Component guide · Intermediate · 4 min read
ESP32 AI Voice and Audio Boards

Espressif ESP32-S3-BOX-3, ESP32-S3-Korvo-2 and ESP32-LyraT-Mini, ESP32-S3 Xiaozhi AI Voice Chat Boards, MINI ESP32-S3 Voice Board, AI-01 ESP32-C2 Voice Board, Waveshare AI Smart Speaker and Hiwonder ESP32-S3 Vision Module
By Geekobot Learning Team · Updated 29 Oct 2026
On this page
These ESP32 boards have microphones, a speaker amplifier and often a screen, for voice assistants and audio projects. ESP32-S3-BOX-3 is Espressif's complete kit with a touch screen and works with Home Assistant voice. Korvo-2 has a microphone array for far-field wake words. The Xiaozhi boards run open-source firmware that talks to cloud AI models. The Hiwonder module adds a camera for vision.
Key specs at a glance
- Chip
- ESP32-S3 with PSRAM on most; ESP32 on LyraT-Mini; ESP32-C2 on AI-01
- Audio in
- One or more MEMS microphones, ES7210 ADC on Korvo
- Audio out
- ES8311 codec and speaker amplifier
- Screen
- 2.4 inch touch on S3-BOX-3; small LCD on some voice boards
- Software
- ESP-SR (wake word, commands), ESP-ADF, ESPHome voice, Xiaozhi firmware
- Connectivity
- Wi-Fi and BLE
Models in this guide
These are alternatives. Pick the one that fits your project; you only need one.
Option 1AI-01 Intelligent Voice Development Board w/ ESP32-C2 & Mic/Speaker₹759
Option 2ESP32-S3-N16R8 AI Voice Chat Robot Learning Board (Unsoldered)₹549
Option 3ESP32-S3-N16R8 AI Voice Chat Robot Learning Board w/ Xiaozhi Firmware₹569
Option 4MINI ESP32-S3 AI Voice Development Board w/ Mic & Speaker₹879
Option 5Waveshare ESP32-S3 AI Smart Speaker Development Board₹2,519
Option 6Espressif ESP32-S3-BOX-3 AIoT Evaluation Kit₹6,819
What the ESP32 AI voice and audio boards (ESP32-S3-BOX-3, Korvo-2, LyraT-Mini, Xiaozhi voice boards, smart speaker) is
ESP32-S3 AI voice and audio boards such as the ESP32-S3-Korvo, ESP-BOX and LyraT-style boards add microphones, an audio codec and a speaker output for wake words and voice control.
For Home Assistant, flash ESPHome's voice assistant firmware to the S3-BOX-3 from the browser, then pick a wake word. For offline commands, ESP-SR recognises a wake word and up to 200 command phrases on the chip itself.
Xiaozhi firmware connects to a cloud AI service; check the server you choose and its privacy terms. Voice boards need a good 5V supply, because the speaker amplifier draws peaks.
ESP32 AI voice and audio boards (ESP32-S3-BOX-3, Korvo-2, LyraT-Mini, Xiaozhi voice boards, smart speaker) pinout
The pins on the ESP32 AI voice and audio boards (ESP32-S3-BOX-3, Korvo-2, LyraT-Mini, Xiaozhi voice boards, smart speaker) and what each one does.
| Pin | Function |
|---|---|
| USB-C | Power, flashing, serial |
| Microphones | On board |
| Speaker | JST connector or on-board speaker |
| Buttons | Boot, reset, mute or action |
| Expansion | PMOD or Grove on S3-BOX-3 dock |
- USB-C
- Power, flashing, serial
- Microphones
- On board
- Speaker
- JST connector or on-board speaker
- Buttons
- Boot, reset, mute or action
- Expansion
- PMOD or Grove on S3-BOX-3 dock
Wiring the ESP32 AI voice and audio boards (ESP32-S3-BOX-3, Korvo-2, LyraT-Mini, Xiaozhi voice boards, smart speaker)
Keep the speaker away from the microphones and set a sensible volume to avoid echo and clipping.
| ESP32 AI voice and audio boards (ESP32-S3-BOX-3, Korvo-2, LyraT-Mini, Xiaozhi voice boards, smart speaker) pin | Connection | Note |
|---|---|---|
| USB-C | 5V 2A supply | Weak supplies cause resets at high volume |
| Speaker | 4-8 ohm to the speaker connector | Within the amp rating |
| Sensors | Expansion header | 3.3V logic |
- USB-C
- Connection:
- 5V 2A supply
- Note:
- Weak supplies cause resets at high volume
- Speaker
- Connection:
- 4-8 ohm to the speaker connector
- Note:
- Within the amp rating
- Sensors
- Connection:
- Expansion header
- Note:
- 3.3V logic
Example code
The package path follows the ESPHome wake-word project; check the current path in ESPHome's documentation before flashing.
