quiet room tone at the end of your track. Now whenever you want to paste over an extraneous noise, hit P, and it will paste the exact duration of quiet room tone over the noise.

Proof

I listen through to check that nothing stands out as a mistake, i.e., obvious misreads, the wrong character voice. For a commission or charity auction, I have someone else proof the audio against the manuscript because I keep zoning out when I do this, and I don’t know what I am mispronouncing. You don't know what you don’t know.

Create M4B file

For chaptered works, I create an audiobook file. This makes it easy for someone to listen to 20 chapters without downloading individual MP3 files and getting lost. I use AudiobookBinder (free for Mac). It is very easy to use:

Embed metadata and cover art to the file itself

At some point during this process, I’ll have made cover art. Probably whilst listening back to the podfic. I like to embed cover art so that when someone is playing the podfic on their phone, they can see the pretty art and the author details etc. I like to embed metadata so if listeners renamed the file, they still know what it is, who podficced it, who wrote it. (I like to have a long-term view of fandom and consider preservation.)

I use a program called kid3. It’s free for Mac, Linux and Windows. You drag the cover art to the bottom right of the MP3 and it somehow does some magic and embeds this art. I put the title, author, etc. in the “tag 2” section and Cmd+S to save. (As AudiobookBinder also does cover art and metadata, the Kid3 step is not necessary when creating M4B files)

It looks like this:

Host

AO3 only hosts text, not art or audio or any other file. I upload my cover art and MP3s/M4Bs to archive.org, because it is free, and it is a longstanding charity, so I hope it’ll last forever. I highly recommend two hosts, for longevity, and at present (2024) the Audiofic Archive is a good option

It's important when getting the URL for the art that you don't open the image in a new window and copy that URL. That link is a temp link and will break a few months down the line. Here is a comparison of the temporary and permanent links to the same image file, so you know what to look out for:

https://archive.org/download/how-i-podfic/temp%20link.png -permanent
https://ia802507.us.archive.org/34/items/how-i-podfic/Picture%2010.png - temporary, will stop working after a few weeks/months

To get the URLs you need, right click the image file and copy the link like I do in this screenshot:

This is how to get the MP3/etc links:

Post

I use the Podfic Posting Helper browser extension to pre-populate the tags, the related link, the warnings, et al., and I have customised the extension to contain Azdaema’s podfic template.

If I have drafted the post days in advance, I change the date to today, so it shows at the top of the listings for new works.

VERY IMPORTANT: I preview the work before posting. There is a known bug that prevents the “Someone has made a related work, would you like to link it” email being sent to the author.

Notifications about related works don’t go out if the work is posted without preview: If you are creating a new work with "This work is a remix, a translation, or was inspired by another work" ticked, choosing to preview at least once before posting will ensure a notification is sent to the author of the inspiring work.

I then add the podfic to my spreadsheet for stats reasons. I have a list of every podfic, word count, length, author.

I hope you found this helpful. Any questions, please ask.