Why this is fast
Native server ffmpeg is roughly 4–8x the throughput of browser WebAssembly on the same job, with real SIMD and threading. Uploads land at the nearest Backblaze B2 point of presence via a presigned URL, the cut runs server-side, the clipped MP3 comes back through Bunny CDN from the same edge. Even a 60-minute podcast trimmed to a 30-second pull-quote finishes in seconds, including upload.
What this tool does, exactly
We run ffmpeg -i input -ss <start> -to <end> -vn -c:a libmp3lame -q:a 2 -f mp3 output.mp3. -ss and -to placed after -i is accurate-seek mode (slower but frame-accurate), -vn drops any album-art video stream, and LAME re-encodes at VBR quality 2 (~190 kbps). Re-encoding on every cut keeps the path predictable across MP3, WAV, M4A, OGG, and FLAC sources — a stream-copy cut from a non-MP3 source into an MP3 container would simply fail.
Three things people cut audio for
Ringtones are the classic case: grab the 20-second hook from a track and load it onto a phone. Podcast and interview editing is the second — pull a quotable 30 seconds out of an hour-long episode to post on social without exporting the whole thing. The third is trimming dead air at the front and back of a voice memo — every recording has an awkward count-in and a fumble for the stop button, and lopping those off makes the file tighter to share.
Sister tools
Need MP3 from a different format first? MP3 converter, M4A to MP3 converter, and audio converter all do the format step. Pulling audio out of a video first? Video to audio converter. Shrink the cut MP3 for email or Discord? Compress audio. Cutting video, not audio? Video cutter. Recording the audio in the first place? Voice recorder captures straight to MP3.