24 AUG 2026
Tingvar on Android
The useful time to find out a plan is wrong is on the walk before the meeting, not at the desk after it. This page says what an Android client would add over the browser, and it says plainly that zero Android devices have been tested and nothing has been lodged with Google Play.
What exists today
In development 24 Aug 2026
No date yet. The client is being built and no Google Play submission has been lodged.
Store listings
Google Play PENDING No Google Play listing exists yet, so there is no badge here and nothing to install.
In the browser you already have.
Be told when it ships
One email, on the day this surface ships, and nothing else. Off unless you tick it, and one click to stop.
19 AUG 2026
What there is to show, and what is measured
What there is to show, and what there is not
Not yet produced
The two pictures a store listing shows before you install. A design render is acceptable on our own site when it is labelled. It is not acceptable in a store listing, where it is a policy breach in most stores, so these wait on the real application either way.
Measured
Voice to voice
1471 ms
Median, on the bench in us-central1, with a real media server
MEASURED 2026-08-19, 122 readings
Real human sessions
0 sessions
Have taken place. Every measured turn so far used pre-rendered audio
MEASURED 2026-08-23, 1 reading
There is no screenshot on this page and no device frame with a render in it. A design render presented as an app screenshot is a lie to a reader and a policy breach in the store it would be submitted to. The measured figures above are what exists to show, and both sit on the evidence page with the method and the sample size.
Why a phone, specifically
A plan gets written at a desk and defended somewhere else. The twenty minutes between the two is when the objections you have not thought of are cheapest to hear, and it is time nobody spends at a keyboard.
A phone is worse than a laptop at everything that happens after a session, and on Android it has one problem the browser does not: the operating system is entitled to stop a long running audio session to save power. That row is in the table below rather than left out of it.
What it would do here
| Capability | Browser | Android |
|---|---|---|
| Joins a room and holds a voice call | Yes | Yes |
| Each persona in a fixed position in the stereo field | Web Audio API | The platform audio engine |
| Mono fallback for a single earbud | Yes | Yes |
| Reading a long transcript afterwards | Better | Worse. It is a small column |
| Holding a session while you walk | Worse | Better |
| Survives the battery optimiser | Not applicable | Not decided |
| Anything to install | No | Yes |
Read about the browser surface
Every cell in that table is a design decision rather than a measurement, because the client is not built. Intended behaviour, recorded 24 Aug 2026. The rows that say Not decided are the ones a store visitor is most likely to be caught by, so they are named rather than omitted.
What you need
| Requirement | What it means |
|---|---|
| Headphones | Strongly recommended and close to a requirement. A phone speaker feeding back into a phone microphone is the worst case of the feedback problem, not the mildest. |
| Microphone permission | The app asks for the microphone so the room can hear you; your voice is used live and no recording is kept. |
| An exemption from battery optimisation | Probably. A long running audio session is exactly what an aggressive optimiser stops, and asking you for that exemption is a thing we would have to explain rather than slip past you. |
| A connection that holds a call | A live room is a voice call. It costs roughly what a voice call costs, and a weak connection costs you audio quality first. |
The app asks at the point you join a room, not at first launch. Refusing leaves you able to listen and not to interrupt, which is half the product.
What the measured figure does not cover
The latency figure on this page was measured server side, against Vertex AI in us-central1 and a real SFU. It was not measured on an Android device. The device side audio leg is NOT MEASURED, and it varies with the manufacturer, the chipset, the Android version and the Bluetooth profile in use. Zero Android devices have been tested.
This block sits at the same size and the same contrast as the capability table above it, because a qualification set in smaller grey type is a qualification the page is hiding. The rule is in the stylesheet rather than in this page, so it is not a matter of taste.
Questions to answer before you install
Will the battery optimiser kill the session?
It is the risk this surface has and the browser does not, and which handsets it bites on is NOT MEASURED. Vendor battery optimisers stop long running background work, and a 45 minute audio session is that shape. We do not yet know which devices need an exemption.
What does a session do to the battery?
NOT MEASURED. A 45 minute room is a continuous two way voice call, which is expensive on any phone, and we would rather publish a figure taken off real handsets than guess at one now. It appears here with its method and its date once the client runs on hardware.
How much mobile data does a session use?
NOT MEASURED. A room is a voice call, so the order of magnitude is a voice call, and the exact figure depends on the codec the media server settles on. The number is published here with the date it was taken.
What happens if a phone call arrives in the middle of a session?
Not decided, and that is a decision we owe you rather than a measurement we have not taken. Android hands the audio focus to the incoming call, so the room loses the microphone whether we like it or not. What the session does on the way back in is written here before the app is submitted rather than discovered by you.
Will Bluetooth headphones add enough delay to make interrupting awkward?
Very likely, and how much is NOT MEASURED. Bluetooth adds its own latency on top of the voice to voice figure on this page, which was taken on a wired path and not on a phone. Interrupting is the part of the product most sensitive to it.
Does a single earbud get a usable mix?
The intent is a mono fold down that keeps every persona audible, and whether it works on real earbuds is NOT MEASURED. The stereo field is how you tell the personas apart, so folding it down costs you something, and how much is a thing we have to hear on hardware.
What happens if I leave the app in the middle of a session?
NOT MEASURED. The room is meant to keep running with the screen off, which needs a foreground service and a notification you can see, and whether it survives on any given handset is exactly what has not been tested.
Where the recording goes
Tingvar does not store your session audio: it is processed while the room is live and no recording of it is kept. What persists is the transcript, the text of what you said and what each persona said, attributed line by line. Only the account that ran the session can read that transcript, and nobody at Tingvar reads one without your written request on a support case. You can delete any session yourself, and deletions are purged from backups within 30 days. The retention schedule is published at tingvar.com/trust/your-data.
Where a subscription is cancelled
The app takes no payment. A subscription is bought on the website and cancelled from your account settings, in three clicks from your account. The refund terms carry the window and what happens when a room drops part way through.
Deleting your account
You delete a Tingvar session yourself from the session list, and deleting your account destroys every session in it. Deletion is self-service and does not go through support. Anything you delete is purged from backups within 30 days, and we publish that number because an unbounded backup window is what fails a security questionnaire. Billing records survive deletion for seven years, because tax law requires it, and we say so rather than implying everything vanishes.
The control is reachable without signing in first, which is what the Play policy asks for and what we would do anyway. It is at the account deletion page.
Where to go from here
IDLE
Nobody has used this yet
Zero real human sessions have taken place. Every measured turn to date used pre-rendered audio clips rather than a live human. What is built and verified is the decision layer: the turn-taking policy, the bid scoring and the configuration guards, with 619 automated tests passing and 81.78 % combined coverage. Everything there says the system works. Nothing there says anyone wants it, and being first is a real risk you are taking.