# Easiest: flash from https://esphome.io/projects/ in the browser.
# Manual config uses the maker's package:
substitutions:
name: s3-box-3
packages:
esphome.voice-assistant: github://esphome/wake-word-voice-assistants/esp32-s3-box-3/esp32-s3-box-3.yaml@main
esphome:
name: ${name}
name_add_mac_suffix: false
wifi:
ssid: !secret wifi_ssid
password: !secret wifi_passwordSetup and tuning
use ESP-IDF with ESP-ADF and ESP-SR for wake word and command recognition; the vendor demo is the best start.
I2S or PDM mics feed the codec; keep them away from the speaker and fan noise.
use the stated speaker impedance (often 4 or 8 Ω) and power from a good 5 V supply.
voice models need PSRAM; pick a module with 8 MB.
for chat assistants, stream audio to a cloud API over Wi-Fi.
What you need
Shop this list11 of 11 found in our store · ₹23,500
AI-01 Intelligent Voice Development Board w/ ESP32-C2 & Mic/Speaker: the part this guide coversAI-01 Intelligent Voice Development Board w/ ESP32-C2 & Mic/Speaker₹759
ESP32-S3-N16R8 AI Voice Chat Robot Learning Board (Unsoldered): another version this guide coversESP32-S3-N16R8 AI Voice Chat Robot Learning Board (Unsoldered)₹549
ESP32-S3-N16R8 AI Voice Chat Robot Learning Board w/ Xiaozhi Firmware: another version this guide coversESP32-S3-N16R8 AI Voice Chat Robot Learning Board w/ Xiaozhi Firmware₹569
Generic MINI ESP32-S3 AI Voice Development Board w/ Mic & Speaker: another version this guide coversMINI ESP32-S3 AI Voice Development Board w/ Mic & Speaker₹879
Waveshare ESP32-S3 AI Smart Speaker Development Board: another version this guide coversWaveshare ESP32-S3 AI Smart Speaker Development Board₹2,519
Espressif ESP32-S3-BOX-3 AIoT Evaluation Kit: another version this guide coversEspressif ESP32-S3-BOX-3 AIoT Evaluation Kit₹6,819
Espressif ESP32-S3-Korvo-2 Development Board: another version this guide coversEspressif ESP32-S3-Korvo-2 Development Board₹6,685
Espressif ESP32-LyraT-Mini V1.2 Audio Development Board: another version this guide coversEspressif ESP32-LyraT-Mini V1.2 Audio Development Board₹2,729
Hiwonder ESP32-S3 AI Vision Module: another version this guide coversHiwonder ESP32-S3 AI Vision Module₹1,700
USB-C cable: for power and flashingUSB-C Power Cable with On/Off Switch for Raspberry Pi 4₹163
MAX98357 I2S amplifier: for boards without a speaker amplifierMAX98357 I2S 3W Class D Amplifier Module₹129Closest match, check the specs
Alternatives and when to choose them
- ESP32-S3 DevKitC: plain S3 board.
- I2S microphones: build your own.
- MAX98357 amplifier: add a speaker.
Common mistakes to avoid
- Choosing a module without PSRAM for voice models.
- Putting the speaker right next to the microphones.
- Expecting full offline speech-to-text.
- Running the speaker from a weak USB port.
Frequently asked questions
Can the ESP32-S3 recognise a wake word offline?
Yes. ESP-SR runs wake words like Hi ESP and a list of commands on the chip.
Does it understand full sentences?
Not offline. Full speech needs a cloud speech or AI service.
Why does voice recognition miss commands?
Noise, echo or a far mic. Move closer and keep the speaker away from mics.
Can I use Arduino IDE for these boards?
Partly, but the voice libraries are best supported in ESP-IDF.
Why does audio crackle?
A weak supply or wrong I2S settings. Check sample rate and power.
Parts for this guide
Keep learning
Sources
Written by the Geekobot team from manufacturer datasheets. Always check the datasheet for the exact part and variant you buy.

