Audio Trimmer & Converter

about this tool

Drag across the waveform

Drop an audio file — or a video, to take the sound out of it — and the browser decodes it into samples. The waveform is drawn from those samples: each column shows the loudest value up and down within the slice of time it covers, which is how an editor draws one. Drag across it to choose a part, or type the times as 0:05, 1:05.5 or 65.5.

The selection's loudest sample is shown in decibels below full scale, so you can see whether there is room to turn it up before you do.

What it does to the sound

In order: the selection is cut, mixed down to mono if you asked, then normalised, then faded. Normalising scales everything so the loudest sample sits at -1 dB, with every channel scaled equally so the balance is kept; silence is left alone rather than amplified into noise. Fades are linear, and if the two together are longer than the clip they are shortened in proportion so they cannot cancel the middle out.

Formats

WAV is written here, by this page: a RIFF header, a format chunk and interleaved samples, at 16, 24 or 32-bit float. That means it always works, whatever encoders the browser has, and the size is exact and predictable — sample rate times channels times bytes per sample, plus 44 bytes.

M4A (AAC) and OGG (Opus) go through mediabunny to the browser's own encoders, and are only offered when the browser reports it can encode them. There is no MP3: no browser has an MP3 encoder, and the usual alternative is a GPL-licensed ffmpeg build of tens of megabytes, which this site does not use.

Limits

The sample rate is whatever the decoder gave, and is not converted; the lossy encoders will resample to 48 kHz themselves, and the panel says so when your file is higher. There is no compression, equalisation or noise reduction, and no multi-track editing: one file, one selection, one output. The whole file is decoded into memory, so an hour-long recording at 48 kHz stereo takes about 700 MB of it. Nothing is stored between visits.

For video rather than sound, video to GIF, trim & convert does the equivalent job. To capture the audio in the first place, the screen, camera & audio recorder.

questions

Which formats can it read and write?

It reads anything your browser can decode, which in practice means MP3, WAV, M4A, AAC, OGG, Opus and FLAC, plus the audio inside an MP4, MOV or WebM. It always writes WAV, because that is written here rather than by a codec, and offers M4A and OGG when the browser has the encoders for them.

Why is there no MP3 output?

No browser ships an MP3 encoder, and the usual way round that is a GPL-licensed build of ffmpeg weighing tens of megabytes, which this site does not use. M4A and OGG are smaller than MP3 at the same quality and play on everything current.

What does normalising do?

It finds the loudest sample in your selection and scales everything so that sample sits just below full scale, at -1 dB. Every channel is scaled by the same amount, so the balance between them is unchanged. It does not even out loud and quiet passages: that is compression, which this tool does not do.

Does 16, 24 or 32-bit matter?

16-bit is what a CD uses and is plenty for listening. 24-bit gives more headroom for further editing and is half again the size. 32-bit float keeps exactly what came out of the decoder, including anything above full scale, and is the largest.

Is the file uploaded?

No. The browser decodes it into samples, this page does the trimming and shaping, and the file is written here or by the browser’s own encoder. Nothing leaves the page and nothing is stored between visits.