SoundStudio Help

Quick start

First time on this PC? See Requirements and Using SoundStudio on your PC. New to audio words like LUFS, dBFS or Skip? The Glossary of terms defines them, and the Audio theory tutorial explains the ideas behind every tool, with diagrams and a step-by-step for cleaning up a song and making a radio spot.

1. Click Import (or press Ctrl+O), or drag audio files onto the page. Bulk Import loads several at once. Loaded files appear in the Files list on the left. Click one to make it active.
2. Press Space to play and pause. Click anywhere on the waveform to move the playhead there.
3. Drag across the waveform to select part of the audio. Most tools work on the selection if there is one, otherwise on the whole file.
4. Pick a tab below the waveform (EQ, Dynamics, Effects, Voice, Noise, Split, Tools, Enhance, Spectrum, Pitch, Helper) to work on the sound.
5. Made a mistake? Ctrl+Z undoes it. Small edits such as click fixes take almost no memory, so hundreds of them can be undone. Big whole-file edits share a memory allowance of about 500 MB; when it fills, the oldest steps drop off.
6. Press Ctrl+S to save into your work folder, or click Export to save as WAV (16 or 24-bit) or MP3. Autosave keeps your session safe if the page reloads.

Stopping a long job

Anything that takes more than half a second shows a bar at the bottom of the screen with what it is doing, how far along it is, and a ■ Stop button. Click Stop or press Esc. The job ends at its next checkpoint, usually within a fraction of a second.

If the whole page ever stops responding, close the tab and open SoundStudio again. Autosave brings your files back from the moment before the job started.

Requirements

SoundStudio itself (web page or installed app)

Local Helper (optional)

Using SoundStudio on your PC (local and offline)

All the audio work happens on your own computer. Your songs are never uploaded to jeffapierson.com or anywhere else: the page only loads the program, and the Helper only listens on your own machine (address 127.0.0.1, which nothing outside the PC can reach).

Three ways to run it

A good local setup

1. Install the app (Files tab, ⬇ Install as app) and pin it to the taskbar.
2. Click 📁 Work Folder and pick the folder where you keep your songs, for example Music\Suno. The browser remembers it; after a restart it may ask once to allow access again.
3. Leave Autosave session on.
4. Optional: run the Helper setup, then open the Helper tab and click Allow when Chrome asks.
5. Optional: set a watch folder in the Helper tab so new Suno downloads are checked and loudness-matched automatically.

Where your work lives

Working offline

Once the app is installed and opened once while online, it opens and works with no internet: editing, repair, Fit to Length, autosave, the Work Folder and WAV/MP3 export. The Helper works offline too, after each AI model has been downloaded once (use each tool once while online). When you are back online the app picks up the newest version by itself.

Keyboard shortcuts

Shortcuts work when you are not typing in a box. If a key seems dead, click on the waveform first.

Playback

View

Editing

Files

Mouse actions

Phones and tablets (touch)

With a selection made, a row of buttons appears under the scrollbar, since phones have no Delete key: Play (just the selection), Zoom (fill the screen with it), Repair, Smooth Cut, Delete, Undo and ✕ (clear). The Sel boxes take exact times too. With the Spike Pen on, a tap repairs the spike under your finger. On a phone the page stacks and scrolls like any web page: the editor on top (the waveform takes about a third of the screen height), then files, then the tool tabs with every control at full height; scroll down to reach them and back up to the waveform. A tablet in landscape, or a phone turned sideways, gives the waveform the most room.

Zoom and scroll

Zoom goes from the whole file down to 5 milliseconds across the screen, deep enough to see single samples. The zoom slider moves smoothly across that whole range. The time ruler adds finer marks as you zoom, down to 1 ms (00:10.003).

The scrollbar under the waveform

• Drag the thumb to move through the file.
• Click the empty track to jump one screen toward that side.
• ◀ ▶ buttons move a tenth of a screen per click and keep going while held. Shift+click moves a full screen.
• The readout on the right shows the exact span on screen, for example 00:10.001 - 00:10.006.
• Go to box: type a time such as 12.345 or 1:02.345 and press Enter. The view centers there and the playhead moves there.

During playback the view follows the playhead and turns the page when it runs off the edge.

Scrubbing (tape-style)

Scrubbing works like rocking reels by hand on a tape machine. The audio plays at the speed and in the direction you move. Drag slowly and it plays slow and low. Drag fast and it plays fast, up to 8x. Drag left and it plays backward. Stop moving and the sound stops.

Scrub only happens with the Scrub button on

Turn on the Scrub button (or press S). While it is on, any drag on the waveform, the playhead tab or the time ruler scrubs, and Alt+mouse wheel jogs in small audible steps. Press S again to go back to selecting. With Scrub off (since v3.19), nothing makes sound when you drag: the playhead tab and the time ruler just move the playhead, and clicking next to the playhead selects like anywhere else.

Hold the mouse near the left or right edge while zoomed in and the view keeps rolling in that direction. When you let go, the playhead stays put and Space plays from there. If audio was playing when you started, it picks up again from where you let go.

Selecting and editing

• The very start and end: keep dragging past the left or right edge of the waveform (even off the window) and the selection runs exactly to the start or end. Starting a drag within 8 pixels of either end snaps to it too. Zoomed in, holding past an edge rolls the view, faster the further out you go. Or use the ⇤ and ⇥ buttons beside the Sel boxes, or Shift+Home / Shift+End, to stretch the selection (or from the playhead) to the start or end.
• Drag across the waveform to select. A drag of less than 2 pixels counts as a click and clears the selection, so zoom in for very short selections: with 50 ms on screen, 0.5 ms is about 14 pixels wide.
• Exact selection: the Sel boxes under the waveform show the selection to 0.1 ms (four decimals). Type a start and end in seconds, for example 12.3450 and 12.3455, and press Enter. Type only a start and the selection is 0.5 ms long. The length shows next to the boxes.
• Select a found click exactly: right-click its mark and choose Select it.
• Cut, Delete, Paste, Duplicate and Remove Silence blend the joins over 6 ms, so edits no longer leave a click at the seam.
• Crop to Selection and Trim Start/End add a 3 ms fade at the new start and end for the same reason.
• Paste inserts at the playhead. It does not replace the selection.
• Very short cuts (under 6 ms, such as cutting out a spike) blend the audio just before and just after the cut instead of reusing the removed samples, so the spike cannot sneak back in. They remove about 2 ms extra.
• Smooth Cut (held notes) in the Tools tab, or Shift+Delete: to shorten a note that runs too long, select the part to remove from the middle of the note (at least 40 ms). The end of the cut is shifted up to 12 ms so the note's waveform continues in step, then the two sides are blended over 20 ms. The message says how well it matched; if the spot is not a steady note it blends anyway. The playhead lands 1 second before the join so you can press Space and listen.
• Effect buttons that say Selection / All work on the selection if you have one, otherwise the whole file, and blend 50 ms at the selection edges.
• Loop (↺) repeats the selection while playing (or the whole file with no selection), and the playhead loops with it. If the playhead is outside the selection, looping starts at the selection. The jump at the loop point may tick in your speakers; that sound is not in the file.

Tracks (up to 4, played together)

Press + Track (under the waveform, left of Sel). Pick Empty track, or one of your open files. Your current file becomes Track 1 and the new one Track 2. Add more up to 4.

• Lanes: every track is drawn as a lane under the main waveform, on the same timeline. The track marked editing is the one drawn big above: every tool, effect, Undo and the Breath Remover work on it. Click another lane (or its name in the Files list) to edit that track. The view, playhead and selection stay where they are.
• Which track you are editing: each track has its own color (Track 1 purple, Track 2 teal, Track 3 orange, Track 4 pink). The main waveform gets a frame in that color and a badge at the top saying Editing Track 2 (with the file name if it has one). The track's lane is outlined in the same color, its bar is tinted, and the Files list shows a T1, T2... tag beside each track. If you open a file that is not one of the tracks, the lanes stay on screen and the badge turns gray and says so; click a lane to go back.
• Putting audio on a track: select audio on any track and Copy (Ctrl+C). Click the track you want, put the playhead where the audio should start, and Paste (Ctrl+V). On empty space (silence) the clip is placed there and nothing after it moves. Pasting past a track's end makes it longer. Pasting on top of existing audio inserts, as it always has.
• Play plays all the tracks together. The track being edited goes through the live effects you have turned on; the others play as they are.
• Each lane has M (mute), S (solo: hear only soloed tracks) and a volume slider from -30 to +10 dB (double-click it for 0 dB). These work while playing.
• Export mixes all the tracks into one file, with Mute, Solo and volume counted and effects applied to the track being edited. Untick Mix all tracks in the Export box to save just the track being edited. Silence left at the end by empty track space is trimmed off.
• x on a lane takes that file out of the tracks. It stays open in the Files list.
• Ducking the music under a voice: click the music lane and add Volume Pins in the Tools tab. Pins show as orange bands on that track's lane, so you can see them while you edit the voice.
• Tracks are part of the autosave, so Restore brings them back.

