v0.9.7 · Public beta open

Clarity for live voice.

One insert per vocal channel or group.
Real-time AI denoise, dereverb and feedback prevention —
at zero latency and ultra-low CPU.

  • Zero latency · ultra-low CPU
  • 48 / 96 / 192 kHz
  • Windows · macOS
Built for the rooms it ships in
Concert Venues Conferences Theaters Streaming studios Broadcast
/ Four jobs, one insert

Four problems your voice channel has,
solved at the same time.

Denoise

Removes HVAC, fans, hiss and console-fan rumble without smearing transients. Strength is one fader.

Feedback prevention

Catches mic-monitor coupling before resonance — not after. No notch filters to draw, no rings to chase.

Dereverb

Tightens room tails and bathroom-stage acoustics. Blend transparently with a single fader.

Debleed

Pulls down spill from drums, monitors and nearby instruments on the vocal mic, so the channel stays clean in a loud room.

0.0 ms latency
3 faders 4 buttons
1 insert per channel or group
/ Specs

Boring details, on purpose.

No surprises at load time.

Format
VST3 · AU
Engine
AI-driven · no notch filters or static EQ
Platforms
Windows 10/11 (64-bit) · macOS 10.13+ (Universal)
Tested in
Fourier Audio transform.engine · Waves SuperRack
Sample rates
48 / 96 / 192 kHz
Latency
0.0 ms
Channels
Mono or stereo
Window
940 × 500, resizable up to 1500 × 880
CPU
Low CPU · GPU-accelerated on Windows (DirectML) when available · run on every channel or group
Memory
~80 MB per instance
License
Per-seat · offline license file, no machine binding · install on all your own machines, one at a time
Coming Soon
/ Pre-launch

Be first in line.

Public beta is open. Drop your email — we'll send the beta download link, plus pre-launch pricing on launch day. No newsletter machine, no drip — only mails when there's something real to share.

We won't share your email. Unsubscribe with one click in the launch mail itself.

License TBA
/ seat
  • Use on all your own machines · one at a time
  • Free updates · v0.x → v1.x
  • Upcoming in-house AI model · free update
Estimated launch · Summer 2026
/ FAQ

Things people ask before buying.

Is it really zero latency?

Yes — internal latency is 0.0 ms. The AI core processes lookahead-free at the host's block size; the plugin reports zero PDC. Roundtrip latency depends on audio driver.

Can I run it on every channel of a 32-channel desk?

CPU permitting, yes. The SOFT (Low CPU) engine is designed for this case. On Windows the engine is also GPU-accelerated (DirectML) when a compatible GPU is available, which drops CPU load further and scales to many more instances. If no GPU is found it falls back to CPU automatically, nothing to configure.

What sample rates are supported?

48, 96 and 192 kHz. At unsupported rates the plugin shows a banner and bypasses audio cleanly.

How does this differ from a noise gate?

A gate cuts below a threshold. Clear Voice Live separates voice from background continuously, so you keep low-level speech and breath while the noise floor goes away.

What kind of filters are used?

No notch filters and no static EQ. Clear Voice Live uses AI-driven voice processing that separates voice from background in real time — at the right strength, the voice itself stays untouched while the noise floor and feedback resonances are reduced.

How does licensing work?

One license is per person, not per machine. After buying you get a personal license file: load it once, completely offline, with no activation and no machine binding. Install it on as many of your own machines as you like (studio, laptop, FOH rig, backup); the only rule is one machine in use at a time per license. Running it on two machines at once needs a second license.

Is there a trial?

A 10-day fully-functional trial ships with the public release. Get beta access.

What buffer size should I use?

For live use, 32 to 64 samples. Anything above ~64 samples introduces enough roundtrip latency that the artist hears their voice twice through the in-ear monitor. The plug-in itself adds 0.0 ms — driver and host buffer are what determine roundtrip.

Can I run multiple instances?

Yes. One instance per vocal channel is the standard setup. Routing many mics through an auto-mixer (Dugan-style) into a single group and running one instance on the group is one option — but because the CPU cost is so low, you can just as easily run a dedicated instance on every single channel for the best, most consistent result.

Where in the signal chain should it go?

Before compressors and EQs. The plug-in expects a relatively clean dynamic input — heavy compression upstream muddies the noise floor and makes the AI work harder.

Why does Dereverb seem to do nothing?

Dereverb is part of the same pipeline as Strength. If Strength is at 0 %, the entire processing chain is bypassed and Dereverb has no effect. Raise Strength first, then bring up Dereverb.

Where do I find the manual?

Online at clearvoice.live/manual. It also installs to your Documents folder under Clear Voice Live — the same place your presets live.

Where is it built?

Graz, Austria. Solo developer, direct support.