Audio Trimmer

Source{{ sourceBadge }} OutputPCM WAV PrivacyLocal only

Source audio:

WAV, MP3, M4A, AAC, OGG, FLAC, or WebM; one file at a time.

{{ sourceFile ? sourceMetaLine : 'The trim stays in local browser memory.' }}

Ignored {{ ignoredFileCount }} extra {{ ignoredFileCount === 1 ? 'file' : 'files' }}.

The demo is a two-second stereo WAV generated locally.

Review source

{{ sourceDuration > 0 ? formatDuration(sourceDuration) : 'Reading duration…' }}

Start time:

Adjust start:

Included boundary; use the slider for rough placement and the field for exact seconds.

End time:

Adjust end:

Exclusive boundary; it must be later than the start and within the source.

Fade in:

Neutral default: 0 seconds. Fade durations share the kept interval.

Fade out:

Neutral default: 0 seconds. The combined fades cannot exceed the selected duration.

Trimmed audio

{{ row.label }}{{ row.display }}

Review output

Verified WAV

The chart renderer is unavailable. The exact interval remains available in the audio artifact and trim ledger.

Signal	Value	Meaning	Copy
{{ row.label }}	{{ row.display }}	{{ row.detail }}

A clean audio edit depends as much on its boundaries as on the material kept between them. Starting a fraction too late can clip the first consonant of a sentence. Ending too early can remove a natural decay, while ending too late can leave a breath, click, or unwanted silence. Reviewing the interval by ear is therefore part of the edit, even when the timestamps look exact.

Digital audio is stored as a sequence of sample frames. Each frame contains one sample per channel at one instant, so a stereo frame contains a left and right value together. A timestamp must eventually land on one of these discrete frames. At 48,000 Hz, adjacent frames are about 0.0000208 seconds apart; decimal timestamps can be precise without mapping to a fraction of a frame.

Start boundary: The first sample frame retained in the new recording.
End boundary: The point immediately after the last retained frame. Treating the end as exclusive prevents one extra frame from entering the output.
Fade: A gradual gain change near an edge. A fade can soften a hard cut, but it also changes the sound inside the kept interval.
PCM WAV: An uncompressed WAVE file containing pulse-code modulation samples. It is widely usable for later editing, though often larger than compressed audio.

Short fades are useful when a cut crosses a non-zero waveform and would otherwise create a click. They are not automatic repairs for poor timing. A fade-in can soften an intended attack, and a fade-out can remove room tone or the tail of a note. The combined fade lengths must fit inside the selected interval so that the two ramps never overlap.

Trimming changes duration and removes material outside the selection; it does not join clips, remove a section from the middle, normalize loudness, change tempo, or preserve a compressed source codec. Keep the original recording until the new WAV has been played from beginning to end and accepted in its next editing or publishing step.

How to Use This Tool:

Choose the audible boundaries first, then add fades only when the cut itself needs smoothing.

Add one Source audio file. Accepted files include WAV, MP3, M4A, AAC, OGG, FLAC, and WebM; the file must be no larger than 16 MiB and no longer than 3 minutes after decoding.
Set Start time and End time in seconds. Use the sliders for rough placement and the number fields for millisecond steps, then preview the selection so speech, music, and room tone begin and end naturally.
Open Advanced only if an edge needs smoothing. Set Fade in or Fade out from 0 to 10 seconds, keeping their combined duration no longer than the selected interval.
Select Trim audio. If the browser cannot decode the container, reports an unsupported channel count or sample rate, or rejects the interval, correct the named source or timing problem and run the trim again.
Play Trimmed audio and check the interval and fades before downloading the WAV. Changing any timing setting makes the prior result stale and requires a new trim.

Interpreting Results:

A ready result has passed three checks: its byte count matches the planned PCM data, its RIFF/WAVE header carries the expected format values, and the generated file decodes to the planned frame count, sample rate, and channel count.

The reported duration comes from retained frames divided by sample rate, so it may differ slightly from the decimal difference between the entered timestamps. That is normal sample-frame rounding. Use the reported start frame, exclusive end frame, and output frame count when an edit must be reproduced exactly.

Verification proves that the WAV matches the trim plan, not that the edit sounds natural. Listen for clipped syllables, cut-off reverberation, clicks, or an overly long fade, then re-trim from the original if any edge needs adjustment.

Technical Details:

A time selection becomes exact only after both timestamps are mapped to decoded sample frames. The start frame is included and the end frame is excluded. This half-open interval keeps the frame count equal to the end index minus the start index and avoids duplicating a boundary when adjacent selections are later assembled.

Formula Core

Let t_s and t_e be the start and end times in seconds, f_s the decoded sample rate in hertz, C the channel count, and N the retained frame count. Timestamp products are rounded to the nearest whole frame.

\begin{matrix} F_{s} & = & round (t_{s} f_{s}) \\ F_{e} & = & round (t_{e} f_{s}) \\ N & = & F_{e} - F_{s} \end{matrix}

Duration and output size then follow from the retained frames. The 44-byte term is the fixed header used by this PCM16 WAVE output, and each channel contributes two bytes per frame.

\begin{matrix} t_{out} & = & \frac{N}{f_{s}} \\ B & = & 44 + 2 C N \end{matrix}

Audio trim formula symbols and boundaries
Quantity	Meaning	Boundary or unit
F_s	First retained frame	Included; integer frame index
F_e	Frame immediately after the selection	Excluded; integer frame index
N	Retained sample frames	At least 1 frame
f_s	Decoded sample rate	8,000 to 96,000 Hz
C	Decoded channel count	1 for mono or 2 for stereo
B	Generated WAV size	Bytes

Transformation Core

The source container is decoded before the selection is written. Compressed frames are not copied into the output, so the result always follows one explicit PCM16 WAVE path.

Audio trimming and verification stages
Stage	Operation	Stopping condition
Inspect	Check the file signature, size, duration, sample rate, and mono or stereo channel count.	Unknown container, failed decoding, empty audio, more than 3 minutes, unsupported rate, or more than two channels
Select	Round the timestamps to an included start frame and excluded end frame.	End not later than start, boundary outside the source, or fewer than one retained frame
Fade	Round each fade duration to frames and apply a linear gain ramp from silence to full level or full level to silence.	Either fade exceeds 10 seconds or both fades together exceed the retained frame count
Encode	Clamp each sample to the valid range, convert it to signed 16-bit PCM, interleave channels, and write a RIFF/WAVE header.	Cancelled processing or a generated byte count that differs from the plan
Verify	Decode the new WAV and compare its frame count, rate, channels, header, and data size with the plan.	Any structural or decoded property differs

Fade frames are rounded separately from selection frames. A one-frame fade is set to silence at that edge. Longer fades include both endpoints, so the first fade-in frame has zero gain and its last fade-in frame has full gain; fade-out follows the reverse path.

Privacy and Compatibility Notes:

The recording is decoded, trimmed, and verified in the browser rather than uploaded to an audio-processing service.

Browser codec support varies. A recognized MP3, M4A, AAC, OGG, FLAC, or WebM container can still fail to decode in a particular browser.
The output is PCM16 WAV regardless of the source format, so file size can increase substantially compared with compressed audio.
Close the tab or clear the source after working with a private recording, and keep the original until the downloaded WAV passes playback in its destination app.

References:

Web Audio API, W3C, 17 June 2021.
Resource Interchange File Format (RIFF), Microsoft Learn.