A podcast is a series of digital audio episodes distributed over the internet through an RSS feed, which lets listeners subscribe once and receive each new episode automatically in a podcast app. What separates a podcast from a single audio file or a live stream is the combination of episodic publishing, on-demand listening and subscription by feed, an open distribution model that no single platform owns.
In audio production, a podcast is also a delivery format with its own technical conventions: speech-optimized recording, loudness around -16 LUFS, compressed files such as MP3 or AAC, and metadata that directories read to list and display each show. The quality of a podcast therefore depends on editorial content, speech intelligibility and correct feed technology at the same time.
How podcast distribution works
Podcast distribution works by publishing a feed, not by uploading to every app. The producer places audio files and an XML feed on a hosting service; directories and apps read that feed regularly and download or stream each new episode to subscribers.
Each episode is an item in the RSS 2.0 feed whose enclosure element points to the media file, with its size in bytes and its MIME type such as audio/mpeg. A stable GUID identifies the episode so that apps do not download it twice. Namespaces such as the iTunes tags and the open Podcasting 2.0 namespace add artwork, categories, explicit flags, chapters and transcripts.
The table below shows the components of the distribution chain and what each one is responsible for.
| Component | Role | Typical failure |
|---|---|---|
| Audio file | Carries the episode content | Wrong loudness or corrupted encoding |
| Hosting service | Stores files and generates the feed | Outage makes episodes unavailable |
| RSS feed | Lists episodes and show metadata | Invalid XML stops directory updates |
| Directory | Indexes shows for search and discovery | Rejects artwork or category errors |
| Podcast app | Subscribes, downloads and plays episodes | Duplicates episodes when the GUID changes |
| Analytics layer | Counts downloads and estimates listeners | Inflated counts without request filtering |
Podcast formats and production styles
Podcasts range from one-person commentary to scripted documentary series, and the format determines recording setup, editing time and audience expectations. Episode length varies widely, from short daily briefings of about 5 to 15 minutes to interview shows that run well over an hour.
The following table compares the most common formats by production effort and technical demands.
| Format | Typical structure | Production effort | Main technical challenge |
|---|---|---|---|
| Solo commentary | One host, loosely scripted | Low | Keeping energy and pacing alone |
| Co-hosted conversation | Two or more regular hosts | Low to medium | Microphone bleed between hosts |
| Interview show | Host with rotating guests | Medium | Matching remote guest audio quality |
| Narrative documentary | Scripted narration with tape and music | High | Editing, scoring and rights clearance |
| Audio fiction | Actors, sound design and score | Very high | Sound design and spatial consistency |
| Video podcast | Recorded conversation with cameras | Medium to high | Synchronizing audio and multiple video angles |
Although many shows now also publish video, the audio version remains the reference product for most podcasts, because a large share of listening happens while commuting, exercising or working, where the screen is not watched.
Audio specifications for podcast delivery
There is no single binding podcast standard, but widely accepted delivery practices exist. The aim is consistent, intelligible speech at a file size that downloads quickly on mobile networks.
The values below are common practice for spoken-word shows; music-heavy or narrative productions often choose higher bit rates.
| Parameter | Common value | Effect |
|---|---|---|
| Integrated loudness | About -16 LUFS stereo, -19 LUFS mono | Similar level to other shows in the app |
| True peak ceiling | -1 dBTP | Avoids clipping after lossy encoding |
| File format | MP3 or AAC | Universal playback on apps and devices |
| Bit rate for speech | 64 to 96 kbps mono, 128 kbps stereo | Balances clarity against download size |
| Sample rate | 44.1 or 48 kHz | Matches recording and editing workflows |
| Metadata | ID3 tags, artwork about 3000 by 3000 pixels | Correct titles and images in directories |
Speech processing usually follows a fixed chain: high-pass filtering around 80 Hz, noise reduction where needed, gentle compression, de-essing and a final loudness and true-peak stage. Over-processing is a common fault, because aggressive noise reduction adds audible artifacts that reduce intelligibility more than the original noise did.
Podcast compared with related audio media
A podcast differs from radio and streaming audio mainly in how it reaches the listener and when. Radio is scheduled and linear, an audiobook is a finished single work, and a platform-exclusive show may look like a podcast but is not distributed through an open feed.
- Broadcast radio: live and scheduled, regulated by broadcast rules, and listened to at the moment of transmission.
- Radio catch-up: a broadcast program offered afterward as a feed becomes, technically, a podcast.
- Audiobook: a complete single work, usually sold or licensed per title rather than subscribed to by episode.
- Platform-exclusive audio show: episodic and on demand, but locked to one app, so it lacks the portability of an open RSS feed.
- Live audio room: interactive and real time, with no persistent episode unless it is recorded and published later.
Measurement, monetization and common mistakes
Podcast audience size is estimated mainly from downloads, not listens, because an RSS file request does not report whether the episode was played. Industry measurement guidelines, notably the IAB podcast measurement guidelines, filter bots and duplicate requests so that figures from different hosts are comparable.
Revenue usually comes from host-read advertising, dynamically inserted ads, listener support and premium feeds. Dynamic ad insertion stitches current advertisements into older episodes at download time, which keeps back catalogs commercially useful.
- Changing the feed URL without a redirect loses subscribers, because apps keep polling the old address.
- Inconsistent loudness between hosts forces listeners to adjust volume; each voice should be leveled before the final loudness stage.
- Recording remote guests over a call codec produces compressed, dropout-prone audio; a local recording on each side is far more reliable.
- Treating downloads as listeners overstates reach, since automatic downloads are not always played.
In everyday language
A podcast is an audio show you can play whenever you like. New episodes arrive on your phone by themselves once you subscribe, a little like a magazine delivered to your door.
Behind the scenes, the show publishes a simple list of its episodes on the internet. Podcast apps check that list and fetch anything new. The key idea is that no single company owns the delivery, so the same show can usually be heard in almost any podcast app.
Seen in the real world
A small public library launches a weekly interview podcast about local history. The host records in a quiet meeting room with a dynamic microphone, while each guest records a local backup file on their own device. After editing, the producer levels both voices, applies light compression and normalizes each episode to -16 LUFS with a -1 dBTP ceiling, then exports a 128 kbps stereo MP3.
The files go to a hosting service that generates the RSS feed, and the feed is submitted once to the major directories. Each new episode then appears in subscribers’ apps within hours. After six months, the library sees about 1,500 filtered downloads per episode, and later moves to a new host using a feed redirect, so no subscribers are lost.