Module 2: Audio Foundations
Every edit sits on top of the raw recording, and the raw recording is where episodes are won or lost. This module teaches you to diagnose what's wrong with a track in the first thirty seconds, run a cleanup chain in the right order using your ears instead of preset stacks, and deliver files at the loudness standard every platform and client expects.
Diagnosing recordings: what can be saved and what can't
Before you touch a single tool, listen to the raw audio on headphones and name its problems. There are four usual suspects. Room echo — the voice bouncing off bare walls, common in spare bedrooms and offices — makes speech sound distant and cheap. Mic distance is its cousin: a guest talking to a laptop across a desk gives you a thin, roomy voice no plugin fully rescues. Levels are the third: one guest loud, one quiet, or a track recorded so hot it clips and distorts on every laugh. Noise is the fourth: fans, traffic, a fridge hum, a dog two rooms away.
Sort what you hear into two piles. Fixable in post: uneven levels, steady background hum, mild echo, mouth clicks, long silences. Not really fixable: heavy clipping, a voice buried under louder noise, two people talking over each other on a single track, echo so strong the words smear. The professional skill is knowing the difference in the first minute and saying so immediately — to the host, with a fix for next time (mic closer, a treated corner, separate tracks), not after you've spent four hours failing to rescue the file.
That honesty is a producer behavior, not an editor one. The editor who quietly ships bad audio gets blamed for it. The producer who flags a bad take the same day, explains the one change that prevents it, and ships the best possible version gets trusted with the whole show. Diagnosis notes — two or three lines per episode, kept with the project — also become your evidence at rate-raise time: a record of problems you prevented, not just ones you survived.
A guest recorded into a laptop mic across a bare room, talking over the host on one shared track, with heavy echo. The host asks you to "just clean it up." What's the producer's move?
The cleanup chain, in plain terms: ears before presets
Cleanup has an order, and the order matters. Noise reduction goes first, while the track is still raw: it targets the steady floor underneath the voice — fan hum, room tone, hiss. A little goes a long way; push it hard and voices turn watery and robotic. EQ comes next, shaping the voice itself: cut the muddy low frequencies where rooms boom, tame harshness up high, and leave the warm middle alone. Compression follows, evening out the performance — whispers and laughs brought closer together so the listener isn't riding the volume knob. De-essing goes last, softening the sharp "s" sounds that the earlier steps often exaggerated.
Every DAW named in this course — Audacity, Descript, Audition — has these tools and a pile of one-click presets beside them. Presets are starting points, not answers, because no preset has heard your guest's kitchen. The working method: solo the voice, adjust one tool at a time, and bypass it every few seconds to compare. If you can't hear the difference, the change isn't doing anything — undo it. Two or three honest adjustments beat eight stacked effects every time.
One trap to avoid: processing a copy into oblivion. Always work on a duplicate of the raw track and keep the original untouched in the project folder. Guests ask for changes, hosts change their minds, and "the version before the noise reduction" is a sentence you want to be able to answer with a file, not an apology. Raw stays raw. That's a rule with no exceptions in this course.
A beginner loads a voice track and stacks every preset that sounds impressive: maximum noise reduction, three EQs, two compressors. The result sounds underwater. What went wrong?
Loudness and delivery: −16 LUFS, formats, and naming discipline
Podcast audio has a loudness standard, and hitting it is not optional polish — it's the line between amateur and professional delivery. Master speech episodes to −16 LUFS for stereo. That's the target the audio editor's rate ($30–150 an episode) already assumes you're delivering. LUFS measures how loud the episode plays overall, not its loudest peak, so a mastered episode sits comfortably next to every other show in a listener's queue instead of blasting or whispering. Check loudness on the final export, after every edit and music bed is in — measuring an intermediate file tells you nothing.
Format discipline is the rest of delivery. Export the master as a high-quality WAV for the archive and an MP3 for upload, since hosting platforms expect MP3 and listeners' apps are built around it. Use a sensible bitrate (128 kbps is the common speech standard) and never upload a WAV where an MP3 is expected — file size punishes listeners on mobile data and some platforms will reject or re-encode it badly.
Finally, naming. A professional file name carries the show, the episode number, and the title: something like Fieldnotes_E042_TheWinterAudit.mp3. What it never says is "final," "final2," or "REAL_final." Your future self, the host, and anyone who inherits the show should be able to find any episode's master in ten seconds from the name alone. Keep the same convention for project files and note files, and keep the raw recordings in the same folder structure. Delivery is a system, not a save dialog — and systems are what the next modules build on.
You're delivering a finished stereo interview episode to a client for upload. Which delivery is correct?
What you can now do
- Diagnose a raw track in the first thirty seconds — room echo, mic distance, bad levels — and say honestly what can be saved and what can't.
- Run the cleanup chain in the right order, noise reduction before EQ, using your ears instead of stacking presets until voices turn watery.
- Master a speech episode to −16 LUFS and deliver it in the right format with disciplined file names.