Account

What new tool do you want?

Video & Audio

Audio Cutter & Trimmer (MP3, WAV, M4A)

Cut and trim audio online — drag to select the part you want, add fade in and fade out, and save as MP3 or WAV. Great for ringtones and clips.

How to use

  1. 1Add an audio file (or a video to cut its sound).
  2. 2Drag the handles on the waveform to select, add fades, preview, then download MP3 or WAV.

Frequently asked questions

How do I make an iPhone ringtone?

Cut a clip under 30 seconds, save as MP3, then convert to M4R (or import via GarageBand) — Android can use the MP3 directly.

Is my audio uploaded?

No — cutting and encoding happen in your browser.

Which formats can I open?

Anything your browser can play: MP3, WAV, M4A/AAC, OGG, FLAC (most browsers) and the audio track of MP4 videos.

Dragging the waveform beats typing timestamps

The selection can be set two ways — dragging the handles directly on the rendered waveform, or typing exact start/end times into the Start and End fields (format m:ss.ss, parsed and clamped to the file's actual length). Dragging is faster for finding a natural cut point by eye, like the silence right before a chorus starts; typing exact numbers is better once you already know precisely where a clip needs to begin, e.g. matching a lyric or sound effect to another edit.

Fades exist to hide a cut, not decorate it

Fade in and fade out (0–10 seconds each) are applied after trimming, so a selection that starts mid-word or mid-note can be softened rather than clipped abruptly. For a ringtone or notification sound this matters less — an instant start is often desirable so there's no delay — but for a clip pulled from the middle of a song, even a 0.5–1 second fade on each end avoids the audible 'pop' of a hard cut landing on a waveform peak.

Worked example: making a 25-second ringtone

Loading a 3:40 song and dragging the selection to roughly 1:12 to 1:37 (the chorus hook) gives a 25-second clip — under iPhone's typical ringtone length expectations. Adding a 1-second fade in and fade out smooths both edges, then 'Play selection' previews it instantly via the Web Audio API before committing to an export, so you can nudge the start by a few hundred milliseconds if the beat doesn't land right, without re-decoding the whole file each time.

No upload, and no re-encoding until you export

The file is decoded once into raw audio samples in the browser and every drag, fade adjustment and preview play works directly on that decoded buffer — nothing round-trips to a server, and nothing is re-encoded until the final MP3/WAV export. That's different from trimming in a tool that re-encodes on every preview: here, previewing the selection a dozen times while fine-tuning the start point costs nothing in quality or wait time, since the decode happens only once per file load.

Long fades on a short clip can cancel each other out

Fade in and fade out each use a quarter-sine curve measured from their own end of the clip, and the two are applied independently — so on a very short selection, setting both anywhere near the clip's full length makes them overlap in the middle rather than one simply following the other. A 2-second selection with 10-second fade in and fade out requested on each end (both get clamped to the clip's own length) never reaches full volume at all: the two curves multiply together and the loudest point, dead center, peaks at only about 49% of the original level. For a clip this short, fades well under half its length — a quarter-second or less — keep the middle at full volume the way a fade is usually meant to sound.

Trimming is sample-accurate, not keyframe-limited

Because trim() slices the decoded Float32Array directly at floor(start × sampleRate) and ceil(end × sampleRate), the cut point lands within one sample of the time you specify — at 44.1 kHz that's under 0.023 milliseconds of error, far tighter than ear or eye can detect. This is a real advantage over cutting the audio track of a video file by re-encoding, which is often constrained to the nearest keyframe or frame boundary (commonly 33–40 ms at 25–30 fps); for isolating a single word or a drum hit precisely, working from the audio waveform directly avoids that frame-size rounding entirely.

Related tools

More: Video & Audio