Speak without turning the room into an audience.
Students often move to hallways, lower their voice, or switch to text when roommates are nearby. VoiceShell is built for that awkward middle ground.
Wearable voice privacy for semi-open spaces
Speak freely. Stay private.
VoiceShell is a lightweight acoustic wearable designed for dorms, shared offices, creators, and professional calls where speaking is useful, but being overheard is not.
835
valid survey responses in early user research
68.7%
reported delaying or avoiding voice calls when others were present
7
seed users across four early verticals
The missing layer in voice-first computing.
AI agents, dictation, online calls, and creator tools make voice more productive. Shared rooms make voice harder to use. VoiceShell focuses on the physical boundary that software cannot provide.
Calls and private work need a boundary.
Open offices solve collaboration, not confidentiality. VoiceShell gives users a way to discuss sensitive work without broadcasting every word.
Record, rehearse, consult, and ideate with less exposure.
From online tutoring to creator scripts, AI drafting, and sensitive consultations, the goal is making speech difficult to understand nearby.
Passive acoustics plus active masking.
The product direction combines a mouth-adjacent acoustic shell with embedded audio processing, aiming for privacy that feels wearable rather than theatrical.
Reverse-speech masking (OLaW)
Plays time-reversed Chinese speech outside the chamber to mask intelligible content. Unlike white noise, it occupies speech-like bands without mirroring the original message.
Personalized neural AEC
A user-specific neural AEC model learns clean speech and masking playback together. Calls stay clearer while reverse-speech masking is active.
Multi-cavity passive chamber
A multi-cavity chamber targets 500 Hz to 4 kHz, the core intelligibility band. It provides physical insertion loss as the base layer for active masking.
Wearable industrial design
Light mouth-adjacent hardware pairs with head or ear support for stable positioning. The form aims to stay practical for calls, study, and creator workflows.
Hear the difference.
Early A/B recordings will show how nearby speech changes with VoiceShell on and off.
Without VoiceShell — recorded 1m away
With VoiceShell — recorded 1m away
Early proof, built around real use.
The project is still in prototype development, but the direction is anchored in user interviews, acoustic tests, and staged hardware milestones.
P0
Prototype phase completed with initial acoustic validation and measurement workflow established.
7
High-value seed users identified across university, creator, medical, and professional-service scenarios.
EVT
Next stage focuses on engineering validation samples, structure refinement, and user testing.
From prototype to first production run.
A staged path: validate engineering, test with seed users, launch crowdfunding, then move toward manufacturing.
EVT samples
Second-generation structure, acoustic materials, and roughly 50 engineering samples.
DVT validation
Mold planning, 200 design-validation units, seed-user paid trials, and patent submissions.
Crowdfunding launch
Kickstarter and Indiegogo launch driven by real user orders and creator demonstrations.
CES showcase & MP preparation
Public launch target and mass-production preparation for the first commercial batch.
Be among the first to try it.
We're recruiting seed users for paid trials in 2026 Q4.
Email
voiceshell@outlook.com
X
@voiceshell_off
For early access, send a short note about your use case and city.
Team & contact
Built by an engineering student team based in Wuhan, China. Contact: Han Yujiang.
Email
voiceshell@outlook.com
X
@voiceshell_off
TikTok
tiktok.com/@voiceshell
Reddit
reddit.com/user/No-Promotion-7155