Audio Kit: open, edit, clean up and convert any sound

Open a sound or a video takes MP3, WAV, M4A, AAC, FLAC, OGG and AIFF files, and the sound inside MP4, MOV and WebM videos, up to 400 MB and 60 minutes. Press Choose a file, drag files onto the box, or press Record to record from your microphone (Stop and open puts the recording in the editor). Pick several files at once and they are joined into one, in the order picked. Open one from your Graphics Library (under the buttons, and Library in the editor) lists your saved sounds and videos, searchable.

What Audio Kit does shows each main job as a card with a picture of it: Pull the sound out of a video, Remove background noise, Edit by the words, Split the voice from the music, and Save as any format. Each card's Try it on a sample opens a real sample file and runs that exact job, so you see and hear it before using your own file. Samples are free; the interview sample comes with its transcript already made.

Your look (top right) sets the editor's color (aiKit, your Brand Kit color, Studio or Soft), the waveform style (Bars, Fine line or Mirror) and how roomy the screen is. It is saved to your account.

Audio Kit opens in Basic mode, made for going to an exact spot in a song or recording and editing it by hand, with no AI. Basic and Advanced at the top switch modes, and your choice is remembered. The waveform fills the editor, with a time ruler along its top that gets finer as you zoom in. Click anywhere to jump there; drag to select a part; drag the white handles at either end of the selection to adjust it. Zoom with the + and − buttons, a two-finger pinch on a phone or trackpad, or Ctrl/⌘ with the scroll wheel; Zoom to selection fills the screen with the selected part, Whole file shows everything, and the slider under the waveform moves along when you're zoomed in. While it plays, the view follows the play point.

Under the waveform: Play (Space), Play selection, and Loop, which repeats the selection until you stop it. Start, End and Length show the selection in plain numbers, to the thousandth of a second; type a time in Start or End (like 1:23.5) to set it exactly. Go to jumps to any time you type. The big buttons work on the selection: Trim to this keeps only the selection, Delete removes it, Cut removes it and keeps a copy, Copy keeps a copy, Paste puts the copy at the play point (or over the selection), Make silent keeps the length but silences it, Fade in and Fade out fade across the selection, and Louder and Softer change it by 3 dB (the whole file when nothing is selected). Marker (M) drops a numbered marker at the play point; click a marker to jump to it, or its × to remove it. Undo and Redo are always at the top.

Every cut is exact to the sample, and each seam gets a tiny automatic crossfade (4 milliseconds) so it never clicks. Keys: Space play or pause, I and O set the start and end at the play point, Delete removes the selection, ⌘Z undo and Shift+⌘Z redo, ⌘X / ⌘C / ⌘V cut, copy and paste, ← and → move the play point (Alt for 10 ms steps, Shift to select), + and − zoom, T trim, S silence, F and G fade in and out, L loop, Esc clears the selection. Ask AI opens the plain-words panel beside the editor when you want it; Save opens the save choices in a window.

Add / Open has Music under it (two tracks): pick a music file or one from your Graphics Library, set each track's volume and how far the music dips while someone talks (it dips by itself), repeat it to the end or start it at the play point, press Hear the mix, then Put it in (Undo takes it out). Re-record this part (under the big buttons) records a new take from your microphone in place of the selected part; Stop and use this take puts it in with clean edges.

Spectrogram (next to Play) shows frequencies over time, low at the bottom and high at the top, so a click, cough or beep stands out. Drag a box around it and press Remove this sound: only the sound inside the box is lowered, and everything around it stays. Press Spectrogram again for the waveform.

Markers can have names: press the pencil on a marker (or double-click it), type a name like Verse one and press Enter. The arrow buttons (or [ and ]) jump to the previous and next marker. Chapters downloads them as a YouTube description (starting at 0:00) or a podcast chapters file, and they go into MP3 and FLAC files as chapters when you save the whole file.

Recipes (in Add / Open) keep a few steps plus how to save, such as trim silence, shorten pauses, level to -16 LUFS, MP3 for podcasts. Three starters are there; Make a recipe saves your own to your account (up to 20). Use on this file runs it and opens Save with its format. Batch (in Add / Open, or Batch them on the first screen) runs a recipe or a plain conversion on up to 50 files in your browser and downloads one ZIP.

In Save, Title, artist & cover art sets the title, artist, album, year, genre, comment and a cover picture, saved inside MP3, M4A and FLAC files. Share for comments turns the sound into a Customer Proof: a page with a waveform player where people leave comments at any second. Nothing is sent from Audio Kit; the comment page opens so you choose who gets the link.

The top line shows the file's name (with a frame from the video when it came from one), length, mono or stereo, sample rate and loudness in LUFS. The two arrow buttons are Undo and Redo for every change (also Ctrl or ⌘ + Z, and Shift for redo). Add / Open has Add a file at the end (marked with a split mark where it starts), Record more, Open from Graphics Library, and Open another file.

