Noise reduction often removes the very sounds that make speech feel human: breaths, laughter, and pauses. This model treats those sounds as part of the foreground voice, aiming to suppress what surrounds a speaker without rewriting the recording. That distinction matters for voice datasets and live agents, where waveform changes can affect both identity and downstream speech recognition.
Voice Isolator
Separates speech from background noise at 48 kHz while preserving the speaker’s timbre, breaths, and laughter. Offers offline and streaming models; it does not restore missing audio detail or select a target speaker.
Introduction
Information
- Websitehuggingface.co
- OrganizationsInworld AI
- Published date2026/11/14
More Items
Transcribes speech audio into text across 45 languages, with automatic language detection. Its 0.5B parameters make it a compact option for multilingual transcription using Transformers or vLLM.
Combines Korean-English language and image understanding with screenshot-driven browser control in one model. It prioritizes retaining general LLM and VLM capabilities rather than optimizing solely for web tasks.
Turns photos into detailed foreground alpha mattes, with particular attention to hair, fur, shadows, and complex edges. Built on a generative image-editing model and distilled to four steps; self-hosted weights are non-commercial only and require substantial GPU memory.