podcast-post-production-workflows / 2026-09-28

Prepare AI Music for Podcast Episodes Without Loudness Surprises

A practical workflow for turning an AI music draft into a podcast-ready handoff with speech room, ducking, format notes, and loudness checks.

The problem often appears after the episode feels finished. The intro sounded polished alone, but the host is small after upload. The app turns the whole episode down. A warm bass hides the discount code. That is not only a taste problem; it is a handoff problem. Podcast music has to survive the path from prompt to mix to loudness target to host upload.

What is EasyMusic.AI?

EasyMusic.AI is an AI music creation platform for turning prompts, style direction, and lyrics into music drafts you can review. In a podcast workflow, use EasyMusic.AI to test intros, beds, transitions, and outros around the human voice, while rights review, platform checks, and publishing approval stay separate. When you need a review or delivery copy, the Audio Converter can help prepare a practical format without fixing a crowded mix.

Start With the Voice Lane

Mark the cold open, guest name, sponsor line, chapter break, and ending. Write where the words must be effortless and where music can step forward for two seconds. Do not generate a full song and then fight it into the episode. Ask for short cues with one job each: six-second intro, low bed under recap, two-second transition, clean outro tail.

Prompt for Mix Headroom

A useful prompt might be: podcast intro bed, 92 BPM, warm keys, soft low percussion, no lead vocal, no busy hi-hats, sparse midrange, clean ending before host speaks. Generate two alternatives: one calmer version and one no-drums version. If the bed will sit under long speech, the winner is not the catchiest hook. It is the draft that leaves consonants, names, numbers, and calls to action clear.

Check Speech Against Background

Use your ears and a meter together. As a practical reference, background sound under speech often needs a wide gap; WCAG Technique G56 describes 20 dB of foreground-over-background difference as a way to keep speech understandable. Do not make the music loud by itself and then force the voice over it. Place the voice first, lower the bed, then use ducking or keyframes before and after key lines.

Make the Host Copy Deliberately

Apple Podcasts recommends overall loudness around -16 dB LKFS, plus or minus 1 dB, with true peak not exceeding -1 dB FS. Spotify for Creators supports MP3, M4A, and WAV for audio episodes. Those facts are not magic settings for every show, but they keep export decisions from being random. Save a clean source master, make an upload copy for the destination, and do not treat AI-generated music as automatic clearance for every use.

Ideas You Can Reuse Today

• Write a voice-first cue map before generating music.
• Ask for no lead vocal and sparse midrange for any bed under speech.
• Start testing the bed roughly 15 to 20 dB below the voice, then adjust by ear.
• Duck music around guest names, offers, URLs, prices, and sensitive claims.
• Keep a handoff card with prompt, draft name, mix target, format, and destination platform.
• Listen on a phone speaker before uploading the episode.

FAQ

Should I normalize the music before placing it under the voice?

No. Mix voice and music first, then check the loudness of the full program.

Is one MP3 enough for every podcast host?

Not always. Keep a source master and create the upload copy after checking the destination requirements.

What if the music is beautiful but covers the host?

Go back to a simpler prompt: lighter percussion, less midrange, no vocal, and a cleaner ending. Lower volume alone may not solve a dense arrangement.

Can I use AI music in a commercial podcast?

It can be part of the workflow, but rights, tool terms, disclosure, and platform rules still need separate review.