Clicks and pops

Open the Noise tab and find Click & Pop Removal.

Recommended workflow

1. Press Find Clicks. Nothing changes in the audio. Each click is marked in red on the waveform.
2. Read the report:
   • Rate: clicks per minute.
   • Typical spacing: the usual time between clicks.
   • Evenly spaced (yellow note): a steady rhythm usually means the recording chain, such as a buffer or sample-rate setting, a USB or clock problem, or electrical interference. Fix that at the source for future recordings.
   • Length and Channels: how long the clicks are and whether they are in the left, right or both.
   • Density strip: bars show where clicks bunch up. Click the strip to jump there.
3. Use Next / Prev, or click a time in the list, to zoom to each click. Scrub across it to listen.
4. Fix them one at a time, or all at once:
   • One click, straight from the waveform: point at its red mark and double-click to fix it, or right-click for a small menu (Fix this click, Ignore, Zoom to it, Listen).
   • One click, other ways: pick it (Next / Prev, a row in the list, or a single click on its mark), then press Fix This Click, F, or Delete. Each row in the list also has its own Fix button. Only that click is repaired, and its mark turns green.
   • Too long or too ugly to repair? Drag across it and press Delete Selection in the Click & Pop panel, or right-click its mark and choose Cut it out. The audio is removed and the two sides are joined with a 6 ms blend so the cut itself cannot click. With nothing selected, Delete Selection cuts out the picked click. Marks after the cut move to stay on their clicks.
   • Not really a click? Press Ignore, I, or the ✕ on its row. It leaves the list and nothing is changed there.
   • Everything left on the list: press Remove Clicks. After Find Clicks it repairs exactly the clicks still listed, skipping the ones you ignored. With no list (or after Clear Marks) it scans and repairs in one pass.
   Every repair can be undone with Ctrl+Z.
5. Run Find Clicks again to confirm. It should find few or none.

Settings

Selection: if any part of the audio is selected, however short, Find Clicks and Remove Clicks only touch clicks inside that selection. With nothing selected they cover the whole file. Click once on the waveform (without dragging) to clear a selection.

Spikes you can see but the scan misses

Some ticks sit inside a noisy sound (a sung "s", a breath, a cymbal) where, statistically, they are barely different from the noise around them. Your eye still catches them because a thin line pokes out above the waveform's outline. Two tools are built for that:

• Spike Pen (button in Click & Pop Removal, or press P): turn it on, then click right on the spike. It finds the exact spike within about 1.5 ms of your click, in whichever channels have it, and fixes it. Choose Repair (rebuilds 0.8 ms: 0.25 ms before the peak and 0.55 ms after, which covers the spike's tail) or Cut out 0.8 ms. Each click is one undo step. Press P again to turn it off. Tested on three spikes in "Do You Hear Me Now" (0:50.31, 1:16.72, 2:43.20): after one click each, all three measured the same as the ordinary noise around them.
• Detector: Spikes above the outline scans for exactly what your eye sees: a 1 ms slice whose peak is several times taller than the typical peaks in the 25 ms around it, with its neighbours clearly lower. Set List order: Strongest first and work down the list with Next. Sensitivity 5 lists only the most obvious (about 30 in a busy song); 10 lists hundreds, strongest first. Anything quieter than -50 dBFS is ignored.

Suno songs

1. Set Detector to Suno ticks, Sensitivity 6 or 7, Repair range Full range.
2. Find Clicks, step through a few with Next and scrub across them to confirm, then Remove Clicks.
3. Switch Detector to Standard and run Find Clicks again to catch anything else. Expect drum hits and consonants to show up here: Ignore those.

In testing on a drum track with 20 planted ticks, the Suno fingerprint found 17 with no false alarms, while Standard flagged over 200 drum hits. On a vocal-style test with ticks placed inside consonants, it found 43 of 50 with no false alarms.

Click lists and Claude analysis

• Export List saves the current clicks as a CSV (time, length, channels, peak frequency, type, fixed or open).
• Import List loads a CSV of click times: first column the time in seconds (or m:ss.ssss), optional second column the length in ms, optional note. Imported clicks behave like found ones: Next / Prev, Fix, Ignore, Remove Clicks.
• Deeper analysis by Claude: upload the song (WAV is best, MP3 works) to a Claude chat and ask for a click analysis. Claude runs the same detectors offline with extra passes and sends back a summary plus <song>_suno_ticks.csv (high and medium confidence ticks) and <song>_clicks_all.csv (everything, including low-confidence and other clicks to review). Load the song in SoundStudio, then Import List.

Click frequency

Find Clicks measures the frequency makeup of every click by comparing the few milliseconds of the click with the audio just around it.

• Summary line: how many clicks are broadband, high ticks, midrange clicks or low thumps, with the typical band and peak.
• Graph (shows when you step to a click with Next / Prev or the list): red is the click, gray is the audio around it, the yellow line is the click's peak, and the shaded area is what Remove Clicks will rebuild.
• List: each row shows the click's peak frequency, or "broad" for broadband ones.

Repair range

Controls which frequencies of each click get rebuilt. Everything outside the range stays exactly as recorded, so the music under the click is left alone.

Repair Selection uses the same setting.

Noise tools (Noise tab)

Spectral Noise Reduction

1. Select a stretch that is only background noise, no voice or music. Half a second or more works best.
2. Press 1. Capture Noise Profile.
3. Set Reduction (how much to take out) and Smoothing, then press 2. Apply Noise Reduction.

Quick Noise Removal

Silence Removal

Remove Silence cuts every quiet stretch below the Threshold that lasts longer than Min Duration. Trim Start/End removes only the silence at the two ends.

Seeing volume by frequency

Spectrogram view

The View menu at the right of the scrollbar row (or the V key) switches the main display between Waveform, Wave + Spectrogram and Spectrogram. In the spectrogram, time runs left to right like the waveform, frequency runs bottom (50 Hz) to top (20 kHz), and brightness is level: black is silent, purple quiet, orange loud, yellow and white loudest. Point at it to read the frequency, the nearest musical note and the level. Selecting, zooming and clicking all work the same as on the waveform.

• A click or spike is a thin vertical line through all the frequencies.
• Hum is a steady horizontal line at 50 or 60 Hz (and its multiples above).
• Hiss is a haze across the top.
• A sung note is a stack of horizontal lines (its harmonics); a held note that goes on too long is easy to see end to end.

Spectrum tab

Shows the level of every frequency as a curve. Analyze Selection / File averages the selection (or the whole file with nothing selected). Live while playing follows what you hear, with an optional peak hold line. Point at the curve to read the frequency, note and level; with smoothing on, the readout also gives the strongest peak nearby. Smoothing (1/6 octave by default) makes the overall tone easier to read; pick None to see single tones and hum sharply. Show L and R draws each channel. The Octave bands list totals the level in each octave from 31.5 Hz to 16 kHz, a quick way to spot too much bass or too little top. A full-scale pure tone reads 0 dB, so a tone at -6 dBFS reads about -6.

The small spectrum in the Meters tab uses the same scales while playing.

Pitch: finding and fixing a voice that drifts

Use this when a voice changes pitch where it should not: a voice ad pieced together from different takes, an AI voice that creeps higher or lower from one sentence to the next, or a line that came out sounding higher after other edits.

Seeing the pitch

• View menu → Wave + Pitch (or press V until it shows) draws the voice's pitch as a line under the waveform, with note names at the left. The dashed teal line is the reference pitch and the green band around it is the tolerance. Each phrase gets a short bar at its typical pitch: green if it is within the tolerance, orange with the amount (for example +3.1 st) if it drifts.
• Point at the pitch line to read the frequency, the note and how far it is from the reference.
• The Pitch tab has a Now meter that follows playback: the needle sits in the middle when the voice is at the reference pitch, and moves left (lower) or right (higher), turning orange or red when it is past the tolerance.

