• Member
• November 6, 2026

ANEK Room

Private conversations in shared rooms

Summary

This note extends the ANEK concept from a family car to a room. The idea is to let a person take a phone call, or talk with a colleague or visitor, in a shared space without the conversation being understood by people nearby and without wearing headphones. The approach combines three things: sound that is focused on the listener and masked for everyone else, an optical or camera-assisted way of capturing the speaker’s voice so that quiet speech is enough, and a digital twin of the room that is used to design the filters and to place reflecting, diffusing and absorbing surfaces deliberately. Light-based sound delivery and ultrasound beams that bounce off a surface are treated as add-ons to be tested, not as the foundation. The honest target is a conversation that is unintelligible and discreet to neighbours, not one that is perfectly silent outside.

1. Promise

In a shared room, your conversation stays yours. A counter clerk, a patient at a registration desk, an employee at a shared desk or a family member in a living room can take a call or hold a conversation, and the people a few metres away hear only a soft, unintelligible background. The promise is made in terms a person can feel: no earpiece to wear and no booth to walk to, and nothing that isolates the person from the room.

The promise has a boundary that should be stated from the start. The system can hide what the listener hears, which is the other party’s voice, because that sound can be delivered locally. It cannot make the speaker’s own voice disappear, because that voice is spoken into the room. What it can do is let the person speak quietly, capture that quiet speech well, and mask what leaks. For this reason the correct claim is discretion and unintelligibility rather than secrecy.

2. Lineage

The idea rests on four lines of earlier work. The oldest and most widely used is sound masking in offices, in which a steady, broadband sound is spread through a space so that overheard conversations become less intelligible without anyone wearing headphones or sitting in a booth [S3]. Practitioners describe masking as not cancelling speech but shortening the distance over which a talker can be understood [S4], and field studies have tested natural sounds such as flowing water for the same purpose [S5].

The second line is personal audio and sound zone control, in which a loudspeaker array delivers separate programmes to different positions in one space. The Southampton work on speech privacy lists open-plan offices and museum exhibits among its contexts [S1]. Its closest result to this concept is that masking noise can itself be focused toward the positions of potential eavesdroppers, that the constant background noise of a room can be used as part of the masking, and that using both reduces the contrast the array must achieve [S2].

The third line is directional sound from light and ultrasound. Modulated ultrasound that demodulates in the air produces audible sound that can seem to come from any surface the beam strikes [S6]. Experiments on reflection show that a rigid wall preserves the beam’s directivity while a diffusive panel destroys it [S7], and other work notes that the beam is reflected and intercepted, which makes it hard to keep the sound in a single place [S8]. Separately, the laser system developed at MIT Lincoln Laboratory creates sound in the air from a laser tuned to water vapour, with sound described as audible only within a very small distance of the target [S9], although with a trade-off between fidelity and loudness [S10]. The fourth line is the digital-twin method behind the original Invisible Duet post, in which a full-wave simulation below 500 Hz is coupled with a geometric model above it, and designs are tested against imperfect drivers. ANEK Room applies that method to a working room rather than a demonstration room.

3. Benefits

For the person speaking, the main benefit is the ability to handle a call or a sensitive conversation in place, without equipment on the ears and without leaving the workspace. For people nearby, the benefit is less distraction, because overheard speech that is not intelligible is much less intrusive than speech that is understood. For organisations that run service counters, such as banks, clinics, pharmacies and government offices, the benefit is privacy for the person being served, who is often standing within earshot of a queue.

There is also a benefit for the design of the space. Today privacy in a room is bought with walls, booths and partitions, which fix the layout and take up area. A system that works largely through software, ceiling and desk hardware and carefully chosen surfaces allows a room to change its use without being rebuilt. These benefits are hypotheses at this stage and should be confirmed with listener tests in a real space.

4. Concept and structure

The system is organised in layers that can be installed one at a time. The first layer is a conventional masking system, with ceiling or wall emitters that provide a uniform background sound tuned to the room. This is mature, inexpensive and needs no tracking, and it is the baseline every later layer must beat. The second layer is focused personal audio. A small loudspeaker array near the desk or counter, or a ceiling array above it, delivers the far-end voice to the listener’s position and sends a masking signal toward the positions where bystanders are likely to stand or sit. The array filters are computed per frequency from the digital twin, which means the reflections of the room are part of the design rather than an uncontrolled disturbance.

The third layer is deliberate use of the surroundings. Walls, ceiling and furniture are treated as acoustic components: rigid, shaped panels where a reflection is wanted, diffusive panels where it must be broken up, and absorbers where it must be removed. A desk-side reflector can also act as a virtual source for a parametric ultrasound beam, so that audible sound appears to come from a surface close to the listener. Windows and glass partitions need particular care, because reflections from them can carry sound or ultrasound toward people outside the intended area.

