What uneven volume usually looks like
A quiet guest and a loud host. The most common case. The host has a good microphone and a set level; the guest is on a call, a laptop, or a phone, and comes through several decibels lower. Every handoff is a jump.
Inconsistent microphone distance. The same person is loud when they lean in and quiet when they sit back or turn to a co-host. The level drifts within a single speaker’s parts, not just between speakers.
Sections that change level. Someone adjusts a gain knob halfway through, a second segment was recorded on different gear, or a remote connection changes quality mid-episode. One part of the file sits at a different level from the rest.
Why turning it up is not the fix
Raising the volume of the file, or normalizing it, applies one change to everything. The quiet parts come up, but the loud parts come up by the same amount and may clip or need limiting. The difference between them, which is what listeners notice, does not change. The same is true of lowering the whole file to tame the loud parts.
Fixing uneven volume means applying different gains to different parts of the recording: more to the quiet turns, less or none to the loud ones, with smooth changes between them.
How Pod Level levels a recording
Pod Level detects speech turns, measures the loudness of each turn, and moves every turn toward the recording’s typical speech level, applying up to 12 dB of correction per turn in either direction. Turns that are too short or too uncertain to measure follow the gain of the speech around them, and gain changes ramp briefly at turn boundaries rather than stepping. The gain within a turn is steady, so the natural dynamics of each voice are kept.
Because the correction follows the turns rather than a fixed clock, it handles all three cases above: a consistently quiet guest, a speaker who drifts, and a section that sits at a different level. After the turns are balanced, the whole programme is normalized to −16 LUFS with true peaks limited to −1 dBTP.
Processing a recording
- Open Pod Level in current iPhone Safari, Chrome on iPhone, or desktop Chrome or Edge.
- Choose the file. WAV, MP3, M4A/AAC, OGG or Opus, WebM, and FLAC are accepted, and the original is opened read-only.
- Check the duration and the proposed output format, then choose a save location on desktop or start directly on iPhone.
- Let analysis and rendering run. You can cancel at any time; nothing is written to the original file.
- Preview a few handoffs before and after, and read the measured input and output loudness, the applied gain range, and the number of turns adjusted.
Limits worth knowing
Correction is capped at 12 dB per turn, so a very large gap is reduced rather than closed completely. Pod Level does not reduce noise or room sound, which becomes more audible when a quiet part is raised. It works on one mixed file; if you have separate tracks per speaker, balance them in your editor first or export a mix and level that. For the case of two people at steadily different levels, see balancing podcast speakers.