Bravo AAC

An open-source Augmentative and Alternative Communication (AAC) application powered by Google Gemini and deployed on Google Cloud Platform.

Bravo AAC helps individuals with communication differences express themselves using AI-generated word suggestions, pictogram symbol boards, and customizable communication pages. It supports scan-based access, multiple AAC board formats (Grid, Tap, Freestyle/Compose), and import from common AAC formats (Accent MTI, TouchChat).

Features

• AI-powered word prediction — Context-aware suggestions using Google Gemini
• Multiple board types — Grid, Tap interface, and Freestyle/Compose modes
• Scan access — Full switch/scan support across all interfaces
• AAC format import — Import from Accent MTI and TouchChat formats
• Custom image support — Upload and manage per-user pictogram images
• Multi-environment — Development, testing, and production environments on GCP
• Firebase Auth — User authentication and per-account data isolation
• Firestore — Real-time database for user profiles, boards, and content

Tech Stack

Layer	Technology
Backend	Python 3.12, FastAPI
LLM	Google Gemini (via google-generativeai)
Database	Firestore
Auth	Firebase Authentication
Storage	Google Cloud Storage
Deployment	Cloud Run (Docker), Cloud Build CI/CD
TTS	Google Cloud Text-to-Speech

Getting Started

Prerequisites

• Python 3.12+
• Docker (for local Cloud Run simulation)
• A GCP project with the following APIs enabled:
  • Cloud Run, Cloud Build, Firestore, Firebase Auth, Cloud Storage, Text-to-Speech, Gemini API
• A Firebase project linked to your GCP project
• A Google Gemini API key

Local Development

1. Clone the repo

   git clone https://github.com/OSUBlakester/BravoGCPCopilot.git
   cd BravoGCPCopilot
   

2. Create a virtual environment

   python3 -m venv venv
   source venv/bin/activate
   pip install -r requirements-local.txt
   

3. Configure credentials

   cp config.py.template config.py
   # Edit config.py with your Firebase project credentials
   

   Or create a .env file from the template:

   cp .env.template .env
   # Edit .env with your environment variables
   

4. Start the server

   ./restart-local.sh
   

   The app will be available at http://localhost:8000.

Cloud Run Deployment

See DEPLOYMENT.md for full multi-environment deployment instructions.

Automated deployments are configured via cloudbuild.yaml — merging to main triggers a Cloud Build pipeline that builds and deploys to Cloud Run.

Required Environment Variables

These must be set in your Cloud Run service (or .env for local dev):

Variable	Required	Description
ENVIRONMENT	Yes	development, testing, or production
GCP_PROJECT_ID	Yes	Your GCP project ID
GEMINI_PRIMARY_MODEL	Yes	Primary Gemini model (e.g. gemini-2.5-flash-lite)
GEMINI_FALLBACK_MODEL	Yes	Fallback Gemini model (e.g. gemini-1.5-flash-latest)
GEMINI_FAST_WORDS_MODEL	No	Defaults to GEMINI_PRIMARY_MODEL
FIREBASE_API_KEY_*	Yes	Firebase API key for your environment suffix
FIREBASE_AUTH_DOMAIN_*	Yes	Firebase auth domain
FIREBASE_PROJECT_ID_*	Yes	Firebase project ID
FIREBASE_STORAGE_BUCKET_*	Yes	Firebase storage bucket
GOOGLE_API_KEY	Yes	Gemini API key (via Secret Manager recommended)
SERVICE_ACCOUNT_KEY_PATH	Yes	Path to GCP service account JSON
EMAIL_FEATURE_ENABLED	No	Set true to enable Gmail integration
GOOGLE_OAUTH_CLIENT_ID	If email enabled	OAuth client ID
GOOGLE_OAUTH_CLIENT_SECRET	If email enabled	OAuth client secret
EMAIL_OAUTH_STATE_SECRET	If email enabled	Random secret for OAuth state
EMAIL_TOKEN_ENCRYPTION_KEY	If email enabled	Fernet key for token encryption

See ENVIRONMENT_VARIABLES.md for full details and GCP Console/CLI setup instructions.

AAC Format Import

Bravo supports importing communication boards from:

• Accent/NovaCHAT MTI files — via accent_mti_parser.py and accent_bravo_mapper.py
• TouchChat .ce/.c4v files — via touchchat_ce_parser.py

See MTI_MIGRATION_PROCESS.md for the migration workflow.

Contributing

See CONTRIBUTING.md for branching workflow and contribution guidelines.

License

Apache License 2.0 — see LICENSE for details.