Fixing it

1. Open the Pitch tab. It reads the pitch automatically (about 2 seconds for a 4 minute file).
2. Pick the reference: Whole file uses the voice's typical pitch; or select a part that sounds right and click Use selection.
3. Set the Tolerance. Normal speech moves up and down within a sentence, so compare whole phrases: 2 semitones is a good start; use 1 for a steady announcer voice.
4. The Phrases list shows each phrase's pitch and how far off it is. ▶ plays it. Fix moves that phrase back to the reference; Fix all orange phrases does them all as one step (one Ctrl+Z undoes it).
5. For finer control, select a part and use Shift by hand (for example -0.5 semitones), or Match selection to reference.

Notes: pitch reading needs the voice by itself, before music is mixed in (for a finished song, split out the vocals in the Helper tab first). Corrections under 2 semitones sound natural; bigger ones start to change the voice's character. If the Helper is running and Fit to Length's engine is set to Rubber Band, corrections use it and keep the voice character better.

Mixing files together (voice and music)

Load the files (for example a voice recording and a music track), then click Mix Files Together in the Files tab (the same button is in the Helper tab). Each file gets a row:

• Checkbox: include the file in the mix.
• Level slider and dB box: how loud that file is. 0 leaves it as is, -6 is about half as loud, -12 sits well in the background. Double-click the slider for 0.
• M mutes the file. S (solo) plays only the soloed files, handy for checking one part.
• Start at: delays the file, in seconds. For a radio spot, start the voice 1 to 2 seconds into the music.

Press Play and move the sliders while it plays; you hear every change right away. Drag the position bar to jump around. The Mix level meter shows the loudest point: green is fine, yellow is close, red means the mix goes over 0 dB. Nothing is turned down for you: if it goes over, Make Mix File says by how much. Lower a level (usually the music) until the meter stays out of the red, or the export will clip there.

To turn the music down only while the voice talks (ducking), put the files on tracks and use Volume Pins on the music track.

Make Mix File saves exactly what you hear as a new file named "... - mix.wav" and opens it. The original files are not changed, and each file remembers its level and start time the next time you open the mixer. Good starting levels: a voice over a music bed, voice 0 and music -12 to -18; singing over a backing track, voice 0 and music -3 to -6.

EQ tab (parametric)

Ten bands plus Low Cut and High Cut. Each band can sit on any frequency from 20 Hz to 20 kHz, and you set how wide it is. The EQ is live: you hear it while playing, and it can be baked into the file on export (Apply Effects) or right away with Apply EQ to Selection / All.

Three ways to adjust a band

• Drag a numbered dot on the curve. Left and right moves the frequency, up and down sets the gain. The mouse wheel over a dot makes it narrower or wider. Double-click a dot to flatten it. On a phone, drag the dots with a finger.
• Use the slider under each band for gain, as before.
• Type exact values in the band editor: click a band's frequency label (under its slider) or its dot to select it, then set Type, Frequency, Gain and Width. The editor also shows the nearest musical note and how wide the band is in octaves or semitones. Reset band puts it back to its standard position, flat.

Types and widths

• Bell: boosts or cuts around one frequency. Most bands use this.
• Low shelf / High shelf: raises or lowers everything below or above the frequency. Bands 1 and 10 start as shelves.
• Notch: removes one frequency completely, with nothing around it touched. Use it for hum (50 or 60 Hz and its multiples) or a single whistling tone.
• Width presets: Wide (about 2 octaves, for tone shaping), Medium (about 1.4), Narrow (about half an octave), Very narrow, and Hairline (for hum). The Q box is the same thing as a number: higher Q is narrower.

While playing, the live spectrum is drawn faintly behind the curve, so you can line a band up with what you hear. The selected band's own shape is drawn in yellow.

Finding a problem frequency

In the Spectrum tab, click a peak on the curve. The app finds the strongest point near your click and offers Cut 4 dB (narrow), Notch it out or Boost 3 dB (wide). It then moves a free EQ band to that exact frequency and opens the EQ tab so you can fine-tune it while listening. A classic way to find a harsh or ringing frequency by ear: boost a narrow band a lot, sweep it slowly until the problem jumps out, then turn that band into a cut.

Presets (Flat, Bass Boost, Treble Boost, Vocal, Loudness, Mid Scoop, Telephone, Warm, Bright, Outdoors, Stadium, Rock, Folk) put all ten bands back at their standard frequencies before setting the gains.

Dynamics tab

Suggested Presets set the Compressor, Limiter and Gate values together. They never switch anything on: tick Enable to hear it, or use the Apply buttons. Press Fit to This Audio after picking one to measure the selection (or the whole file) and set its Threshold, Makeup and Gate to suit the level:

• Music: 2:1 with a soft knee, Threshold just above the song's average level, so only the loud parts are touched by a few dB. Keeps the song's level. Limiter at -1 dB. No gate.
• Voice: 3:1, Threshold a little under the voice's average level, Makeup set so the average lands near -18 dB. Gate set 10 dB above the room noise, lowering the gaps at least 15 dB (more if the makeup raised the noise).
• Advertising: 5:1 with fast attack and release, Makeup set so the average lands near -13 dB for a dense, loud spot. Firmer gate.
• All Off puts everything back.

The line under the presets shows what was measured and set. If the room noise is within 18 dB of the voice, the gate is left off (it would chop words); clean the noise in the Noise tab first. If a recording is so quiet it needs more than +30 dB of makeup, Normalize it first.

• Compressor: Threshold, Ratio, Attack, Release, Knee, Makeup Gain. Makeup Gain is the only gain it adds: before v3.17.3 the browser's built-in compressor quietly added up to 12 dB more on top. Tick Enable to hear it live, or press Apply Compressor to Selection / All.
• Limiter: sets a Ceiling the level cannot pass, with 5 ms look-ahead so even the first edge of a peak is caught. It leaves anything under the Ceiling alone. Release is how fast the level comes back after a peak: 30 to 100 ms suits most material. Much shorter distorts bass; much longer makes everything duck after each loud hit. Apply Limiter to Selection / All writes it into the audio.
• Gate: turns down anything under the Threshold, handy between phrases. Hold keeps it open a moment after the sound drops so word endings are not chopped. Range is how far down it goes: Mute (-80) silences the gaps, while -10 to -20 dB just lowers the room noise and sounds more natural on voice. Set the Threshold a few dB above the noise between phrases. It works live when Enabled, in Export, and with Apply Gate to Selection / All.
• Normalize: raises or lowers the whole file so the loudest peak hits the Target Peak. RMS Normalize goes by average level instead.

Effects tab

Stereo Width: 0% folds the file to mono, 100% leaves it as recorded, 200% is extra wide. It works by turning up or down only what differs between the left and right channels (the "side"), leaving what they share (the "mid", usually the voice and bass) alone. Above 150% the app warns you: very wide mixes can lose parts when played on a mono speaker, so check with Width at 0%. Mono files are not changed. Double-click the slider to return to 100%. Works live, in Export, and with Apply Width to Selection / All.

Pre-delay is the gap before the reverb starts (it keeps a voice clear of its own echo). Damping darkens the reverb tail: 0% bright, 100% rolled off above 2 kHz.

Suggested Presets: Music (plate-style reverb about 1.75 s at 15% with 20 ms pre-delay, soft 375 ms echo), Voice (small room, 6%, just enough that a dry read does not sound boxed in), Advertising (almost dry, 4%, so the voice stays up front). All three reset Chorus and Distortion and leave them off. All Off turns everything off.

Reverb (Room, Hall, Cathedral, Plate, Spring), Delay/Echo, Chorus, Distortion and Stereo Width. Each has an Enable box for live listening and an Apply button to write it into the audio.

Pitch Shift changes the pitch by up to 12 semitones and keeps the length. Time Stretch (pitch kept) changes the speed from 25% to 400% and keeps the pitch. (Before v3.5 both worked like a tape speed change, so pitch and length always moved together.)

Fit to Length (radio time compression)

Radio calls this time compression (or time squeezing): making a spot exactly :30 or :60 without the voice sounding sped up or chipmunked. It is in the Effects tab.

