01 · Measure
Loudness analysis.
We measure integrated loudness and true peak across the whole episode with R128 methods. No guessing from a moving average.
Technical guidelines
You don't have to care about loudness. We do.
Deliver your audio however it comes. If you want to help, aim for around −24 LUFS with a little headroom. Either way, every episode is conformed to spec before it publishes.
Loudness
Aim for −24 LUFS. Or aim for nothing.
Loudness is measured in LUFS, integrated across the whole episode. Every platform normalizes to its own target. Delivering so your show benefits from each of them is our job, not yours.
You don't need to be the loudest thing on the planet. Resist the urge to crank it. Send it quiet, around −24 LUFS, dynamics intact, with at least −2 dBTP of headroom. That is the archival loudness we like to receive. If your tools can hit it, great. If they can't, send whatever you have. We are not fussy about the input.
We handle the loudness. Castcore runs the exact same pipeline we build for broadcasters around the world, not a fixed gain offset bolted onto every file. We measure with R128 methods and bring every episode to the AES77 −18 LUFS target, leaving the right headroom so platforms like Spotify can apply their own boost without clipping.
We take the standards for everything they're worth. Knowing exactly how each platform normalizes, and delivering so your show benefits from all of them, is the part that takes expertise. That is the part we are good at.
Integrated loudness · LUFS
deliver ≈ −24
−32−28−24−20−16
Send it quiet, around −24 LUFS, and let us do the work. We store at −18 LUFS with a −5 dBTP ceiling — headroom calculated so each platform's boost lands clean.
Deliver around−24 LUFS
Input ceiling−2 dBTP
We store at−18 LUFS
Output ceiling−5 dBTP
StandardAES77
What to deliver
A clean master beats a loud one.
Upload the best-quality file you have, in any format. We re-encode every episode ourselves, so the one thing that matters is that what you send is clean. Garbage in, garbage out.
- File format
- Anything FFmpeg can decode — which covers everything you're realistically working with, and quite a few formats you probably aren't. No need to convert anything before you upload.
- Best quality
- A 24-bit / 48 kHz master. Sample rates above 48 kHz are strictly irrelevant for publishing. They add size and nothing else.
- Headroom
- Limit true peaks to −2 dBTP, or don't — we apply conformance limiting regardless. We'd skip it in a perfect world, but a certain Swedish streaming platform has its own ideas about loudness targets, and we need the headroom.
- Channels
- Stereo, interleaved.
- Loudness
- Aim for −24 LUFS if you measure it, or aim for nothing at all. We conform every episode to spec regardless. Just make sure you don't see red anywhere.
- Cover art
- Square RGB, 1400×1400 minimum. We rescale up to 3000×3000 for the directories.
- Duration
- 150 minutes at most per episode, and that's 2.5 hours per episode. Split your episodes into parts if you exceed this limit.
What happens after upload
Every upload, conformed to one target.
02 · Process
Dynamics to AES77.
Solo: loudness normalization to −18 LUFS and true-peak limiting. Producer: full upward and downward compression to target LRA, loudness normalization, and conformance limiting. Either way, the output meets spec.
03 · Encode
Fraunhofer AAC.
We encode above the floor bitrate with the Fraunhofer AAC encoder. Audibly cleaner than minimum-bitrate output at the same size.
04 · Artwork
Normalised to 3000×3000.
Apple Podcasts requires exactly 3000×3000 JPEG. We rescale and convert your cover automatically on every upload — whatever size or format you send.
Before you publish
A 30-second checklist.
- No clipping. Peaks at −2 dBTP or below, or let us limit.
- Highest-quality source. Export from your master, not a previous MP3.
- Stereo. Deliver both channels.
- Square cover, 1400px or larger. RGB, no transparency.
- Title & description written. They feed the index and search.
- Show notes, if you have them. More context means better discovery in the index.
Want the engineering behind all this?
LUFS vs dBFS, dithering, encoding artifacts, the Dynamic Score: the people who build Castcore's audio pipeline write it all up at Fremen Guru. Castcore keeps it practical. Fremen goes deep.
Clean file in, on-spec episode out.
That's the whole deal. Start a show and your first upload comes out to spec.