Audio for Podcasters: Recording, Mastering, and Delivery Specs

Record every microphone to a separate track at 24-bit, 44.1 kHz or 48 kHz, mono. 24-bit gives you 144 dB of dynamic range — enough headroom that a loud laugh or excited interjection will not clip even if your gain staging is conservative. Use 44.1 kHz for audio-only podcasts (matches consumer playback systems) or 48 kHz if you also plan to publish a video version (matches video industry standard). Avoid recording directly to MP3 or AAC — lossy encoding during the recording stage means every subsequent edit, EQ, compressor, and noise reduction operates on already-degraded audio. Most multitrack recorders (Zoom F-series, RodeCaster Pro II, Squadcast, Riverside.fm) default to WAV; verify before each session. Save uncompressed multitrack archives in [FLAC](/convert/wav-to-flac) for storage efficiency.

Inside the DAW, audio is processed as 32-bit floating-point PCM regardless of the source file format. Keep imports as WAV. Apply noise reduction first, then EQ (high-pass at 80-100 Hz to remove rumble, gentle de-essing), then compression (3:1 to 4:1 ratio, attack 5-10 ms, release 50-100 ms, threshold targeting 6-10 dB of gain reduction on peaks), then a brick-wall limiter at the final stage. Bounce the mix to a 24-bit WAV master before encoding to delivery format. Keep this WAV master forever — if a network or platform asks you to redeliver the episode in a different format six months later, encoding from the WAV master is one lossy step instead of two. Podcast post-production guidance: see [audio-for-voice-actors](/guide/audio-for-voice-actors) for related noise floor and gating workflow.

The de-facto podcast loudness standard is -16 LUFS integrated for stereo, -19 LUFS for mono, with a true peak ceiling of -1 dBTP. Apple Podcasts normalises to -16 LUFS automatically; Spotify targets -14 LUFS for podcasts; Google/YouTube Music -14 LUFS. Publishing at -16 LUFS sits comfortably between platforms — quieter shows get audibly turned up by the platform's normaliser (and noise comes with them); louder shows get turned down (wasting your loudness work). Measure with Youlean Loudness Meter (free), iZotope Insight, or Auphonic. The -1 dBTP true-peak ceiling matters because lossy codecs (MP3, AAC) can introduce inter-sample peaks above the digital ceiling during decode; -1 dBTP gives the codec 1 dB of safety. Reference: [what-is-audio-limiter](/guide/what-is-audio-limiter).

Voice-only episodes should be delivered mono. A pure mono podcast at 96 kbps sounds as good as a stereo podcast at 192 kbps because a single voice has no stereo information to preserve. Mono halves the bitrate (and the bandwidth bill for the host). Episodes with music beds, stereo SFX, or remote co-hosts panned for spatial separation should be delivered stereo. The cleanest workflow: edit a stereo timeline, then export mono if no stereo material made it into the final mix. Some podcast hosts (Megaphone, Art19) request stereo regardless; check the spec sheet. Bluetooth earbuds and most car stereos fold stereo to mono on the listener side, so stereo is rarely heard as stereo on mobile.

RSS podcast feeds reference one delivery file per episode in the enclosure tag. The format must be MP3 (audio/mpeg) for maximum app compatibility — Apple Podcasts and Overcast accept M4A (audio/x-m4a) but Stitcher, older Sonos firmware, and many in-car apps do not. Use CBR (constant bitrate), not VBR — older podcast players seek by frame and miscalculate remaining time on VBR files. Bitrates: voice mono 64-96 kbps CBR, voice stereo 96-128 kbps, music+voice stereo 128-192 kbps. Encode via [WAV to MP3](/convert/wav-to-mp3) selecting CBR explicitly. Sample rate 44.1 kHz. ID3v2.3 tags for title, episode number, season, artwork (1400x1400 to 3000x3000 JPEG, RGB).

Most podcast hosts re-encode uploads to a standardised target. Buzzsprout 'Magic Mastering' transcodes to 96 kbps stereo or mono at -16 LUFS. Libsyn passes the file through largely unmodified but enforces 320 kbps as a maximum. Spreaker re-encodes everything to 128 kbps. This means delivering at 320 kbps is wasted bandwidth — the host will degrade it to 96-128 kbps anyway. Conservative best practice: deliver 96 kbps mono CBR for voice shows, 128 kbps stereo CBR for shows with music, and let the host pass the file through. If the host re-encodes and the result sounds worse than expected, raise a support ticket — many platforms can disable re-encoding for premium accounts. When an episode bumps against your host's per-file size cap, the [audio compressor](/audio-compressor) drops bitrate or downmixes to mono in the browser without re-rendering from your DAW.

Music beds are usually delivered as MP3 320 kbps or WAV. Always import them at the highest quality available; downsampling at the delivery stage is one operation, repeated downsampling through edit cycles compounds. If you receive a music bed as M4A or AAC, convert to WAV via [M4A to WAV](/convert/m4a-to-wav) before importing into the DAW so the editing chain operates on PCM. For royalty-free libraries (Epidemic, Artlist, Musicbed), download the WAV stems where offered. If a stock track is longer than your intro slot, [trim the music bed](/audio-trimmer) down to the exact length before importing — the DAW timeline stays cleaner when source clips already match their final duration. Final tip: keep a 'show package' folder per episode containing the multitrack WAVs, the mix WAV master, the delivered MP3, and the show notes — when listeners report audio glitches three months later, you will have the source intact. See [audio-for-voice-actors](/guide/audio-for-voice-actors) for the related VO workflow.