This article describes a step-by-step installation procedure for “AI Stackchan-2”, a voice dictation AI system that utilizes Google Speech-to-Text to recognize and replay using OpenAI’s cloud service and generates real voice using VOICEVOX. Since the original GitHub documentation is in Japanese, I aim to provide an English version with detailed information to setup.
0. System Diagram
1. Download “AI Stackchan-2 PLUS” through M5Burner
Install M5Burn into your PC then download and burn it to your M5stack Core2.
2. Connecting M5stack to WiFi
Prepare a Micro SDcard and initialize it with FAT32 format, then make a “wifi.txt” file which includes the following SSID/PW text pair.
3. Got OpenAI API key
Registered your ID on OpenAI.
Go to your account menu and chose “View API keys”, then you should create a new secret key.
4. Got the VOICEVOX API key
5. Got Google-STT API key
Registered your ID on GCP and create a new project first, then enable Speach-to-Text API.
Go to API then select Credentials.
Select Create Credential on the top menu and API key.
6. Connecting M5stack to those Services
Make an “apikey.txt” file on the SDcard that includes the above API keys.
7. How it works