Spend the available bitrate on the voice and programme that matter
Source-aware LAME MP3 settings for interviews, podcasts, audiobooks and narration, including mono, ABR and practical low-pass choices.
MP3 Settings for Interviews, Podcasts and Storytelling
Start with the purpose of the recording
Speech does not automatically need a narrow-band encode. A modern podcast recorded with a good microphone can contain ambience and vocal overtones well beyond the core intelligibility range. Interviews captured in noisy environments, long-form narration and archival speech may benefit more from mono encoding and a deliberately limited upper bandwidth.
Suggested starting points
| Scenario | Starting point | Why |
|---|---|---|
| Stereo podcast / storytelling | -V4 or --abr 160, stereo, low-pass Auto | Keeps stereo ambience, music and sound design while remaining efficient for spoken-word programme audio. |
| Natural podcast | -V4 or --abr 128, low-pass Auto | Preserves a natural top end while remaining efficient. |
| Mono interview | -m m --abr 96 | Avoids spending bits coding two channels when the useful source is mono. |
| Long-form narration | -m m --abr 80 --lowpass 12 | Prioritises speech and duration over full-band ambience. |
| Very bandwidth-conscious speech | -m m --abr 64 --lowpass 8 | Deliberately voice-centred; expect a less open sound. |
Do low-frequency cleanup before LAME
A conventional podcast rumble filter is often placed around the tens of hertz to roughly 80–100 Hz depending on the voice and recording. LAME's built-in high-pass cannot operate that low at normal 44.1/48 kHz sample rates, so do that cleanup in the PCM editor or restoration chain before encoding.
Command examples
lame.exe -V4 stereo-story.wav stereo-story.mp3A good first example when the programme genuinely uses stereo ambience, music or sound design.
lame.exe -m m --abr 96 interview.wav interview.mp3lame.exe -m m --abr 80 --lowpass 12 narration.wav narration.mp3Use the narrower examples only after listening. A storytelling production with music, stereo ambience or sound design should generally be treated more like programme audio than a bare voice recording.