We have a chatbot built on Retool and powered by ChatGPT that uses audio for inputs and outputs.
We need to run this on a Raspberry Pi with a mobile cellular connection. The Raspberry Pi will need a
microphone and a button which can be pressed to receive the input and a speaker to play the output.

Example – https://app.omnie.ai/embedded/public/74d5e117-777b-4b03-9336-62a0ee1ce008

Project stages:
1. Hardware – Recommend a set of hardware components with consideration for power consumption.
The unit will be powered by a battery and solar panel and as such the power usage should be as low
as possible. Consider alternatives to Raspberry Pi if a lower power consumption can be achieved.
The components are Raspberry Pi (or similar), cellular module, speaker, microphone and illuminated
push button.
2. Operating System – Write a program on the Raspberry Pi and enable access to our Retool app
(https://app.omnie.ai/embedded/public/74d5e117-777b-4b03-9336-62a0ee1ce008) using a keyboard
and screen.
3. Audio – Add a microphone, speaker and push button to the Raspberry Pi to enable interaction with
the Retool app using voice. Add a volume control rotary potentiometer. The push button will illuminate
when activated and rermain illuminated until the unit has finished processing a conversation.
4. Cellular – Add a cellular module to the Raspbery Pi to enable remote use of the product. Investigate
the option to disconnect when not actively in use to save power.
5. Power Supply – The unit will be powered by a battery and connected solar panel. We will have a
controller which will monitor the battery voltage and shut the unit down when the voltage drops below
a predefined level to allow the solar panel to charge the battery. The unit will then power back up
when the voltage reaches a preset level.

The process should work like this:
1. Press and release the button
2. Bot speaks the welcome message via a speaker
3. Message ends and the microphone is enabled to listen for the response
4. User speaks their input which is recorded via the microphone
5. Bot waits for a 2 second silent period, disables the microphone, processes the input and
delivers it response via the speaker
6. The process continues until the end of the conversation.

30% deposit
Milestone 1: Working prototype demoed and tested – 40% payment
Milestone 2: Deployed to our own hardward. Tested and working correctly. Processes documented – 30% payment

Budget: $6,000

Posted On: January 29, 2024 15:03 UTC
Category: Scripting & Automation
Skills:Raspberry Pi, Raspberry Pi Firmware, ChatGPT API Integration

Country: United Kingdom

click to apply

Powered by WPeMatico