Skip to content

VoiceCleaner 1.1.4

Choose a tag to compare

View all tags
@BuddyDoQ BuddyDoQ released this 14 Oct 16:09
· 17 commits to main since this release
v1.1.4
0aac853

VoiceCleaner 1.1.4 by Steamburger Studios

What's new in 1.1.4

  • Opens more than WAV. Phone voice memos (M4A/AAC), MP3, FLAC, Opus/OGG, AIFF, WMA and
    AMR, and the sound of video files such as Zoom recordings (MP4, MOV, MKV, WebM, AVI): in the
    Enhance tab, Batch, Compile and Playback. The first audio track is used; surround is mixed
    down to stereo. Results are WAV (or MP3), named after the original. M4A, AAC, WMA and video
    are decoded by an LGPL build of FFmpeg that ships with VoiceCleaner (its source code is
    attached to the release).
  • Batch preview: turn on Preview in the Batch tab (P) and click a file to hear a
    10-second sample (the part with the most speech) enhanced with the current settings, A/B
    at equal loudness. Clicking another file previews that one, and the sample follows setting
    changes, so you can tune before enhancing everything.
  • Command line: VoiceCleanerCLI --batch recordings out --preset Natural enhances files
    and folders without the window (--format mp3, --bits 16, --recursive, --no-smart,
    --list-presets), with one progress line per file and an exit code for scripts.
  • Recording options (Enhance tab, Record): a 3-second countdown before recording
    starts, and Record 1 s of room tone first: a quiet second (Stay quiet → Speak now)
    that becomes the noise profile of the take. Big, colour-coded cues show what to do: GET
    READY · 3    , STAY QUIET, SPEAK NOW. Both options are remembered.
  • Clipping repair. Recordings made too loud have their cut-off peaks rebuilt (sparse
    declipping): every unclipped sample stays as recorded, and the lost peaks are filled in
    from the surrounding sound. It runs automatically when clipping is detected (Advanced →
    Repair clipping    ). On clipped test speech, intelligibility (STOI) after enhancing rose
    from 0.88-0.97 to 0.94-0.99.
  • Enhance pauses playback (whichever track is playing) until it is done; Play waits until
    then, so you never hear a half-finished state.
  • Fixed: the installer's side panel showed the VoiceCleaner name cut off (1.1.3).
  • README: download links and what VoiceCleaner does come first; building from source
    moved to BUILDING.md.

Speech cleanup and enhancement for WAV recordings: noise and room reduction, neural voice
re-synthesis, EQ and tonal balancing, dynamic leveling, loudness, recording (microphone and
desktop audio), sessions and playback. Works offline.

Which download?

Download Size
Most PCs (standard, CPU) VoiceCleaner-1.1.4-win64.msi (installer) 290 MB
VoiceCleaner-1.1.4-win64-portable.zip (no install) 353 MB
NVIDIA graphics card (faster) VoiceCleaner-GPU-1.1.4-web.exe: small installer that downloads the rest 1 MB (downloads 2.25 GB)
VoiceCleaner-GPU-1.1.4-win64-portable.zip.001 + .002 (no install) 3.00 GB
  • Voice re-synthesis model: the 490 MB model is downloaded once, from NVIDIA's
    official repository, the first time you use Voice Re-synthesis. Everything else works
    offline.
  • Standard vs GPU: both editions install to the same place, and installing one
    replaces the other.
  • GPU portable ZIP: open the .001 file with 7-Zip. Or join
    the parts in a command prompt with `copy /b VoiceCleaner-GPU-1.1.4-win64-portable.zip.001
    • VoiceCleaner-GPU-1.1.4-win64-portable.zip.002 VoiceCleaner-GPU.zip`, then unzip.
  • GPU installer without the web installer: download the VoiceCleaner-GPU-1.1.4-win64.msi
    and every vcgpu*.cab file into one folder, then run the .msi.
  • Unsigned builds: Windows SmartScreen may warn on first run. Choose More info →
    Run anyway. Verify downloads with SHA256SUMS.txt.
  • FFmpeg: VoiceCleaner ships an LGPL build of FFmpeg to open M4A, AAC, WMA and video files.
    Its source code is attached (ffmpeg-source-*.tar.gz).

Requirements: Windows 10/11 64-bit, 8 GB RAM. GPU edition: NVIDIA driver 570 or newer.

macOS (arm64)

Download Size
Mac (Apple Silicon, M1 or later) VoiceCleaner-1.1.4-macos-arm64.dmg: open it and drag VoiceCleaner to Applications 197 MB
VoiceCleaner-1.1.4-macos-arm64.zip: the same app, zipped 197 MB
SHA256SUMS-macos.txt: checksums to verify the downloads

Requires macOS 12 or newer. Noise reduction runs on the CPU and Voice Re-synthesis on the
Apple GPU (Apple Silicon). The 490 MB voice re-synthesis model downloads the first time that
feature is used; everything else works offline.

First launch: this build is not notarized by Apple, so macOS blocks it the first time.

  1. Open VoiceCleaner from Applications. When macOS says it cannot verify it, click Done.
  2. Open System Settings > Privacy & Security, scroll down and click Open Anyway
    next to the VoiceCleaner message, then confirm with your password.

You only need to do this once. (On macOS 14 and earlier you can also right-click
VoiceCleaner and choose Open; macOS 15 Sequoia removed that shortcut.)

Desktop audio: System audio (all apps) records what the Mac plays (macOS 14.2 or newer;
macOS asks once for System Audio Recording permission). On older macOS, install a loopback
driver such as BlackHole; it is then listed under Desktop audio. Microphone recording works
on every version; macOS asks for permission the first time.

Opens more than WAV: M4A voice memos, AAC, MP3, FLAC, Opus/OGG, AIFF, WMA, AMR and the
sound of videos (MP4, MOV, MKV, WebM, AVI). Formats beyond WAV/FLAC/MP3/OGG/AIFF are decoded by
an LGPL build of FFmpeg inside the app, compiled from the source attached to this release.

Files: preferences in ~/Library/Application Support/VoiceCleaner, logs in
~/Library/Logs/VoiceCleaner, sessions in ~/Documents/VoiceCleaner Sessions.

Assets 17
Loading