Device path
Choose the real headset and microphone, see connection state, and recover after they are plugged back in.
Technology and compatibility
It sits beside the software you already use.
The first target is a Windows 11 desktop. CRELAR supplies a processed listening path and microphone path. Calling, monitoring, and livestream apps keep doing their own work.
This is not a music widener or a bass booster. It is for voices in calls, monitoring, talkback, and spoken streams.
Shared base
Three industries share one desktop base.
They do not share default settings or acceptance tests. Contact centers, monitoring, and livestreams are validated separately.
Processing and format
Enhancement runs in audio processing on the user’s computer. Format conversion and bandwidth extension are different things.
A way back when something fails
Closing the window is not the same as stopping audio processing. A fault needs a fallback, and mute has to stay mute.
Where the scenarios differ
One engine, different default jobs.
| Capability | Contact centers | CCTV | Livestreams |
|---|---|---|---|
| Listening enhancement | Validate at launch | Validate at launch | Optional for call-ins |
| Microphone enhancement | Validate at launch | When talkback is used | Validate at launch |
| Original versus enhanced | Used in the pilot | Kept in playback | In the product, not on this site |
| Central configuration | Needed for deployment | As a project requires | Not a default screen |
| Many channels and integration | Platform integration later | Multi-channel later | Not needed at first |
Compatibility and data
Limits written into the page.
- Windows 11 desktop first
- Softphones, monitoring clients, livestream apps, and headsets are listed per environment. They are not pre-declared as universally compatible.
- Resampling is not bandwidth extension
- Turning telephone audio into 48 kHz only changes the processing or output format. Restoring missing frequencies needs its own evaluation, and the far end is still limited by the telephone codec.
- On-device processing is still to be checked
- The architecture aims to process audio on the agent or creator PC. A public claim that audio is never uploaded waits until model, diagnostic, update, and related service behavior is confirmed.
- Closing the window is not exiting the engine
- Closing the interface, bypassing the effect, and stopping audio processing are three actions. Sound does not automatically return to the previous device because the window closed.