Published by PitchChanger · Updated · How these guides are made
Hear the difference
Compare the final two seconds of a steady held tone.
Pause your workspace preview before comparing these clips.
Original synthesis plus three real browser exports; no preference or physical-device test.
How to use this workflow
Keep the intended note length
Choose an end after the musical release you want. A fade cannot recover sound beyond the selection.
Set one duration for all candidates
Enter 2000 ms in Fade-out milliseconds. Keep the start, end and fade-in unchanged.
Compare three fresh WAVs
Select each curve, choose Preview result, and download WAV. Listen through the transition into silence, not only the final sample.
The same fade-out duration can follow different gain paths
For the fade portion, let p be progress from 0 at its beginning to 1 at its end, and u = 1 − p. This trimmer uses Linear gain u, Exponential gain u², and S-Curve gain 3u² − 2u³. These are this implementation’s amplitude formulas; another editor can use the same label for a different curve.
At halfway, Linear and S-Curve both reach 0.5 amplitude, about −6.02 dB relative to the unfaded signal. Exponential reaches 0.25 amplitude, about −12.04 dB. The curves still differ before and after that midpoint.
Swipe or scroll sideways to read all columns.
| Fade progress | Linear gain | Exponential gain | S-Curve gain |
|---|---|---|---|
| 25% | 0.7500 | 0.5625 | 0.84375 |
| 50% | 0.5000 | 0.2500 | 0.50000 |
| 75% | 0.2500 | 0.0625 | 0.15625 |
One held note, three actual exports
The original is an authored four-second 220 Hz sine with a short attack and an intentionally abrupt ending. Each comparison keeps the full selection, Fade in = 0 and Fade out = 2000 ms. The fade therefore begins at two seconds.
The three WAVs were exported through the real browser Audio Trimmer. Each retains 192,000 frames at 48 kHz, or 4.000 s. The example isolates the envelope on a steady signal; it is not a musician’s natural release or a listening-preference study.
Choose the ending you need to hear
If the held note should keep its level initially and soften later, audition S-Curve. If it should retreat sooner, audition Exponential. Use Linear as a straightforward reference. These are listening starting points derived from the formulas, not universal production rules.
A longer fade can obscure an expressive release or quiet final syllable. Move the fade start or shorten its duration before reaching for a different curve. Audacity’s fade documentation is useful background, but its curve definitions should not be assumed identical to this tool’s formulas.
Check overlap and the saved endpoint
Fade-in and fade-out are independent. If they overlap, their gains multiply, so a comparison with a long fade-in is no longer a fade-out-only experiment. Keep Fade in at zero for the examples on this page.
After any selection or fade change, generate a fresh preview and download. Reopen the saved WAV and check its last phrase at normal listening volume. WAV is the direct output; make a separate converted copy if you need MP3.
Common questions
Does changing the curve shorten the file?
No. With the same selection it changes sample gain, not the selected frame count.
Is Exponential the same in all editors?
No. Here it is the squared amplitude formula shown above; other tools may use different definitions.