Trim Audio Online

A trim keeps one continuous range and discards everything before the start time and after the end time. It does not remove a section from the middle: the audio between your start and end is copied in order, with nothing cut out of it. If you need to delete a middle section, trim the two surviving parts and place them back to back in a separate join step.

Times accept several shapes. Type plain seconds like 12.5, clock style like 1:05, or full hours like 1:02:03. The two sliders cover the same values with the arrow keys, so a keyboard-only user can move the start by a tenth of a second and hear the change. The label under the range updates as you edit, which prevents the common mistake of trimming the wrong minute.

The selected range is checked before anything runs: the start must come before the end, both must sit inside the file, and the clip must be between a tenth of a second and thirty minutes. A range that fails any of those checks explains itself instead of producing a broken file.

Phones export M4A, recorders write WAV, and most shared audio arrives as MP3. All three open in the same page. The source is mounted into a verified local audio engine, decoded, and re-encoded to MP3, so a 48 kHz WAV and a 22.05 kHz M4A both come out at a consistent 44.1 kHz stereo.

Trimming is a media edit, not a transcription. This page never calls a speech model, never opens a microphone, and never deducts transcription minutes. The verified engine is downloaded after you start a trim; that runtime download is separate from your media and your file is not part of it.

Accepted input is one MP3, WAV, or M4A file up to 100 MB and three hours long. The selected range can be up to thirty minutes, and the exported MP3 is capped at 64 MB for browser memory safety. The browser checks these limits when you choose a file and the execution step rechecks size, decodability, duration, and the encoded result before it returns anything.

Trimming keeps audio and removes time. The output is still sound: a shorter clip you can send, loop, or archive. It changes nothing about what was said, because it does not read the words. A trim is done the moment the encoder finishes, with no language model and no minutes billed.

Transcription keeps the words and removes the listening. The output is text with timestamps, and it usually costs account credits because a speech model runs on a server. It is the right tool for notes, captions, or search, and the wrong tool when you only need a cleaner audio file.

The two combine well. Trim first so the speech model only processes the part that matters, then download the MP3 and take it to the MP3 to Text Converter. Trimming itself always uses zero transcription credits, so the clip is free to prepare before any billed step.