1. Load the read. To fit only part of it (for example the voice, leaving a music tag at the end alone), select that part first.
2. Type the length you need in Target: 30, 29.5 or 0:30. The line under it shows the current length and how much faster that is.
3. Choose Pauses:
   • Tighten lightly first (default): takes up to 40% out of each gap between phrases, always leaving at least 150 ms, before speeding anything up.
   • Tighten firmly first: up to 70% of each gap, leaving at least 90 ms. Punchier, closer to how a producer edits a read by hand.
   • Leave pauses alone: the whole read is sped up evenly.
4. Click Fit to Length. The result lands exactly on the target, to the sample. The report says how much came from pauses and how much from speeding up. Ctrl+Z undoes it.

How it works: gaps are found as stretches more than 30 dB below the loud parts, and each cut is blended over 8 ms so no clicks appear. The rest is sped up by cutting the audio into 30 ms pieces and overlapping them a little closer together, lining each piece up with the one before so the pitch and tone stay the same. Both channels are handled together so the stereo image does not move.

Voice tab

Breath Remover

Finds every breath in the selection (or the whole file, if nothing is selected) and fixes them all at once.

• Find and Fix All Breaths: one click, done. Ctrl+Z undoes all of it in one step.
• Find and Review: lists them first. Click a row to select that breath on the waveform, press its play button to hear it with a little before and after, untick any that are not breaths, then Fix Ticked Breaths. Worth doing the first time on a new voice or room.
• Action: Turn down by 3 to 40 dB (15 is a good start; the voice still sounds like it breathes, just not in your ear), Silence (same length, breath gone), or Cut out (removes the breath and closes the gap, which also shortens a spot).
• Sensitivity: Gentle takes only clear breaths, 0.18 s or longer and at least 14 dB under the words. Normal takes 0.13 s and up, 10 dB under. Strong also catches short, louder breaths but may touch soft "f", "th" and "h" sounds, so review it.

How it decides: a breath is noise with no pitch, quieter than the words, and at least a tenth of a second long. Words with pitch (vowels, m, n, l) are never touched, and "s" sounds are skipped because their energy sits high above 4 kHz. The edges fade over 12 ms so nothing clicks.

• Extract Voices keeps the vocal range and reduces the music. Remove Voices (Karaoke) does the opposite. Results depend on the mix. Center-panned vocals separate best.
• De-Esser tames harsh S and SH sounds around the chosen Frequency.
• Vocal Enhancer: Presence, Warmth and Air.

Split and naming (Split tab)

• Auto Split by Silence: finds gaps and drops markers. Adjust Silence Thresh, Min Silence, Min Segment and Padding, then Detect Split Points.
• Manual: Split at Playhead, Split at Selection, or Shift+click on the waveform.
• Time-Based: Auto Split Every N sec or Split at Next N sec.
• Detect Chapters suggests longer natural breaks.
• Naming Template: Base Name, Pattern and Start #. Rename any segment in the Split Regions list.
• Import loads split points from a CSV (time in seconds or MM:SS.mmm in the first column, optional name in the second) or a standard .cue file.
• Export All Segments saves each piece as its own file, optionally as one ZIP. Each file gets a 3 ms fade at its start and end so it does not click.
• Split markers have their own undo: Ctrl+Shift+Z and Ctrl+Shift+Y.

Ducking and the volume envelope

Ducking means turning one sound down while another plays, then bringing it back: the music bed (underscore) drops when the voice talks and comes back up in the gaps. Radio spots, podcasts and video voiceovers all do it. A typical duck is 10 to 18 dB under a voice, with the music already down by the time the first word lands.

The volume envelope is the line that shows a track's level over time. Flat at the top is the track as it is. Every duck (or boost) bends the line down (or up) for a stretch, with a ramp in and a ramp out. In SoundStudio the envelope is drawn as an orange line over the waveform and you shape it with Volume Pins: each pin is one bend in the line.

Why not just lower the whole music track? Because then the music is too quiet in the gaps and at the start and end, where it carries the spot. Ducking keeps it full when nobody is talking.

Volume Pins

In the Tools tab. A pin turns the track you are editing down (a duck) or up (a boost) for a stretch, with its own level, ramp times and curve. Up to 24 pins per track.

Quick way: duck music under a voice

1. Put the voice and the music on tracks, then click the music lane so the badge says you are editing it.
2. Make pins one of three ways:
   • While listening: press Play, press B where the voice starts and N where it stops. Each B and N pair makes a pin. (The Mark Start and Mark End buttons do the same.)
   • From a selection: select the stretch and click Add Pin from Selection.
   • From the voice: open Suggest pins from the voice, pick the voice track, click Suggest Pins. It finds each stretch of talking (pauses shorter than Join pauses under stay inside one pin; sounds shorter than Ignore under are skipped). Suggestions are gray and dashed and do nothing until you click Accept on each, or Accept All Suggestions. Preview on a suggestion lets you hear it first.
3. Set one pin's Level, ramps and Curve the way you like, then click Copy to all pins on it so every pin matches (their times stay).
4. Tick Hear pins while playing and play it through. Adjust.
5. Apply Pins writes them into the music track when you are happy, as one Undo step.

How a pin sits

The level changes during the Ramp in, which ends at the pin's Start, stays at the pin's Level until its End, then comes back during the Ramp out, which starts at End. So if Start is where the voice begins, the music is already down when the first word lands.

Tight control, per pin

• Start and End: type 1:23.456 or 83.456 and press Enter. ◀ ▶ move them by the Nudge step (1 ms, 10 ms, 100 ms or 1 s). To playhead sets it to where the playhead is, even while playing. Go moves the playhead there.
• Level: -60 to +12 dB in 0.5 dB steps. Minus is a duck, plus is a boost (green on screen). -10 dB sounds about half as loud; -12 dB leaves a quarter of the signal.
• Ramp in and Ramp out: 0 to 10,000 ms each. 200 to 500 ms in and 600 to 1,500 ms out sound natural under speech; music usually comes back slower than it goes down.
• Curve: see below. A small picture next to it shows the pin's shape.
• On the waveform, the bar along the bottom has a block per pin: drag a white end to move Start or End, a thin end to set that ramp, or the middle to move the whole pin. Everything else on the waveform works as usual above the bar.
• On a phone or tablet: the pin bar is taller and you drag the same parts with a finger: a white end, a thin ramp end, or the middle. A finger can land a little above the bar and still catch it. A second finger cancels the drag and pinches to zoom instead.
• Overlapping pins: any duck wins over a boost, and the deepest duck wins.

Curves: how smooth the ramps are

The curve decides how the level travels from where it was to the pin's Level during a ramp. All of them reach the same level in the same time; they differ in where the change happens. The ear notices corners (a sudden start or stop of the change) more than it notices the speed, which is why the smooth curves sound most natural.

• Smooth (S-curve), the default: starts gently, moves fastest in the middle, lands gently. Good for nearly all ducking.
• Extra smooth: even softer at both ends and a bit quicker through the middle. Best for long ramps (a second or more) and exposed music such as solo piano, where you want no hint of where the ramp starts or stops.
• Straight: an even change in dB from start to end. Simple and predictable, but the start and end can sound like small corners on long ramps.
• Quick, then settle: most of the change happens right away, then eases into place. For getting the music out of the way fast, when a voice starts hard. On the ramp out it comes back slowly at first.
• Slow, then quick: hangs on, then finishes fast. Lets the music ring until just before the voice. On the ramp out it jumps back quickly.

All curves work in dB, the way the ear hears level, and playback follows them to within 5 ms. If a duck still sounds abrupt, a longer ramp helps more than a different curve: try doubling the Ramp in.

Hearing and applying

• Hear pins while playing starts unticked. Ticked, playback follows the pins on every track, at any speed. Preview In plays from 2 s before the ramp in to 3 s after Start, Preview Out from 3 s before End to 2 s after the ramp out, Play Pin the whole pin. The Preview buttons play with the pins heard without ticking the box.
• Apply Pins writes every pin that is On into the track's audio, as one Undo step. Applied pins stay on screen, gray and dashed. Undo takes the audio back and makes those pins editable again; other pin changes you made since are kept. If a boost pushes peaks over 0 dB, the message says so; nothing is turned down.
• Export writes what is in the audio. Pins not applied are not in the file, and the Export box says so in red.

