Network-controlled audio announcement system — ESP32-P4 + Ethernet + REST API

An industrial audio player that receives audio files over Ethernet and plays them through a speaker/megaphone system. Designed for factories, warehouses, public transport stations, and any environment where network-controlled announcements are needed.

A remote server (or app) sends pre-generated audio to the device via REST API. The device caches tracks on flash storage, so repeated announcements play instantly without re-uploading.

What it does:

• Receives audio files over Ethernet via HTTP REST API
• Caches tracks on internal flash (LittleFS) with hash-based versioning
• Plays 44.1 kHz 16-bit stereo PCM audio through ES8311 codec + power amplifier
• Supports DHCP or static IP configuration
• Status LED indicates client connectivity
• Configurable volume, port, and network settings via menuconfig

┌─────────────────┐    Ethernet (RJ-45)    ┌───────────────────────┐
│  Control Server  │ ─────────────────────► │   ESP32-P4 Device     │
│                  │                        │                       │
│  POST /update    │   Upload audio file    │  HTTP Server (:8080)  │
│  POST /play      │   Play cached track    │  LittleFS Storage     │
│  POST /check     │   Check if cached      │  I2S + ES8311 Codec   │
│  GET  /health    │   Health check         │  Power Amplifier      │
└─────────────────┘                        └───────────┬───────────┘
                                                       │
                                                       ▼
                                               ┌───────────────┐
                                               │   Speaker /   │
                                               │   Megaphone   │
                                               └───────────────┘

ESP32-P4 Pin Assignment
─────────────────────────────────────
I2C:     SDA = GPIO 7,  SCL = GPIO 8
I2S:     MCLK = GPIO 13, BCLK = GPIO 12
         WS = GPIO 10, DOUT = GPIO 9, DIN = GPIO 11
PA:      Enable = GPIO 53
Ethernet: MDC = GPIO 31, MDIO = GPIO 52, RST = GPIO 51
LED:     Status = GPIO 2

Upload a new audio track. The device stores it on flash with a hash for versioning.

curl -X POST http://192.168.1.100:8080/update-audio \
  -H "Content-Type: application/octet-stream" \
  -H "X-Message-Text: fire_alarm" \
  -H "X-Audio-Hash: abc123" \
  --data-binary @alarm.pcm

Play a previously uploaded track. Returns 404 if not cached.

curl -X POST http://192.168.1.100:8080/play-message \
  -H "Content-Type: application/json" \
  -d '{"message_text": "fire_alarm", "audio_hash": "abc123"}'

Check if a track exists with matching hash before uploading.

curl -X POST http://192.168.1.100:8080/check-audio \
  -H "Content-Type: application/json" \
  -d '{"message_text": "fire_alarm", "audio_hash": "abc123"}'
# Response: {"exists": true}

Industrial-Megaphone-ESP32/
├── main/
│   ├── main.c                # Entry point, init sequence
│   ├── audio_player.c/h      # I2S + ES8311 codec driver
│   ├── audio_storage.c/h     # LittleFS track caching
│   ├── ethernet.c/h          # Ethernet PHY + IP config
│   ├── http_server.c/h       # REST API endpoints
│   ├── CMakeLists.txt         # Build config
│   ├── Kconfig.projbuild      # menuconfig options
│   └── idf_component.yml      # Dependencies
│
├── screenshots/
└── README.md

All settings are configurable via idf.py menuconfig → Audio Server Configuration:

cd Industrial-Megaphone-ESP32

idf.py set-target esp32p4
idf.py menuconfig        # optional
idf.py build flash monitor

Plug in Ethernet cable. The device gets an IP via DHCP (or uses the configured static IP). Check Serial output for the assigned address.

# Health check
curl http://192.168.1.100:8080/health

# Upload audio
curl -X POST http://192.168.1.100:8080/update-audio \
  -H "X-Message-Text: test_announcement" \
  -H "X-Audio-Hash: v1" \
  --data-binary @test.pcm

# Play it
curl -X POST http://192.168.1.100:8080/play-message \
  -H "Content-Type: application/json" \
  -d '{"message_text": "test_announcement", "audio_hash": "v1"}'

Ethernet over WiFi — Industrial environments need reliable, low-latency connectivity. Ethernet eliminates wireless interference issues common in factories.

Hash-based caching — The server sends a hash with each track. The device only downloads new audio when the hash changes, saving bandwidth and time.

PSRAM-first allocation — Audio buffers are allocated in PSRAM when available, falling back to internal RAM. This allows playback of large audio files.

Mutex-protected playback — The audio player uses a FreeRTOS mutex to prevent concurrent playback requests from crashing the I2S driver.

Temur Eshmurodov — @myseringan

MIT License — free to use and modify.