ESP32's voice assistant

Project Overview:
This project is a voice assistant based on the ESP32 microcontroller. It features speech recognition and can answer simple questions.
Project Functionality:
This design is based on the ESP32-S3. It recognizes the input speech as text, connects to a big data model to obtain the answer, and then regenerates the text into speech for feedback.
Project Principle:
The main control chip used is the ESP32-S3-WROOM, which supports multiple communication protocols and can connect to Wi-Fi, Bluetooth, and Zigbee to meet different wireless communication needs.
The voice module uses the INMP441, which sends the PCM audio input from the i2s microphone to the SST speech recognition service to convert it into text. The obtained text answer is then converted back into speech via TTS for playback (the "speech-to-text-to-speech" process uses the iFlytek API).
The language model used is Volcano Engine (https://www.volcengine.com/product/ark)
, which is relatively easy to configure and suitable for beginners. WebSocket connections are used to process audio data, enabling streaming transmission and ensuring dialogue continuity.
The interactive components consist of two parts: a 1.8-inch touchscreen display and a MAX98357A audio power amplifier for voice playback and human-computer interaction.
The software
code references the large language model AI voice assistant from the Bilibili blogger "Bao Gai Chef Zhang Zuoren".
Important note:
Because the speech-to-text conversion and language model are from different websites, keyword fields in the code should be modified to ensure correct API connection.
The Wi-Fi name and password should be configured in advance in the code.
[Image of the actual product]