Pins follow your edits

Cut, Delete, Paste (insert), Duplicate, Crop, Smooth Cut, the Breath Remover's Cut mode, Fit to Length and Time Stretch move the pins with the audio: cut 2 s before a pin and the pin moves 2 s earlier. A pin entirely inside a removed part goes with it, and the message says so. Undo puts the pins back with the audio, and pins you added or changed after the edit are kept (they move with the audio too). Silence Removal and Trim cannot be followed; after those the panel warns that the length changed so you can check them, and They are fine clears the warning.

Saved layouts

Open Saved layouts. Save Layout keeps the pins (times, levels, ramps, curves) under a name in this browser. Load Layout adds them to the track you are editing, either at the same times or At playhead (moved so the first pin starts at the playhead). Save Layout File downloads them as a small .json file you can keep or move to another PC, and Load Layout File reads one back. Layouts are also saved in a project (Projects tab).

Pins are part of the autosave, so Restore brings them back. The Hear box starts unticked every time.

Delivery Check (loudness targets)

In the Enhance tab. Pick where the audio is going, click Check, and it measures the finished mix and says pass or fail. It changes nothing.

• Measure: As it would export mixes the tracks the way Export does (Mute, Solo, track volume) and includes pins only if they are applied. As you hear it includes the Volume Pins too.
• The report gives integrated loudness (LUFS), true peak (dBTP), sample peak, loudness range (LRA) and the loudest 3 seconds, then what to do: how many dB to turn down, or how far it can go up before the true peak passes the limit.
• Use this target in LUFS Normalization copies the target into the LUFS Normalization slider above it (it does not normalize).
• Live effects are not measured; the report reminds you if any are on.

Tools tab

Enhance tab

• One-Click Enhance: ready-made chains for Music, Voice/Speech, Podcast, Master, Loudness, Cleanup, Warmth, Brighten, Outdoors, Stadium, Rock and Folk.
• Auto Analyzer & Enhance measures true loudness (LUFS, the streaming standard), true peak, clipped peaks, infrasonic rumble and tonal balance across the whole song.
  • Analyze only reports and changes nothing.
  • Music mode (for finished songs, including Suno) makes one even volume change to your loudness target and never goes past -1 dBTP. If the target would need more, it stops short instead of squashing the song. Optional: rebuild genuinely clipped peaks (on by default, only if found), remove infrasonic rumble below 20 Hz, gentle tone correction up to 2 dB (off by default). No gate, no added compression. On a test Suno song the result differed from the original only by the volume change.
  • Voice mode (raw speech) adds an 80 Hz high-pass, gentle compression and a smooth fade-down between phrases.
  • Ctrl+Z undoes the whole thing.
• Loudness Analysis and Normalize to Target LUFS set overall loudness. About -14 LUFS suits streaming, -16 suits podcasts. With Use limiter unticked (how it starts) it stops short of the target if going further would push peaks past -1 dB, and the message says where it stopped. Tick Use limiter to reach the target, with the limiter holding peaks at -1 dB.

Export

Bulk Apply (header) runs the checked steps on every loaded file, in this order: Repair clipping, Noise reduction (needs a captured noise profile), EQ, Compression, Auto Enhance, Loudness normalize (LUFS), Peak normalize, then optionally exports each one as WAV or MP3 named name_processed. Each step uses the settings currently in its panel. Pick a work folder first so the exports do not ask for each file. Every file keeps its own Undo.

Analyze (header button)

Click Analyze at the top. It measures the audio and opens a window with tabs. Open in new window moves it into its own browser window (not a tab) so it can sit beside the editor or on a second screen; its buttons still work on the editor. Analyze only measures. Every change is a button you press, and those buttons only set up the tool: nothing is written into the audio until you press that tool's Apply or Normalize.

• What to analyze: the track being edited, every track, or just the selection. Purpose (Podcast or voice, Music master, Radio or ad spot, Video or social) sets the targets for the recommendations; it starts as the project's purpose, or a guess from the style.
• Overview: loudness, true peak, range, length, tempo and style, then each problem with a color: red fix it, yellow worth a look, blue for your information.
• Sound Profile: level by frequency band, in 8, 16, 32, 64 or 128 bands from 20 Hz to 20 kHz. Each band shows its average level and the highest it ever reaches (red when a band reaches 0 dB). The table lists the numbers.
• Problems, each with what it is, what to do, and buttons:
  • Hum (50 or 60 Hz mains hum, found in the quiet parts) and Buzz (hum with many harmonics): lists exactly which frequencies to notch; Notch buttons put a notch (Q 12) on each in the EQ.
  • Harsh "s" sounds (sibilance): how many and the frequency where they peak; Cut sets a 5 dB cut there (Q 3).
  • Clipping (including clipping that went through MP3): Show selects the first one; Go to Repair Clipping takes you to that button in the Tools tab.
  • Clicks or pops: Show goes to the first; the Noise tab's Click and Pop Removal repairs them. Drum hits can look like clicks, so listen.
  • Background noise, Rumble below 30 Hz (with low cut buttons), DC offset, left and right out of phase, ringing or resonant tones in a voice (narrow peaks, with Cut buttons), and was an MP3 or AAC at some point (a sharp loss of top end that no WAV can bring back).
• Recommended Tweaks: the fixes in the order to do them for the purpose you picked (clipping, noise, hum, low cut, "s" sounds, clicks, evening out the level, loudness to target, peaks, trimming silence), each with a button that takes you to the tool with the values set.
• Track and File: sample rate, channels, sample and true peak, loudness, loudness range, RMS and crest, noise floor, stereo correlation and balance, DC offset, silence at each end, tempo with confidence, and the file as opened (type, size and bitrate, date, and whether it was ever an MP3).
• Metadata: Title, Artist, Album, Year, Genre, Track number and Comment. They are read from the file when it opens (MP3 ID3 tags, WAV INFO tags), you can edit them, and Save keeps them with the file (autosave and projects). Exports write them: WAV as INFO tags, MP3 as ID3 tags. When tracks are mixed, the tags of the track being edited are used (or the first track that has some).
• Style: an estimate made right in the app (spoken word, voice over music, music, speech with background, or music without a clear beat), with the tone (dark, balanced or bright), tempo and dynamics, and why it thinks so.

Playback Check: file, PC or headphones?

In Analyze, the Playback Check tab plays test sounds straight from the app, not from your file, so anything wrong with them is in your playback chain (PC, cable, interface, headphones or speakers).

The tab also shows your output (sample rate, channels, delay). A long delay usually means Bluetooth, which can drop to a muffled mono "hands-free" mode when a microphone app is open; wired headphones are best for editing. It lists the Windows settings worth checking (Audio enhancements, Spatial sound, Mono audio, Format), since a web page cannot read them.

A/B Compare at matched loudness

In Analyze, the A/B Compare tab. Pick A and B from: any open file as edited now, the same file as first opened, with its Volume Pins, all tracks mixed, or a saved version from the open project. Play from the playhead plays both at once, and Hearing A / Hearing B (or the Space bar in that window) switches instantly at the same spot. With Loudness: Matched (how it starts), the louder one is turned down to the quieter one's loudness (measured in LUFS), so you judge the sound and not the volume: a louder version almost always sounds "better" at first, which is why matching matters. Nothing is changed in either version.

Projects (Projects tab)

A project keeps everything for one job in its own folder on your PC: the tracks, numbered saves you can go back to, optional autosaves, and a log of what was done. It needs Chrome or Edge (the installed app counts).

Starting one

1. Open the files (and set up tracks) the way you want, then click Projects (left side, next to Meters) and New Project.
2. Type a name and pick the purpose (Podcast or voice, Music master, Radio or ad spot, Video or social). Analyze uses the purpose for its recommendations.
3. The first time, the app asks for your projects folder. The usual place is Documents\Sound Studio: in the folder picker open Documents, click New folder, name it Sound Studio, and pick it. After that, new projects go there without asking (tick Put it in a different folder to choose another).
4. It then asks whether to turn on autosave. It is off unless you say yes.

What is in the folder

SoundStudio never deletes anything in a project folder. The Projects tab shows the project's size (MB or GB); remove old saves or autosaves in File Explorer yourself if it gets big.

