HiViCon Voice Assistant – Waveshare ESP32-S3 AI Smart Speaker Development Board for Home Assistant Assist

Local voice satellite for Home Assistant Assist based on the Waveshare ESP32-S3-AUDIO-Board.

This repository documents a working configuration for the board variant with:

• ESP32-S3R8
• ES7210 microphone ADC
• ES8311 speaker/DAC codec
• TCA9555 GPIO expander
• WS2812 RGB LED ring
• Home Assistant Assist
• local wake word detection via micro_wake_word
• local STT via Wyoming Whisper
• optional local TTS via Wyoming Piper

The key finding: the microphone path must use the ES7210 ADC. A plain i2s_audio microphone plus ES8311-only setup can stream data but does not provide usable speech recognition on this board.

Status

Working:

• Wake word detection: Okay Nabu
• ES7210 microphone ADC
• ES8311 speaker output
• I2S pins verified
• Home Assistant Assist pipeline
• STT recognition
• local response playback
• RGB status ring
• onboard keys via TCA9555

Example confirmed recognition:

Speech recognised as: " Schalte Lichtküche ein."
Response: "Lichtküche existiert nicht"

The response was semantically correct: Home Assistant understood the command but did not find an entity with that alias.

Hardware

Tested board:

Waveshare ESP32-S3-AUDIO-Board
ESP32-S3R8
ES7210 microphone ADC
ES8311 audio codec
TCA9555 GPIO expander

Verified important pins:

Function	Pin / Bus
I2C SDA	GPIO11
I2C SCL	GPIO10
I2S MCLK	GPIO12
I2S BCLK	GPIO13
I2S LRCLK	GPIO14
Microphone data input	GPIO15
Speaker data output	GPIO16
RGB LED ring	GPIO38
Amplifier control	TCA9555 pin 8
Key1	TCA9555 pin 9
Key2	TCA9555 pin 10
Key3	TCA9555 pin 11

Detected I2C devices:

Address	Likely component
0x18	ES8311
0x20	TCA9555
0x40	ES7210
0x51	RTC / board peripheral

Repository structure

.
├── esphome/
│   └── hivicon-voice-wohnzimmer.yaml
├── docker/
│   ├── docker-compose.yml
│   ├── wyoming-whisper.sh
│   └── wyoming-piper.sh
├── docs/
│   ├── setup.md
│   ├── home-assistant.md
│   ├── troubleshooting.md
│   ├── hardware-notes.md
│   └── known-bad-configs.md
└── logs/

Quick start

1. Copy esphome/hivicon-voice-wohnzimmer.yaml into your ESPHome config folder.
2. Add your Wi-Fi and API secrets in ESPHome secrets.
3. Compile and flash the board.
4. Install Wyoming Whisper for local STT.
5. Optional: install Wyoming Piper for local TTS.
6. Add the ESPHome device in Home Assistant.
7. Configure Assist aliases for the entities you want to control.

ESPHome secrets

Example:

wifi_ssid: "YOUR_WIFI"
wifi_password: "YOUR_WIFI_PASSWORD"
api_encryption_key: "YOUR_ESPHOME_API_KEY"
ota_password: "YOUR_OTA_PASSWORD"

Required Home Assistant components

• ESPHome integration
• Assist pipeline
• Wyoming Protocol integration
• Wyoming Whisper for speech-to-text
• Optional Wyoming Piper for text-to-speech

Local Whisper Docker

docker run -d \
  --name wyoming-whisper \
  --restart unless-stopped \
  --network host \
  rhasspy/wyoming-whisper \
  --model tiny \
  --language de

Then in Home Assistant:

Settings → Devices & services → Add integration → Wyoming Protocol
Host: <docker-host-ip>
Port: 10300

Local Piper Docker

docker run -d \
  --name wyoming-piper \
  --restart unless-stopped \
  -p 10200:10200 \
  -v /volume1/docker/piper:/data \
  rhasspy/wyoming-piper \
  --voice de_DE-thorsten-high

Then in Home Assistant:

Settings → Devices & services → Add integration → Wyoming Protocol
Host: <docker-host-ip>
Port: 10200

Important lesson learned

The ES8311 is not the microphone ADC on this board. It is the DAC/speaker codec.

The working setup requires:

audio_adc:
  - platform: es7210

Without ES7210 initialization, wake/STT may appear partially alive, but Whisper receives unusable audio.

License

MIT License. See LICENSE.