ESP-NOW Audio Source

Wireless audio transmitter built on the ESP32. Captures stereo audio from a PCM1808 ADC over I2S, compresses it with IMA ADPCM, and broadcasts it over ESP-NOW. Designed to pair with the ESP-NOW Audio Sink receiver.

---

Architecture

---

Overview

This is one half of a wireless audio link. The source board reads 48 kHz / 24-bit stereo PCM from a PCM1808, encodes each frame into IMA ADPCM (roughly 12:1 compression), and sends it as a broadcast ESP-NOW packet. No pairing or handshake is needed -- any sink on the same Wi-Fi channel will pick it up.

Each packet carries 96 stereo sample-pairs (~2 ms of audio), plus a small header with the ADPCM predictor state and a microsecond timestamp so the sink can track clock offset.

Key specs

Parameter	Value
Sample rate	48 kHz
Bit depth	24-bit (I2S 32-bit slot)
Channels	2 (stereo)
Codec	IMA ADPCM, interleaved L/R nibbles
Frame size	96 samples (~2 ms)
Packet size	115 bytes (19 header + 96 payload)
Wi-Fi channel	11 (configurable)
TX power	15 dBm (configurable)
Clock source	APLL (low-jitter)

---

Hardware

Components

• ESP32 dev board (dual-core, any variant with enough GPIOs)
• PCM1808 ADC module -- stereo 24-bit audio ADC, I2S/left-justified output

Wiring

PCM1808 Pin	ESP32 GPIO	Function
SCK (BCLK)	GPIO 27	Bit clock
WS (LRCK)	GPIO 25	Word select
DOUT	GPIO 26	Serial data in
SCKI / MCLK	GPIO 0	Master clock
VCC	3.3V	Power
GND	GND	Ground

Set the PCM1808 FMT pins for I2S / Philips mode. The ESP32 provides MCLK at 256x the sample rate (12.288 MHz) via APLL on GPIO 0.

---

Requirements

• ESP-IDF v5.5.2
• An ESP32 target (not ESP32-S3). The project is configured for plain ESP32 in sdkconfig.defaults.

---

Build and Flash

# Set up ESP-IDF environment (adjust path to your installation)
. $HOME/esp/esp-idf/export.sh

# Build
idf.py build

# Flash (replace /dev/ttyUSB0 with your port)
idf.py -p /dev/ttyUSB0 flash

# Monitor serial output
idf.py -p /dev/ttyUSB0 monitor

On Windows with PowerShell:

D:\esp-idf\export.ps1
idf.py build
idf.py -p COMXX flash monitor

---

Configuration

All tunable values are #define constants at the top of main/main_adpcm.c:

Define	Default	Description
TEST_TONE_MODE	0	Set to 1 to output a 440 Hz sine instead of I2S input
SAMPLE_RATE	48000	Audio sample rate in Hz
PAYLOAD_SIZE	96	ADPCM bytes per packet (1 byte = 1 stereo sample-pair)
WIFI_CHANNEL	11	Must match the sink
WIFI_TX_POWER_QDBM	60	TX power in quarter-dBm (60 = 15 dBm)
PIN_MCLK	GPIO 0	I2S master clock
PIN_BCLK	GPIO 27	I2S bit clock
PIN_WS	GPIO 25	I2S word select
PIN_DIN	GPIO 26	I2S data input from PCM1808

---

How It Works

1. I2S capture -- The ESP32 acts as I2S master, providing MCLK/BCLK/WS to the PCM1808. Audio comes in as 32-bit slots in Philips format; the upper 24 bits are extracted and clamped.

2. ADPCM encoding -- Each frame of 96 stereo sample-pairs is encoded into 96 bytes of IMA ADPCM. Left and right channels are interleaved per-nibble within each byte (high nibble = left, low nibble = right). The encoder maintains separate predictor/index state for each channel.

3. Packet assembly -- A 19-byte header is prepended with the magic marker, a sequence number, the current ADPCM predictor state for both channels, and a source timestamp in microseconds. The sink uses the predictor state to initialize its decoder (no state dependency between packets), and the timestamp for clock-sync.

4. ESP-NOW broadcast -- The assembled packet is sent to the broadcast MAC address. No encryption, no peer pairing. Any ESP-NOW receiver on the same channel will get it.

5. Task pinning -- The audio capture/encode task runs on CPU core 1 to stay off the Wi-Fi core (core 0). A low-priority status task on core 0 prints throughput and peak-level stats every 5 seconds.

---

Serial Output

At runtime the status task logs something like:

I (5000) ADPCM_SRC: tx=940 ok=938 fail=2 peak=1234567 rate=188/s pps=188/s

• tx -- total packets attempted
• ok -- successful sends
• fail -- send failures (usually TX queue full, recovers on its own)
• peak -- peak absolute sample value (24-bit scale, max 8388607)
• rate -- successful sends per second
• pps -- packets per second

---

Project Structure

esp-now-audio-source/
  CMakeLists.txt            Root project file
  sdkconfig.defaults        ESP32 build defaults (240 MHz, WDT, Wi-Fi buffers)
  main/
    CMakeLists.txt           Component registration
    idf_component.yml        IDF component dependencies
    main_adpcm.c             All source code (capture, encode, transmit)
  docs/
    images/                  Component photos for reference

---

License

This project is provided as-is for personal and educational use.