The fourth layer is the uplink. A camera and a microphone array can find the speaker’s mouth and extract the voice from that direction, and similar camera-assisted systems already separate speakers in vehicles [S11, S12]. In a room this lets the person speak softly and still be heard clearly by the other party, which directly reduces what neighbours overhear. Because it uses a camera, it needs a clear consent and data-handling design, and the camera should be optional. The fifth layer is a controller with sensing for who is in the room and where, and with simple modes such as private call, shared conversation and quiet. The photoacoustic laser path is kept outside the first designs as a research track, because of eye-safety certification, the need to track the ear, and the unproven quality of speech at conversation level.

5. Feasibility

The acoustic part is feasible in principle, because focused personal audio and masked privacy have been demonstrated in research and the room’s reflections can be modelled with the hybrid approach described in the post. The difficulty is practical rather than theoretical. A room has more variation than a car: people move, furniture changes, doors open, and the low-frequency behaviour depends on room modes that must be modelled or measured. Bass cannot be confined to a small zone in any ordinary room, so the system should deliver and mask mainly the speech band, where focusing and masking are most effective.

The parametric ultrasound option is feasible only in fixed positions with controlled geometry, and it carries limits on bass and distortion and exposure limits for ultrasound that must be checked with the relevant standards. The laser option is not ready for a product. The optical uplink is the most mature of the add-ons, but it depends on lighting, camera placement and on acceptance of a camera in a private conversation. Cost, installation effort and regulatory requirements have not been assessed in this note and should be estimated with a systems partner before any commitment.

6. Novelty

Each element of the concept has prior art. Masking, focused personal audio, masking directed at eavesdroppers, parametric audio spots and camera-assisted speech separation are all known. In the searches carried out for this note, no source was found that combines a reflection-aware, digital-twin-designed room, directed masking, an optical uplink and an optional ultrasound or laser spot into one system aimed at private conversations. That is a statement about the searches made, which were limited, and not a finding of absence.

A defensible claim would therefore be narrow. It would concern the integrated system and its design method, including the deliberate placement of reflecting and absorbing surfaces using a validated twin and the testing of designs against hardware and occupant variation. A pure patent claim on any single element is unlikely to succeed. A patent and freedom-to-operate search should come before any public claim of originality.

7. Market potential in India

No market sizing has been prepared for this note, and none should be inferred from it. The natural entry points are places where a private conversation happens in a shared space and where people already complain about being overheard: bank and government service counters, hospital and clinic registration and pharmacy desks, co-working spaces and open offices, and shared rooms in homes. The strongest case appears to be the counter, because the person being served has no easy alternative and the staff member and visitor stand in a known position. The weakest case is the call-centre desk, where headsets are cheap and already the norm.

The competition is real. Sound masking is established in offices and has years of commercial experience behind it. A new system must show that it improves privacy beyond masking alone, at a price that an organisation will accept, and without adding discomfort or concern about cameras. Vendor surveys on speech privacy in offices exist, but they are promotional and are not used here to size demand.

8. Validation and decision gate

Before building anything large, the concept should be tested in a single real space, such as a service counter or a cluster of desks. Three set-ups should be compared with listeners placed at the positions where neighbours would sit. The first is masking plus focused personal audio only. The second adds a parametric reflected spot at the desk. The third adds the optical uplink so that the person can speak quietly. The measure should be how much of the call a listener can understand, judged both by listening tests and by a speech intelligibility index calculated from the measured fields, together with comfort, annoyance and the acceptability of cameras.

The decision rule is simple. The second or third set-up is worth pursuing only if it clearly lowers the masking level, or the leaked intelligibility, that is needed for privacy compared with the first, and does so at an acceptable cost. If it does not, the simpler masking-plus-focusing design, or ordinary earbuds, is the better answer. The same rule applies to any light-based element: it should be added only if it improves privacy enough to justify the safety work it requires.

Sources and caveats

S1  Wallace and Cheer, Design and Evaluation of Personal Audio Systems based on Speech Privacy Constraints, University of Southampton (eprints.soton.ac.uk/439056).

S2  Wallace and Cheer, Combining Background Noise and Artificial Masking to Achieve Privacy in Sound Zones, preprint 2021 (eprints.soton.ac.uk/451889).

S3  Sound Directions (UK), How a Sound Masking System Can Transform Your Workplace.

S4  FMLink, Solving Your Noise Problem (promotional article).

S5  Open-plan office study on water sounds as speech masking (core.ac.uk/download/294779148.pdf).

S6  Sound from ultrasound, overview article.

S7  Zhong, Wang, Kirby and Qiu, Reflection of audio sounds generated by a parametric array loudspeaker, JASA 148, 2020.

S8  Nakayama and Nishiura, spatial fusion of carrier and sideband waves with a parametric array loudspeaker, 2016.

S9  MIT Lincoln Laboratory, Targeted Acoustic Laser Communication (TALC).

S10 The Register, report on laser-beamed audio, January 2019.

S11 Mitsubishi Electric, speech signal separation using a camera and microphone array.

S12 Hi Auto, audio-visual speaker separation (press coverage, 2019).

 

This note is based on a limited literature and product search. Claims about benefits and the privacy value of each layer are hypotheses to be tested. Cost, safety limits for ultrasound and lasers, camera consent rules and patent status have not been verified, and no market size has been estimated.