Voice Isolator Models by Inworld
Inworld releases · 3 in the last 12 months · this family litChangelog →
2 models2026
Last refreshed 2026-11-19. Next refresh: weekly.
Links
WebsiteAbout
Inworld’s offline and streaming speech-denoising models.
Decision facts
- Best fit
- speech enhancementcoding
- Capability starting point
- Voice Isolator v1
- Lowest tracked input
- Not tracked
- Closest related family
- SAVA
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
2 in view
Voice Isolator v1Current
Use when the workload needs speech enhancement, 64M parameters, and audio.
2026-11speech enhancement64M parametersaudio
Voice Isolator v1 RealtimeCurrent
Use when the workload needs speech enhancement, 58M parameters, and audio.
2026-11speech enhancement58M parametersaudio
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Voice Isolator v1 | Use when the workload needs speech enhancement, 64M parameters, and audio. | 2026-11 | speech enhancement64M parametersaudio | Current |
| Voice Isolator v1 Realtime | Use when the workload needs speech enhancement, 58M parameters, and audio. | 2026-11 | speech enhancement58M parametersaudio | Current |
Release Timeline
1 release group2026-11
2 current
Voice Isolator v1
Currentspeech enhancement64M parametersaudio
Voice Isolator v1 Realtime
Currentspeech enhancement58M parametersaudio
Specifications(2 models)
| Model | Released | Parameters |
|---|---|---|
| Voice Isolator v1 | 2026-11 | 64M |
| Voice Isolator v1 Realtime | 2026-11 | 58M |