# Voice Note Transcribe > Drop in a WhatsApp voice note (.opus / .ogg / .m4a) and get back a transcript, a > structured summary and an infographic in one pass. Everything runs in the > browser with the user's own OpenRouter key — there is no backend, nothing is > uploaded to us, and nothing is tracked or stored server-side. Free beta; > a typical pass costs the key holder well under £0.01, with a declared maximum > shown before each run. Built in the open by an agentic team; every claim on > this site links to its source in the public repository. The app is a JavaScript tool published as `window.__tool` (an SgToolApi named `whatsapp-transcribe`). If you are an agent operating the app or explaining it, the machine-readable surfaces below are maintained for you and are always in step with the deployed build. ## The app, for agents - [API skill](https://whatsapp-voice-transcription.sgraph.ai/app/skills/SKILL__api.md): every `window.__tool` action with params and returns — run a pass, read results, drive the chat, read the workflow and its execution trace - [Browser-use skill](https://whatsapp-voice-transcription.sgraph.ai/app/skills/SKILL__browser.md): operating the app through its UI - [Human guide](https://whatsapp-voice-transcription.sgraph.ai/app/skills/SKILL__human.md): the app explained for people - [Action manifest](https://whatsapp-voice-transcription.sgraph.ai/app/manifest.json): the full action list, machine-readable - [The declared workflow](https://whatsapp-voice-transcription.sgraph.ai/app/workflows/standard.json): the one-pass as a JSON state machine — steps, models, per-step spending ceilings, transitions, failure behaviour. The app executes from this file ## Live state, machine-readable - [status.json](https://whatsapp-voice-transcription.sgraph.ai/engineering/status.json): the deployed version, commit, and the per-layer test results this deploy passed - [issues.json](https://whatsapp-voice-transcription.sgraph.ai/engineering/issues.json): the full issue queue (open/blocked/done) as of this deploy - [docs.json](https://whatsapp-voice-transcription.sgraph.ai/engineering/docs.json): the public documentation inventory - [versions.json](https://whatsapp-voice-transcription.sgraph.ai/versions/versions.json): every release with date, headline and changes - [updates.json](https://whatsapp-voice-transcription.sgraph.ai/updates/updates.json): every Updates post, structured - [videos.json](https://whatsapp-voice-transcription.sgraph.ai/videos/videos.json): the demo videos - [RSS feed](https://whatsapp-voice-transcription.sgraph.ai/updates/feed.xml) - [guide.json](https://whatsapp-voice-transcription.sgraph.ai/user-guide/guide.json): the user guide as data — each step with its test-captured screenshot - [baseline-changes.json](https://whatsapp-voice-transcription.sgraph.ai/user-guide/baseline-changes.json): every time the product's screenshots moved, tied to the commit and CI run that moved them (the agent-review surface for UI change) ## Pages - [How it works + pricing](https://whatsapp-voice-transcription.sgraph.ai/): the landing page — today's beta is bring-your-own-key; the credit product is labelled PROPOSED - [The app](https://whatsapp-voice-transcription.sgraph.ai/app/) - [User guide](https://whatsapp-voice-transcription.sgraph.ai/user-guide/): step-by-step usage — every screenshot taken by a passing test on the same build - [Engineering](https://whatsapp-voice-transcription.sgraph.ai/engineering/): how it is built — pipeline, testing, docs, security posture (including open findings, honestly labelled), team, concepts - [Concepts](https://whatsapp-voice-transcription.sgraph.ai/engineering/concepts/): the concept scheme behind the interface vocabulary — meaning stored in a definition, one label per culture, and the English review that found three words naming the wrong thing. NOT wired to the shipped strings; it is a review surface - [concepts.json](https://whatsapp-voice-transcription.sgraph.ai/engineering/concepts/concepts.json): that scheme as data — definitions, scope notes, per-culture labels, verdicts and broader/narrower/related relations - [Design candidates](https://whatsapp-voice-transcription.sgraph.ai/design/): the A/B hub — three theme prototypes and four culture packs, each runnable with synthetic data and no key. Design references under evaluation, not the shipping product (noindex, absent from the sitemap) - [Browser unit tests](https://whatsapp-voice-transcription.sgraph.ai/tests/browser/): the app's unit suite, runnable in your browser against the deployed modules - [Updates](https://whatsapp-voice-transcription.sgraph.ai/updates/): the dated build-in-public record - [Versions](https://whatsapp-voice-transcription.sgraph.ai/versions/): every release, with diffs - [Videos](https://whatsapp-voice-transcription.sgraph.ai/videos/): demos and shorts - [Library](https://whatsapp-voice-transcription.sgraph.ai/library/): every document behind the product ## Source - [Repository](https://github.com/sgraph-ai/SGraph-AI__SaaS__WhatsApp__Audio__Transcription): everything — code, briefs, decisions, reviews, this file - [Reality doc](https://github.com/sgraph-ai/SGraph-AI__SaaS__WhatsApp__Audio__Transcription/blob/dev/team/roles/librarian/reality/index.md): what exists today vs what is proposed — the honesty contract this site is written under