Audio Too Big: Subtracting Bitrate and Channels
A two-hour lecture recording sits at 200MB and the platform caps at 50MB. Audio math is much simpler than video — as long as you cut in order of least damage first.
Cut one: lower the bitrate (least loss)
Bitrate is the direct driver of audio size, and "enough" is far lower than most assume: speech-only content (lectures, interviews, meetings) is fully clear at 64kbps, audiobooks survive 48kbps. Music needs a higher floor — 128kbps is the general tier, 160kbps covers most critical listening.
Identify the content type before cutting: speech can drop straight to 64 for a halving of size with no audible difference. The most common waste is speech encoded at 192kbps — the extra bits buy frequency detail that was never there.
Cut two: halve the channels (speech only)
Stereo carries left-right differences, but spoken content has almost none — a lecture, an interview, a meeting. Converting to mono loses nothing, and at the same total bitrate each channel gets double the effective data: same size, better clarity — or same clarity, half the bitrate.
Music (especially arrangements with deliberate left-right imaging) should keep stereo. The test takes seconds: listen on headphones — clearly different left/right means keep stereo; nearly identical means mono is safe.
Cut three: container, then verify
If two cuts are not enough, switch formats: M4A (AAC) sounds better than MP3 at equal bitrate at similar size; lossless formats (WAV/FLAC) belong in editing pipelines, not archives — convert them to lossy for delivery. Three rounds turn a 200MB lecture into under 60MB.
Verify three spots after compressing: the first ten seconds (encoding artifacts show up here first), the loudest passage (check for clipping), and the quietest passage (check whether noise floor got amplified). Any failure — back off one bitrate step and re-encode.