Overview

Product & platform overview

SATE Companion

A speech-capture and analysis platform for speech-language pathologists — from a handheld recorder and a wearable pendant, through an AI-powered pipeline, to an interactive clinician report.

Get started How it works

Handheld recorder Wearable pendant Mobile & web apps pipeline: async AI

SATE Companion captures a patient's speech on a hardware device, uploads the audio to a cloud backend that transcribes and analyses it with an AI model, and presents the result to the clinician as an interactive report they can review, edit, and annotate.

This site is a high-level overview of what the platform is, how its pieces fit together, and how each part behaves. It is written for a general reader — no source code or implementation internals required.

Start here

Getting startedThe one core idea, and the fastest reading path for your task. ArchitectureData-flow diagrams, transports, and why processing is async. Known issuesCurrent limitations and status.

Component guides

RecorderThe handheld touchscreen device — the flagship capture tool. PendantA small wearable that streams audio to the phone. Mobile appBridges the devices to the cloud and keeps them in sync. Web appClinician report: playback, transcript editing, annotations. Backend pipelineAsync AI: assemble, transcribe, analyse, store. Plaud integrationSupport for a third-party recorder as an audio source.

The system at a glance

Components

Component	Platform	Role	Guide
SATE Recorder	ESP32-S3 touchscreen handheld	Records to local storage, then uploads over Wi-Fi or via the phone; updates itself over the air	Recorder
SATE Pendant	nRF52840 wearable	Streams 16 kHz mono audio over Bluetooth to the mobile app	Pendant
Mobile app	iOS (React Native)	Bridges the devices to the cloud, integrates the Plaud recorder, and syncs automatically	Mobile app
Web app	Browser (React)	Clinician report: audio player, transcript editing, annotations, metrics	Web app
Backend	Cloud (serverless + container)	Async AI pipeline: assemble, transcribe, analyse, and store	Backend pipeline
Plaud	Third-party recorder	Optional recorder integration as an additional audio source	Plaud integration

Connection methods (transports)

The system deliberately uses a different transport per link — each chosen for the constraints of the device on either end.

Link	Transport	Why
Recorder → backend	Wi-Fi upload, with a Bluetooth bridge through the phone when offline	The recorder has its own Wi-Fi; the phone bridge is the fallback when there's no network
Pendant → app	Bluetooth audio stream	A tiny wearable with no Wi-Fi streams live to the phone
Plaud → app	Vendor Bluetooth SDK	The third-party recorder connects through its own SDK
App → backend	Secure upload over the internet	Standard authenticated upload from the phone
Backend AI call	Long-lived request from a dedicated worker	Transcription can run long, so it lives in a process without a short time limit
Firmware updates	Over-the-air pull from the cloud	Devices fetch updates themselves rather than being pushed to

See Architecture for the full data-flow, and each component guide for more detail on how that part works.