Saving and going back

• Save (or Ctrl+Shift+S) stores a new numbered version of everything open. Type a note first if you like ("before EQ", "client mix 2"); it shows in the list. Ctrl+S still saves the active file into the Work Folder, as before.
• Under the Save button it says All changes saved or Changes not saved to the project.
• Saved versions lists every save, newest first. Open on one rolls back to it: the files open now are closed (they stay in Recent files with their edits) and that version's files, tracks and pins open exactly as they were. If you have unsaved changes it offers Save first, then open.
• Autosave every 2, 5, 10, 15 or 30 minutes: stores a numbered autosave only when something changed. Autosaves are listed under Autosaves and open the same way.
• Open Project opens a project folder (pick the project's own folder, the one with project.json). Recent projects lists the last ones. After the app restarts, click Reconnect to give it access to the last project again (the browser asks once).
• The browser's own autosave (Files tab) keeps working too, separately from projects.

Saving, autosave and the work folder

• Autosave (Files tab, on by default) stores every open file's current audio, the tracks, each file's Volume Pins, the selection, the playhead and the click list in the browser's own storage a few seconds after each change. If the page reloads, crashes or you close it, the next visit shows a bar at the top: Restore it brings everything back. Start fresh keeps the old autosave until you change something, then replaces it. Undo history is not kept across a reload. Autosave works in every browser and without internet.
• Unsaved reminder: a yellow ● in front of a file's name (in the Files list and on the Recent list) means it has changes you have not saved to a file yet. Save (Ctrl+S) or Export clears it. If you close the tab, close the app window or reload while any file shows the dot, the browser asks "Leave site? Changes you made may not be saved" first; choose Cancel to stay. Browsers do not let a page change that wording. Turn it off with Warn before closing if not saved in the Files tab. Phone browsers often skip this box, so save before switching apps. Even if you leave, autosave usually has your latest edits on the Recent list.
• Recent files (Files tab, under the open files): every file you close stays in the browser with your edits, the click list, name and length. Click one to open it exactly as you left it. Keep last sets how many closed files are kept (3, 5, 10 or 20; default 10); when a newer one is closed, the oldest drops off. The x next to a file removes it (you are asked first). The line underneath shows how much browser storage they use: about 115 MB per 5-minute stereo song, so 10 songs is roughly 1 GB. Recent files need Autosave on. Like autosave, they live in this browser only, so Save or Export anything you want to keep for good.
• Work Folder (Chrome and Edge) connects a folder on your computer. The audio files in it are listed under the file list, click one to open it. Save or Ctrl+S writes the active file there as name (edited).wav in 24-bit. If that file already exists from before, you are asked before it is replaced. Exports and Bulk Apply exports go to the same folder.
• The browser remembers the folder. After a reload it may ask once to allow access again: click Work Folder or Save and approve.

Local Helper (Helper tab)

The SoundStudio Helper is an optional program on your PC that does jobs too heavy for a browser. SoundStudio works fully without it; when it is running, the Helper tab lights up (green dot). It lives in %USERPROFILE%\SoundStudioHelper and starts by itself when you sign in to Windows. The first time a page uses it, Chrome asks to let the site reach apps on this device: choose Allow. To install it on a PC, see Helper setup.

Other programs: Audacity Labels (Noise tab, next to Export List) saves the click list as an Audacity label track, File, Import, Labels in Audacity. iZotope RX does not import label lists, so use the times in the CSV to jump there.

Helper setup

1. Download the setup, either way works:
   • SoundStudioHelper-setup.zip (130 KB, recommended): the helper files plus the installer. Right-click the zip, Extract All, open the extracted folder.
   • or just install.cmd (5 KB): it downloads the helper files itself. Chrome may say the file type can harm your computer: click Keep.
   install.cmd is a plain text script; right-click it, Edit, to read every step.
2. Double-click install.cmd. If Windows shows "Windows protected your PC", click More info, Run anyway. If winget asks to accept terms, type Y.
3. Wait 5 to 15 minutes. It installs FFmpeg, uv, Node.js (if missing), Python 3.11 and the AI tools into %USERPROFILE%\SoundStudioHelper, adds a Startup shortcut, and starts the helper. It ends with "Done".
4. Open SoundStudio, click the Helper tab. Chrome asks to let the site reach apps on this device: click Allow. The dot turns green and lists the tools.
5. First use of each AI tool downloads its model once: Demucs about 80 MB, Whisper Small about 480 MB, DeepFilterNet about 10 MB.

Install as an app

SoundStudio still runs as a normal web page. In Chrome or Edge you can also install it:

• Click ⬇ Install as app in the Files tab (or the install icon at the right end of the address bar, or the browser menu, Install SoundStudio).
• It gets its own window, a Start menu and taskbar entry, and no browser tabs or address bar, so the editor has more room.
• Works offline after one visit: the page, help and the MP3 and ZIP encoders are kept on your computer. When you are online, it always loads the newest version.
• Open audio files with it: in File Explorer, right-click a WAV, MP3, FLAC, M4A or other audio file, Open with, SoundStudio. The file opens in the app window that is already open.
• Autosave, the work folder and everything else work the same as in the browser. To remove it, open the app, click the three dots at the top, Uninstall SoundStudio. Your files are not touched.

Files, presets and meters (left side)

• Files: every loaded file. Click one to switch. Each file keeps its own Undo and Redo history, so Undo never reaches into another file.
• Presets: quick presets, plus Save Current as Preset for your own settings.
• Meters: stereo dB meters for L and R, scaled 0 to -60 dBFS. The solid bar is the peak (falls 24 dB per second), the pale inner bar is the average (RMS), the white line holds the recent peak for 1.5 seconds, and the red box at the top lights when a channel reaches 0 dB (click the meter to reset it). Numbers underneath show peak, RMS and hold. The meters run during playback and scrubbing, show what you hear (after the Vol slider and live effects), and a spectrum runs below them.

Troubleshooting

What changed

v3.26 · Oct 9, 2026

• Volume Pins can be dragged with a finger on phones and tablets: a taller pin bar, finger-sized grab areas, and a second finger cancels the drag and zooms instead.

v3.25.1 · Oct 3, 2026

• Fixed: on the website, v3.25 opened as a page of code instead of the app. The web host adds its own script to every page, and a piece of the new Analyze code confused where it put it. Nothing else changed.

v3.25 · Oct 2, 2026 (quality pass)

• Undo after an edit that moved Volume Pins now keeps pins you added or changed afterwards (they move with the audio). Before, Undo of any edit could put the pins back the way they were before that edit and lose later pin changes. Undo of Apply Pins now only un-applies those pins.
• Projects never write over an existing save or audio file, even with the project open in two windows. "Save first, then open" no longer opens the old version if the save failed. Files closed by opening a version keep their pins and tags in Recent files.
• Volume Pins on the other tracks follow the audio correctly while looping a selection.
• Loading a damaged pin layout file can no longer make silent or broken pins.
• Delivery Check says so when every track is muted, instead of hanging.
• WAV tags are placed correctly when the audio data has an odd length (24-bit mono).
• Analyze: fewer false alarms (clipping and hum on music, clicks on drum hits, background noise on continuous music), the "s" frequency is measured from the "s" sounds themselves, and long files keep the page responsive while analyzing. Its Repair Clipping button now takes you to the tool instead of running it, and A/B loudness is a choice (Matched or As they are) instead of a ticked box.
• Enhance tab, Loudness Analysis: now shows real integrated LUFS and true peak (it showed an RMS estimate labeled approx before) on separate lines.
• Glossary: new terms (A/B compare, BPM, Buzz, Correlation, Delivery target, Envelope, Metadata, Project, Ramp and curve, Resonance, Rumble), and Volume Pins is filed under V.

v3.24 · Oct 2, 2026

• Analyze button (top bar): opens a window you can move into its own browser window. Sound profile in 8 to 128 bands with the highest level per band; problems (hum, buzz, harsh "s" sounds, clipping, clicks and pops, noise, rumble, DC offset, phase, ringing, past MP3 compression) with the exact frequencies and a button for each fix; recommended tweaks in order for the purpose; track and file details; style estimate.
• Metadata: title, artist and other tags are read when a file opens, can be edited in Analyze, and are written into WAV and MP3 exports.
• Playback Check: test tones, sweeps, left/right, phase and a silence test to tell whether a problem is in the file, your PC or your headphones.
• A/B Compare at matched loudness: original against edited, versions from the project, with or without pins.
• The top bar has one new button, Analyze, after Export Options.

