Cut any audio file by start and end timestamp: MP3, WAV, OGG and more, processed locally in your browser, never uploaded.
Your audio is processed locally with FFmpeg running in your browser. Nothing is uploaded; the file never leaves your device. First use downloads the ~31 MB engine once. Files up to 200 MB.
Times accept seconds (90), MM:SS (01:30), or HH:MM:SS (00:01:30.5), decimals allowed. Leave start empty to cut from the beginning. The end time is required and must be later than the start.
Loading engine…
Cutting audio is routine work: trimming silence off the start of a voice memo, keeping the best 30 seconds of an interview, splitting a long recording into sections. This tool cuts precisely between the two timestamps you type; the start defaults to the beginning of the file when left empty. Because the seek happens before decoding (a fast seek straight to the start point), even long files begin processing immediately.
The kept segment is re-encoded rather than copied, and that is deliberate. Compressed audio stores sound in frames, so a stream-copied cut must land on frame boundaries and can begin or end with a partial frame that plays as a click or glitch. Re-encoding lands exactly on your timestamps: sample-accurate boundaries, no artifacts. At MP3 quality 2 or Opus 128 kbps the re-encode is transparent to most listeners, WAV involves no loss at all, and audio encoding is fast even in the browser.
Cutting runs locally with FFmpeg compiled to WebAssembly, so your audio never leaves your device. Related jobs: trimming video files with the Video Trimmer, pulling the track out of a video first with the Audio Extractor, or changing the format of the cut with the Audio Converter.
Precision. Compressed audio is stored in frames, and a stream-copied cut can only land on frame boundaries, typically every 20 to 26 milliseconds. That sounds small, but a cut that starts mid-frame plays as an audible click or pop. Re-encoding decodes the audio and rebuilds it from your exact timestamps, so the boundaries are sample-accurate and artifact-free. At these settings the re-encode itself is transparent.
Plain seconds (90), MM:SS (01:30), and HH:MM:SS (00:01:30), each with optional decimals (1:23.5). The start time is optional and defaults to 0; the end time is required. The end must be later than the start, and both should fall inside the track; otherwise you get a clear error instead of a broken file.
The kept segment passes through one re-encode. MP3 at variable bitrate quality 2 (~190 kbps) and Opus at 128 kbps are transparent for most listeners and most material; WAV output involves no encoding loss at all. If the source is already a compressed file, the extra generation of loss is inaudible in practice.
Not with this tool, by design. Lossless audio cutting requires frame-aligned cut points and careful handling of bit reservoirs, and even then it frequently produces glitched edges. This tool always re-encodes the segment to guarantee exact, clean boundaries. If you need a truly bit-exact cut, a dedicated desktop audio editor can do it manually.
FFmpeg stops decoding at the actual end of the track, so the output contains everything from your start time to the final sample. If the timestamps are so far out that nothing is produced (for example a start time beyond the track length), the tool stops with a clear error instead of downloading an empty file.