Home
To see the progress development check out its GitHub Project.
This project is composed of various pieces:
| Module | Purpose |
|---|---|
| Sound Detector | Discriminate and pick up specific sounds using AI. |
| Playback Distributor | Route the detected sound and bounce it back from multiple speakers on multiple locations. |
| Sound Player | Play any given sound when told by the distributor. |
| Journal Web App | Visualize the sound occurrences. Useful to have an objective overview. If case gets to lawyers it might be useful. |
This is a diagram of how they intercomunicate:
If you are only interested in the provisioning and deployment of the architecture:
| Document | Explains How To... |
|---|---|
| Provisioning & Deployment | Flash a new card, plug into a Raspberry Pi and provision it as master or slave and deploy the project |
| Raspberry Pi Sound Setup | Connect multiple Bluetooth speakers, create a combined sink of various speakers, set up microphone. |
Further information on how to get a development environment and how the issues of running container sound on the host (be it a MacBook or a Raspberry Pi):
| Document | Explains How To... |
|---|---|
| Docker Container Sound | Offer host audio I/O to the container through PulseAudio's TCP interface. |
| Development Workflow | Develop the application container on a non-Pi host with keeping the LSP IDE features and still be able to use the microphone and speakers through PulseAudio. |
Other documentation:
| Document | Explains How To... |
|---|---|
| Stack Recipes | Run hello worlds, proof of concepts and basic examples with the stack tools: InfluxDB, NFS and Ansible. |
| AI: Transfer Learning | Repurpose an already trained classification neural network to classify specific sounds. |
| Modes of Operation | Explains the various ways in which the inference (prediction) can be run. |