v3.23 · Oct 2, 2026

• Projects tab (left side, next to Meters): a folder per project, usually in Documents\Sound Studio. Save (Ctrl+Shift+S) stores numbered versions (v001, v002...) with an optional note, and Open on any of them rolls back. Optional autosave (off unless you turn it on; the app asks when you make a project) stores numbered autosaves. A change log records what was done. The project's size shows in MB or GB. Nothing in a project folder is ever deleted by the app.
• The four tabs on the left are a little narrower so Projects fits.

v3.22 · Oct 2, 2026

• Volume Pins: B and N (or Mark Start and Mark End) make a pin while you listen. Suggest pins from the voice proposes gray pins where the voice talks, which do nothing until you Accept them. Copy to all pins gives every pin the same level, ramps and curve. Boosts up to +12 dB as well as ducks. Pins move with Cut, Delete, Paste, Crop, Smooth Cut, Breath Remover cuts, Fit to Length and Time Stretch, and Undo puts them back. Saved layouts (in the browser or as a file). Up to 24 pins per track.
• Five ramp curves (Smooth, Extra smooth, Straight, Quick then settle, Slow then quick), each with a picture of its shape and a note on when to use it. Fade down and Fade up are now called Ramp in and Ramp out, since a pin can go up too.
• Delivery Check in the Enhance tab: measures the mix against TV (US and Europe), podcast, streaming, social or your own target and says pass or fail and by how much. Changes nothing.
• Help: new sections on ducking and the volume envelope, and on curves.

v3.21 · Oct 2, 2026

• Volume Pins in the Tools tab: turn a music bed down in up to 8 places per track, each with its own Drop, Fade down, Fade up and curve, set to the millisecond by typing, nudging, To playhead or dragging on the waveform. Heard live with Hear pins (starts unticked) or the Preview buttons, written into the audio only with Apply Pins (one Undo step). See Volume Pins.
• The Export box warns when a track has pins that are not applied.
• Nothing else in the app changed.

v3.20 · Oct 2, 2026

• Clearer indicator of the track being edited: a frame and an "Editing Track N" badge on the main waveform in that track's color, matching colored lanes, and T1 to T4 tags in the Files list.
• The track lanes no longer disappear when you open a file that is not one of the tracks. The badge says that file is not a track, and clicking a lane goes back.

v3.19.4 · Sep 29, 2026

• Nothing happens automatically any more. Every feature is still there; each one does only what you click.
• Export writes exactly what is in the editor: no limiting, no turning down. A peak at 0 dB is 0 dB in any other app. If the audio goes over 0 dB, the message says by how much (a file cannot hold more, so it clips there).
• Apply buttons, RMS Normalize, Repair Clipping and Make Mix File no longer limit or turn anything down on their own. If the result goes over 0 dB, the message tells you and you decide.
• Apply Effects in the Export box starts unticked. Every Enable box starts unticked each time the app opens.
• Presets (Dynamics, Effects, Reverb) set slider values only and never tick an Enable box. All Off unticks them. Dynamics presets no longer measure your audio on their own: press Fit to This Audio when you want that.
• LUFS Normalize's Use limiter box starts unticked.
• MP3 export of audio above 48 kHz: MP3 cannot hold more than 48 kHz, so the MP3 is saved at 48 kHz and the message says so. WAV keeps the full rate.

v3.19.3 · Sep 29, 2026

• The Export box lists the live effects you are hearing. With Apply Effects unticked it warns in red that they will not be in the file, which is the usual reason an export sounds quieter than the app. It also warns when the Vol slider is not at 100% (Vol changes only what you hear).
• After an export, a message gives the file's peak and loudness (LUFS), and repeats any live effects that were left out.

v3.19.2 · Sep 29, 2026

• Repair Clipping now finds clipping that went through MP3. When clipped audio is saved as MP3 the flat tops come back with a small ripple, so the old check (exactly flat tops only) said "No flat-topped peaks found" on audio that was plainly clipped. On a test song clipped 6 dB and saved as MP3 it now finds and rebuilds 876 peaks; on clean songs it still finds none.
• When there is nothing to repair it says why: the loudest peak level, that peaks only touching 0 dB light the meter without being clipped, and that distortion already softened into the audio cannot be rebuilt.
• LUFS Normalization reaches the target: a new Use limiter box (on) raises the audio to the target and holds peaks at -1 dB, instead of stopping short. Untick it for the old behavior.
• The Target Peak and Target LUFS sliders show their value as you move them. Normalize says where the peak ended up. RMS Normalize can no longer leave the file clipping.

v3.19.1 · Sep 29, 2026

• No more turning whole files down. Before, whenever an effect, an EQ boost or an Export pushed any peak past -1 dB, the entire file was turned down until that one peak fit, so a single bass hit could lower a whole song by 3 dB or more, and it added up with every step. Now only the peaks are held by the limiter and everything else keeps its level. If peaks go more than 3 dB over, the part above 3 dB is turned down evenly so the limiter never squashes more than 3 dB.
• Export no longer touches a file that does not clip. Before, every Export with Apply Effects ticked turned a normal song down about 1 dB even with no effects on.
• Selections can no longer clip when an effect is applied to them.
• Reverb and Delay keep the original at full level up to 50% Mix and add the effect on top. Before, any Mix turned the original down.

v3.19 · Sep 29, 2026

• Tracks: up to 4 tracks drawn as lanes on one timeline, played together, mixed on Export, with Mute, Solo and volume on each. Paste onto empty track space places the clip without moving anything else.
• Selecting to the very start or end: drag past the edge (even off the window), snap within 8 pixels of either end, the ⇤ and ⇥ buttons, Shift+Home and Shift+End. Selection drags keep working when the mouse leaves the waveform.
• Scrub only happens with the Scrub button on. Before, dragging the playhead tab or the ruler, Alt+drag, and clicking within 5 pixels of the playhead all scrubbed, which got in the way of selecting at the start of a file. Now those just move the playhead or select.

v3.18 · Sep 25, 2026

• Breath Remover in the Voice tab: finds every breath and turns them down, silences them or cuts them out in one step, with an optional review list to listen to each one first.

v3.17.4 · Sep 25, 2026

• Stereo Width works. Before, the slider only changed the number next to it. Now it works live, in Export, and with a new Apply Width button, with a mono-check warning above 150%.

v3.17.3 · Sep 25, 2026

• Compressor and Limiter rebuilt. The browser's compressor they used added its own hidden makeup gain (about +6 dB at the old defaults, up to +12 dB) and was not a real ceiling: with the Ceiling at -10 dB, loud audio still came out at -3 dB. Now the Compressor adds only the Makeup you set, and the Limiter holds the Ceiling exactly. What you hear live, Apply and Export all run the same code.
• Dynamics presets measure the audio and fit the Threshold, Makeup and Gate to it. The fixed numbers in v3.17.2 made Voice and Advertising quieter on normal recordings and too loud on quiet ones, and left room noise louder in the gaps.
• Gate Attack and Release are now the time to open and close fully, so 150 ms means 150 ms.
• Reverb Pre-delay and Damping now do what they say. Before, both sliders were ignored.

v3.17.2 · Sep 25, 2026

• Suggested presets on the Dynamics and Effects tabs: Music, Voice, Advertising and All Off.
• The Gate now does something. Before this version its controls were there but it never touched the sound. It works live, in Export, and with the new Apply Gate button.
• Gate Range (mute, or just turn the gaps down) and Hold.
• Limiter Release (was fixed at 10 ms, now 50 ms by default) and an Apply Limiter button.

v3.17.1 · Sep 25, 2026

• Ctrl+V pastes again. Since v3.14 the V key (change view) was catching Ctrl+V first, so paste only worked from the Paste button.
• Copy, Cut, Paste and Select All also answer to Cmd on a Mac and work with Caps Lock on.
• Paste with nothing copied now says so instead of doing nothing.

v3.17 · Sep 25, 2026

• Pitch tab and a Wave + Pitch view: see the voice's pitch over time, a live pitch meter while playing, and a list of phrases that drift higher or lower than a reference pitch. Fix one phrase or all of them, or shift a selection by hand in tenths of a semitone. Length never changes.

