FindASet is an offline-first Android training & nutrition companion. It builds and runs your workouts with a neural voice announcer, counts reps, tracks weight-per-set and PRs, logs your macros and body-weight, and reads nutrition labels with an on-device scanner — everything lives on your phone. No account, no cloud, no ads, no tracking; the one feature that can reach the internet is an optional, off-by-default barcode lookup you switch on yourself.
Closed-source, proprietary application. Free to use (including commercially) — not for sale, not redistributable, not modifiable, no reverse-engineering. See LICENSE and DISCLAIMER.
💸 Every rep you logged here was paid for with someone else's unpaid evenings, and your contribution to that total is a personal best in the wrong direction. PayPal: paypal.me/FAMarco Bitcoin:
bc1qv92c3eyeqvhgfnez7spfd7v2aytkhpshsl65yv
Author / copyright: © 2026 Marco Aurelio Fattizzo (@eVersor-HN). This is the official distribution repository — get FindASet only from here: https://github.com/eVersor-HN/FindASet
- Open the Releases page and download
the latest
FindASet-<version>.apk. - Verify it is the genuine original (see below) before installing it.
- On your phone, open the APK and allow installing from this source if prompted ("Install unknown apps"). Follow the installer. FindASet is sideloaded — it is not on the Play Store.
There is nothing else to install — the app ships with everything it needs (voice models, food database, label-scanner OCR) and downloads nothing on its own. The only network request it can ever make is the optional barcode lookup, and only after you turn it on.
Every official release publishes the SHA-256 checksum of the APK. Comparing the checksum of your download against the published value proves the file is the unmodified original and was not tampered with. (The same repository address and this verification hint are shown inside the app under Settings → About.)
v1.13 — FindASet-1.13.apk:
19bf95c31dfe07e1fcf22f49eb16daff4425cb440da953b6c24b93dda1142453
The authoritative value for each release is in that release's notes and in its
SHA256SUMS.txt asset.
Check it:
# Windows (PowerShell)
Get-FileHash .\FindASet-1.13.apk -Algorithm SHA256# macOS # Linux
shasum -a 256 FindASet-1.13.apk sha256sum FindASet-1.13.apkThe printed hash must match the value above (case-insensitive). If it does not match, do
not install the file — it is not the genuine FindASet build. The APK is signed with the
author's release key (CN=Marco Aurelio Fattizzo, O=FindASet, C=DE); Android will refuse any
later update that is not signed with the same key.
- Boot terminal — the app opens on a status report rather than a menu: a self-test, the day's subsystem levels, what is scheduled and a direct start. It reads today's plan, so a rest day, an easy day and a hard day each get their own readout.
- Guided workout builder — a simple, step-by-step wizard walks you from a name through your exercise list to timing and schedule, with the estimated total workout duration shown as you build.
- Guided workout player — timeline, circular timer, voice cues, spoken rest countdowns and periodic time call-outs.
- Rep & set logging — optional auto-tempo and count-aloud, weight per set, estimated 1RM tracking, PR detection announced by the voice.
- Plans & calendar — weekly training plans, month/week logbook, heatmap, streaks, per-day notes. Call a single day off from the dashboard and its macro targets, reminder and readouts switch to rest without touching the plan; press again to put it back.
- Macro tracking — quick-log dropdown, food library with favourites, fuel-tank goals, a multilingual food database (15 languages) and a FILL MACROS calculator that solves food amounts to hit the day's remaining macros.
- Multilingual label scanner — on-device OCR reads kcal / protein / carbs / fat from a photographed nutrition label across many languages, tells calories from kilojoules, and runs an energy cross-check that repairs gaps. Optionally switch on a barcode lookup to fetch exact values online when a label is unreadable. (Always verify scanned values.)
- Neural offline voice — an on-device announcer (female & male) with selectable styles, personas and effects. Rendered entirely on-device.
- Body-weight tracking — daily log with 0.1 kg steppers, goal line and trend.
- Home-screen widget — today's calories and protein / carbs / fat at a glance; tap to open.
- Colour-vision support — deuteranopia, protanopia, tritanopia and monochrome modes. The alarm colours move to hues that survive the deficiency, statuses gain text markers so meaning never rests on colour alone, and the accent picker shows each swatch as you actually see it.
- Portrait and landscape — every screen lays itself out for the orientation it is in; the player, calendar and label scanner switch to side-by-side arrangements rather than squeezing.
- No account, no cloud, no ads, no tracking — your data stays on the device, with local backup export/import built in. The one network feature — the optional barcode lookup — is off until you enable it, and even then only a scanned barcode number is sent.
- Android 8.0 (API 26) or newer, arm64-v8a device.
- ~95 MB free space for the APK (voice models, food database and scanner models included).
FindASet is proprietary, closed-source software under the FindASet EULA
(full text in LICENSE):
- ✅ Use it for any purpose, including commercially (companies may use it internally).
- ✅ It is free of charge. To share it, point people to this repository.
- ❌ No redistributing or passing on copies, no selling, no modifying / adapting, no reverse-engineering, decompiling or disassembling — except where a bundled third-party license or mandatory law (e.g. § 69e UrhG) provides otherwise.
It bundles third-party components under their own licenses — AndroidX / Jetpack Compose, Kotlin,
kotlinx.serialization, CameraX (Apache-2.0), ML Kit Text Recognition & Barcode Scanning (ML Kit
ToS, on-device models), the neural-TTS stack sherpa-onnx (Apache-2.0) / ONNX Runtime (MIT) / Piper &
piper-phonemize (MIT) / eSpeak NG (GPL-3.0, remains under its own license), the Piper
voice models kristin & john (public-domain LibriVox recordings) and food data derived from
USDA FoodData Central (public domain). Full details and license texts:
THIRD_PARTY_NOTICES.md.
No warranty — FindASet is a personal bookkeeping tool, not medical, dietary, or fitness
advice; figures (including scanned label values) are estimates. Exercise carries risk — train
within your limits. See DISCLAIMER.md.