A PRACTICAL AUDIO GUIDE
Audio Editing Prompts You Can Copy and Adapt
A useful audio prompt names the change and the part of the recording it should affect. You do not need technical language. Start with one change and listen before the next.
Upload audio and start editingFor clearer speech
Try “make the voice clearer and slightly more even in volume; keep it natural.” This can translate into gentle EQ, compression and peak normalization. It does not replace badly recorded words. If the audio has several speakers, a whole-recording adjustment affects all of them.
For cuts and pauses
Use “remove 0:12 to 0:18” for an exact cut, or “keep only 0:30 to 1:15” for an excerpt. For silence, ask “remove long silent pauses but keep breathing room.” The assistant uses detected quiet ranges for pause removal; it does not invent a transcript or identify every filler word.
For a music ending
Try “fade out smoothly over the last 4 seconds; keep the rest unchanged.” For a loop or short preview, specify the section to keep before adding the fade. If you request several changes at once, listen closely to their order: cutting first changes the duration that the fade applies to.
For a correction
Choose the earlier version and describe the new direction. “That was too bright; make a gentler clarity adjustment” is more useful than repeatedly stacking EQ on an unwanted result. Versions let you branch from a sound you already like.
Say what the editor cannot infer
“Make it better” leaves too much room for interpretation. Explain whether this is a spoken message, a podcast excerpt, a music clip or a background recording. Add timestamps for precision. Enable the short listening preview only when you want the AI provider to hear that section.
A prompt to try
Remove the first 2 seconds, reduce steady background hiss gently, and make the voice easier to hear. Keep every spoken word after the cut.
Play the first word, the quietest phrase and the final seconds. Those three checks often reveal unintended changes.
Common questions
Can I say “remove all the ums”?
The editor has no word-level transcript editing. Identify exact time ranges for those cuts.
Can I ask for a different singer or a new song?
Music generation, voice replacement and cloning are not available. The editor changes an existing recording with supported audio operations.