v3.16 · Sep 25, 2026

• Mix Files Together is now a live mixer: press Play and set each file's level while you listen, with Mute, Solo, a start time for each file and a mix level meter. Make Mix File saves exactly what you hear. Also in the Files tab.
• Files with different sample rates can be mixed; the mix uses the highest rate.

v3.15 · Sep 24, 2026

• Parametric EQ: move any band to any frequency, set its width (Q), and pick Bell, Low shelf, High shelf or Notch.
• Drag the dots on the EQ curve (wheel for width, double-click to flatten), or type exact values in the new band editor.
• Click a peak in the Spectrum tab to send that frequency straight to an EQ band.
• The live spectrum shows behind the EQ curve while playing.

v3.14 · Sep 24, 2026

• Spectrogram view (View menu or V) with frequency, note and level under the pointer.
• Spectrum tab: level by frequency for the selection, the whole file, or live while playing, with octave bands.
• The Meters spectrum now has frequency and dB scales.

v3.13 · Sep 24, 2026

• Stop button (or Esc) for long jobs, with a progress bar. Helper jobs stop on the PC as well (helper v1.1). Stop jobs on PC button in the Helper tab.
• Effects no longer add an Undo step when they are stopped or fail.

v3.12 · Sep 24, 2026

• Reminder before closing or reloading with unsaved changes, and a yellow dot on files that are changed but not saved or exported.

v3.11 · Sep 24, 2026

• Recent files: the last 10 files you worked on (you can choose 3 to 20) are kept with their edits and click lists and can be opened again from the Files tab.
• The transport's Scan control is now called Skip, which is what it does: how far « and » jump.

v3.10 · Sep 24, 2026

• New pages: Glossary of terms (over 110 terms, searchable) and Audio theory tutorial (16 sections with diagrams). Both linked from the top of this page.
• Scan explained in the transport bar tooltip and the glossary.
• Help tables scroll sideways on phones instead of widening the page.

v3.9.1 · Sep 24, 2026

• Phones: the tool panels are no longer squeezed into a small scrolling box; the whole page grows and scrolls. Taller waveform on phones. Phones held sideways and small laptop screens no longer run off the right edge.

v3.9 · Sep 24, 2026

• Touch screens: drag to select, drag selection ends, tap to move the playhead, drag the playhead or ruler to scrub, pinch to zoom, two fingers to scroll, touch scrollbar.
• Selection buttons for touch (Play, Zoom, Repair, Smooth Cut, Delete, Undo, Clear).
• Phone layout: header, transport and scrollbar rows wrap to the screen instead of running off the side; bigger touch targets.

v3.8 · Sep 23, 2026

• Helper setup can be downloaded as a zip (server now allows downloads from the helper folder only).
• Help: requirements, using SoundStudio on your PC and offline, Helper setup download and install steps, helper troubleshooting.
• Helper setup program (install.cmd) for any Windows PC.

v3.7 · Sep 23, 2026

• Helper tab: stem separation, Mix Files Together, Rubber Band engine, AI noise reduction, lyrics timing, FLAC/MP3/M4A/Opus export, open any format, watch folder.
• Click list can be saved as Audacity labels.

v3.6 · Sep 23, 2026

• Install as an app: own window, works offline, opens audio files from File Explorer.

v3.5 · Sep 23, 2026

• Fit to Length for radio spots: type 30 and a 35 second read comes out exactly 30 seconds, pauses tightened first, pitch unchanged.
• Time Stretch now keeps the pitch. Pitch Shift now keeps the length.

v3.4 · Sep 23, 2026

• Autosave and restore after a refresh or crash.
• Work folder: open from and save straight into a folder, Ctrl+S saves a 24-bit WAV there.
• WAV 24-bit export. OGG removed (it was a WAV in disguise). Split export as MP3 now really writes MP3.
• Export with effects can no longer clip.
• Repair Clipping rebuilt: flat tops are rebuilt from the surrounding audio instead of averaging single samples.
• Loudness normalize now measures real LUFS (BS.1770) and stops at -1 dBTP instead of clipping.
• Bulk Apply now does every step it lists, and can export each file.
• Each file has its own Undo history. Removing a file from the list no longer switches the editor to the wrong file.

v3.3 · Sep 23, 2026

• Edits, repairs, cuts, Undo and Redo made while playing are now heard right away. The browser plays a snapshot taken when playback starts, so before this you kept hearing the old audio until you stopped and played again.
• Ctrl+Z / Ctrl+Y now work even right after using a dropdown or slider (only typing in a text box keeps them for the text).

v3.2 · Sep 23, 2026

• Spike Pen: click a spike you can see to repair or cut it.
• New detector, Spikes above the outline, plus a Strongest first list order and a strength score on each click.
• Smooth Cut for shortening held notes.
• Very short deletions no longer mix any of the removed audio back in.
• Repair Selection on a tiny selection over a spike now covers the spike's tail.

v3.1 · Sep 23, 2026

• Playback speed control. The playhead now loops with the audio. Stereo dB meters that run during playback and scrubbing.

v3.0 · Sep 23, 2026

• Auto Analyzer & Enhance rebuilt: the old version gated quiet passages in 20 ms chunks, bent every peak near 0 dB, boosted treble on nearly every song, always compressed and pushed up to +12 dB into a fake limiter. The new one measures properly and only changes what is safe.
• Whole-file effects and One-Click Enhance presets can no longer clip: the result is turned down evenly if it would pass -1 dBFS.
• Sub-millisecond click search: actual click length shown, Only clicks shorter than filter.
• Suno fingerprint no longer mistakes the pulses of a buzzy voice or synth for ticks, and ignores the first and last 50 ms of a file.

v2.9 · Sep 23, 2026

• Suno tick fingerprint detector (Detector: Standard, Suno ticks, Both).
• Import List and Export List for click lists, including lists from a Claude analysis of an uploaded song.

v2.8 · Sep 23, 2026

• Select down to 0.1 ms: typed Sel boxes, finer drag, Select it on click marks, tools no longer ignore selections under 10 ms.
• Detection of very short spikes (0.1 to 0.3 ms, the kind Suno adds) no longer skips them at small Max click length settings, and marked spans are tighter.
• Repair Selection grows to cover the whole click when the selection falls short.

v2.7 · Sep 23, 2026

• Fixed the freeze after a few dozen edits. Undo now saves only the part each edit changes: 85 edits on a 4-minute song went from about 7.5 GB to about 180 MB. Undo and Redo also restore click marks.
• Delete Selection in the Click & Pop panel, and Cut it out in the click right-click menu.

v2.6 · Sep 23, 2026

• Point at a click mark: double-click fixes it, right-click opens a menu.
• Jumping to a click keeps your zoom by default (new Jump to click setting).
• The page no longer scrolls the waveform and buttons out of view: only the panels below scroll, and the header stays on one line on laptop screens.
• Detection handles very clean recordings and loud bursts better (no more whole-block false events).

v2.5 · Sep 23, 2026

• Fix or ignore one click at a time: Fix This Click, Ignore, per-row Fix and ✕ buttons, click a mark to pick it, F, I, Delete.
• After Find Clicks, Remove Clicks repairs exactly the listed clicks.
• Long pops: Max click length now goes to 100 ms, each event is followed until it dies away, and Repair Selection accepts up to 250 ms.

v2.4 · Sep 23, 2026

• Click frequency analysis: summary by type, per-click graph, peak frequency in the list.
• Repair range: Full, Auto (each click's own band), Above cutoff, Below cutoff.
• Much cleaner rebuilds of longer clicks over low notes. The model now scales with the click length.

v2.3 · Sep 23, 2026

• This help page, a Help button in the header, and the ? / F1 keys.
• Crop to Selection fades its new edges so it cannot click.

v2.2 · Sep 23, 2026

• Tape-style scrubbing: playhead tab, ruler drag, Scrub mode (S), Alt+drag, Alt+wheel jog.
• Version and date shown in the header.

v2.1 · Sep 23, 2026

• New Click & Pop Removal: Find Clicks report with rate, spacing and pattern check, marks, Next/Prev, Remove Clicks, Repair Selection. The old Remove Clicks button did nothing and now runs the new remover.
• Cut, Paste, Delete, Duplicate and Remove Silence blend their joins. Trim and exported split segments fade their edges.
• Timeline: scrollbar, fine wheel and arrow-key scrolling, Go to box, zoom down to 5 ms, millisecond ruler, faster drawing on long files.