On the waveform, click to move the play point and drag to select a part. Zoom in, Zoom out and Whole file change how much you see; Ctrl or ⌘ with the mouse wheel zooms where you point. With the waveform focused, the arrow keys move the play point, Shift with the arrows selects, Home and End jump to the ends, and Space plays. Play the selection plays only the selected part. After a clean-up, the sound before it is drawn faintly behind, and Hear it before the change plays the earlier version from the same point.

Selected part: Keep only this, Cut it out, or Make it silent (the length stays). Split and join: Split at puts a split mark at the play point, so Save can save each part as its own file; Add 1 s of silence here puts a short gap at the play point. Fade: fade in at the start or out at the end over half a second to 8 seconds.

Volume makes the selection (or the whole file) louder or quieter by up to 18 dB. Leveling sets an even, standard loudness with peaks kept safe: Podcast (-16 LUFS), YouTube and Spotify (-14) or Broadcast (-23). Speed goes from half to twice as fast; Keep the voice's pitch stops the chipmunk effect. Pitch moves it up or down by up to 12 semitones without changing the length. Long pauses finds every pause longer than the time you pick, says how many and how much time they add up to, and Shorten them leaves a short natural gap in each.

Remove background noise learns the noise from the quietest moments (hiss, hum, fans, room noise) and lowers it everywhere while the voice stays. Pick Light, Medium or Strong; the waveform shows the change moving through the sound while it works. Make the voice clear cuts rumble and boom, adds presence, evens out quiet and loud words and levels it for podcasts. Both: noise + clear voice does the two in one step. Make it mono mixes both sides of a stereo file into one. All of these are free and run in your browser.

Split the voice from the music: the Quick voice remover takes out what sits in the middle of a stereo song, which is usually the lead voice, for a karaoke version. It is free and instant, and rough on some songs; it needs a stereo file. Studio split, an AI split into voice, drums, bass and the rest, is shown as not switched on yet. When it is switched on it shows its tokens before it starts, and a split that fails costs nothing.

Make the transcript listens to the sound and lays every word under it with its time. The button shows the tokens it uses (about 1 token for up to 10 minutes). Language can be found for you or picked from the list. While it listens, the words appear as each minute is done.

Click a word to jump there. Click another word with Shift held to select the words between, then Delete the selected words: the sound is cut with them. Ums and uhs are underlined; Remove the ums and uhs cuts them all in one go, and Shorten long pauses cuts pauses longer than 0.8 seconds. When you cut, speed up or join, the words move with the sound. .srt, .vtt and .txt download captions or a plain transcript for video players, YouTube or your notes. A thumbs up or down under the transcript tells us how good it was.

Made for picks the settings in plain words. Best quality keeps everything as good as the original allows (24-bit for WAV, FLAC and AIFF). Podcast makes one clear mono track leveled to -16 LUFS. Small file for email makes a small mono file that is easy to attach. Ringtone keeps up to 30 seconds (select a part first to choose which 30), makes it loud and clear with a soft ending, and saves an M4A as .m4r for iPhone; use MP3 for Android.

Format: MP3 (plays everywhere), M4A and AAC (Apple, iPhone, podcasts), WAV (full quality), FLAC (full quality, about half the size of WAV), OGG (web and Android) and AIFF (full quality for Mac apps). What to save: all of it, only the selection, or each part separately as a ZIP when there are split marks. The line under it shows about how big the file will be. Download makes the file in your browser; M4A, AAC and OGG need a current Chrome, Edge or Safari on a computer, and the screen says so if your browser can't make one. Save to Graphics Library keeps MP3, M4A and WAV files in an Audio Kit folder in your Graphics Library, ready for your other tools.

What's next? under Save sends the sound on: Put it in a video opens Video Kit with it as the music under your video, and Put it under an animation opens Motion Maker with it on its own track. The sound is saved to your Graphics Library first if it isn't there yet.

The panel beside the editor (under it on a phone) takes any change in your own words, typed or said with the microphone: "cut the first 5 seconds and fade in", "remove the background noise and make my voice clear", "remove every um", "make a 30-second ringtone from the chorus". It reads your request against the sound (its length, loudness, selection and, when there is one, the transcript; the sound itself is not sent) and runs Audio Kit's own steps. The answer lists anything Not done yet first, with why and what to do instead, then what was Done. Undo this change puts it back in one step, and a thumbs down with what went wrong can ask for a fix. Each request uses about 2 tokens; the edit itself is free. Changes so far lists every change in order; click one (or Original file) to go back to that point, and Redo goes forward again.

Voice Kit voiceovers and Music Maker songs have Clean up the sound in their What's next? row, and each Screen Recorder recording has a Clean up the sound button next to Download: it opens Audio Kit with the sound loaded (from a recording, the sound is pulled out of the video). Anything in your Graphics Library opens at /tools/audio-kit?asset=<its id>. Nothing is uploaded twice.