01Phase 01
1. Transient Spike Detection
Scans sample-level energy acceleration to pinpoint high-frequency non-harmonic saliva clicks and lip smacks.
Cubic De-Clicker · Zero Artifacts
Mouth Sound & Saliva Pop Remover
Eliminate distracting lip smacks, saliva clicks, and microphone transient pops without dampening speech clarity or uploading raw recordings to external servers.
Mouth Sound & Saliva Pop Remover
Ready for a local audio or video file
Local Browser Audio Engine
Scanning speech & quiet pauses…
Reading media and computing RMS energy…
1Reading file
2Decoding audio
3Scanning quiet
4Building map
Loudness:-16.0 LUFS
Peak:-1.0 dBFS
0:00 savedPacing Preset
Speech cushionsFine-tune detection parameters
Balanced
Balanced (~-38dB): Calibrated natural balance recommended for podcasts, YouTube videos, and dialogue.
0.80s
0.20s (all micro-pauses)2.50s (only long pauses)
0.32s
0.00s (flush / jump)0.80s (spacious)
Automatically removes silence before first speech and after last speech.
Pause #01✂️ Cut Proposed
00:14.2 → 00:16.8 (2.6s gap)
1.6s quiet gap; keeps 0.32s breathing room.
Tip: Press Del or Backspace to cut selected silenceR to keep
Silence Map (0)
0 silences found (0.0s dead air) · 0 will be cut, 0 kept
Target ScopeAll Clips
All edits reversible
Studio DSP & Ambience Rack
Real-Time DSP✨
1-Click Full Background Noise Removal
Removes 80Hz rumble, 50/60Hz hum, room hiss & mouth clicks
Anti-Blackout
Fills shortened silence gaps with authentic room tone synthesized from your recording, eliminating unnatural "vacuum dropouts".
Butterworth
-40 dBFS
+2.5dB Clarity
Sibilance
Cubic Spline
-1.0 dBFS
Subtitle Range Protection
Protect Intentional Sentences & Pauses
Importing your subtitle transcript locks dialogue boundaries so dramatic reflections and deliberate speech pauses inside sentences are never auto-cut.
No subtitles loaded
Hesitation Splicer (Fillers & Stutters)
Removes verbal hesitation sounds (um, uh, er, like) and repeated stutters.
0 fillers0 stutters
Merged with acoustic cuts
A/V Dual-System Sync
Drift ConformerEdit Evidence & Provenance
Manifest v1.0Total Proposals
0
Active Cuts
0
Restored Pauses
0
Protected Lines
0
Structural Export Validation:
✔ Zero server uploads (100% Client-Side)
✔ True-Peak Limiter ceiling ≤ -1.0 dBFS
✔ Duration conservation verified (Δ ≤ 5ms)
✔ 8ms anti-click zero-crossing micro-fades
Media Preflight & Capabilities
Source Format:Audio Track
Sample Rate:44.1 kHz
Channels:Stereo (2-ch)
Memory Footprint:~12 MB RAM
Local Processing:100% Client-Side
Verified Export Options:
- ✔ 16-Bit PCM WAV: Uncompressed lossless master audio.
- ✔ Audacity Labels: Drag-and-drop label track for Audacity / Reaper.
- ✔ CSV Cue Sheet: Tabular marker list with exact timestamps.
- ✔ EDL / FCPXML: Timeline cut lists: EDL for Premiere & Resolve, FCPXML for Final Cut & Resolve.
- ✔ Evidence Manifest: JSON & CSV audit trail with a CRC-32 source checksum.
Episode Show Notes
Silence #01 detected (00:04.4 → 00:05.0)
🎬Sequence Track1 clip
Pause Map & Timeline
Click / drag to scrub0:000:00
✨
✨ Cleaning Noise...Background Noise Remover
Multi-band spectral Wiener filtering & room rumble removal
Strong (75%)
0% (Off)50% (Balanced)100% (Studio Max)
Space Play/Pause · Del Trim · R Keep · J/L Jump 2sDouble-click waveform to trim
Lossless 16-Bit Master Audio & Edit Ledger Export
Rendered locally in browser memory with 8ms anti-click micro-fades. Zero server upload.
Detected from video
Chain in Studio FlowTimeline files reference your media by file name only. Final Cut Pro, DaVinci Resolve or Premiere Pro will ask you to relink the clip to the original file.
✨ Hover any tool icon to see what it does · Click to launch
100% browser-local processing · No files uploaded to any server
Acoustic Engineering Guide · Architecture
How this tool works in browser memory
How this tool decodes, analyzes, and masters your dialogue locally with zero cloud latency.
02Phase 02
2. Hermite Cubic Spline Repair
Replaces corrupted click samples with smooth cubic polynomial interpolation, preserving surrounding vocal harmonics.
03Phase 03
3. Multi-Level Sensitivity Control
Select Conservative, Balanced, or Sensitive modes to adapt to spoken dialogue, ASMR, or dynamic podcast speech.
Connected finishing workflows
Explore companion utilities for dialogue cleanup, speech pacing, and mastering.
Studio Audio Enhancer
Apply broadcast presence EQ, de-essing, and vocal warmth.
Background Noise Remover
Eliminate low-frequency rumble and AC hum.
Audio Silence Remover
Cut dead air with calibrated breathing cushions.
De-Esser
Tame harsh S and SH sounds with a split-band de-esser.
Audio Standards & Research
Engineering guides & deep-dive workflows
Explore the acoustic principles, ITU-R loudness standards, and pacing techniques behind this tool.
De-Clicker6 min read
How to Reduce Mouth Clicks and Lip Smacks in Recorded Speech
Understand the mechanics of saliva transients (2kHz–10kHz) and learn how cubic spline de-clicking repairs micro-bursts without dulling speech consonants.
Saliva Clicks7 min read
How to Remove Saliva Clicks and Mouth Noise from Voiceover Recordings
The physics of transient saliva clicks: detecting 1-4ms high-frequency energy spikes (>4.5σ) and repairing them with Hermite cubic spline interpolation.
Recording Technique7 min read
How to Make a Voice Recording Sound Clearer Before You Start Editing
The source-first clarity framework: capsule proximity, 45-degree plosive deflection, reflection absorption, and 24-bit headroom management.
Frequently Asked Questions
Common Questions & Technical Clarifications
How does the de-clicker detect mouth sounds, lip smacks, and saliva pops?+
The algorithm computes a 2nd-order discrete acceleration derivative across audio samples to detect abrupt, non-harmonic high-frequency energy spikes that exceed 4.5 times the local running standard deviation.
How are detected clicks and pops repaired without creating audio holes?+
When a transient burst (1ms–4ms) is detected, the corrupted samples are excised and reconstructed using 4-point Hermite / Catmull-Rom cubic polynomial spline interpolation, smoothly bridging the adjacent clean waveform slopes without audible dropouts.
Will the de-clicker alter crisp vocal consonants like 't', 'k', or 'p'?+
No. Consonants have wider spectral and temporal energy signatures than mouth clicks. The transient detector restricts repair exclusively to micro-burst spikes between 1ms and 4ms, leaving speech phonetics untouched.
Is the de-clicker processing performed locally on my computer?+
Yes. All transient scanning, matrix interpolation, and audio rendering execute entirely client-side using JavaScript and the Web Audio API without cloud servers.
What formats are supported for de-clicking voice recordings?+
You can process WAV, MP3, M4A, FLAC, AAC, OGG, and video containers (MP4, WebM, MOV) with immediate lossless 16-bit WAV export.