ESP32 Audio Streamer

An ESP32 firmware for turning an I2S microphone into a small network audio source.

It was built for always-on outdoor/bird-audio use, especially with BirdNET-Go, but it also works with common players such as VLC and ffplay.

Tested on a Wemos D1 Mini32. Other ESP32 boards should work if the I2S pins are adjusted.

• Captures mono audio from an I2S MEMS microphone
• Streams live audio over:
  • HTTP: http://<device-ip>:81/stream.wav
  • RTSP/TCP: rtsp://<device-ip>:8554/audio/
• Provides a browser UI at http://<device-ip>/
• Lets you configure Wi-Fi from a setup access point on first boot
• Includes BirdNET-Go-friendly audio profiles:
  • Quality — 48 kHz
  • Stability — 24 kHz
• Can optionally sleep at night and wake at civil dawn

Default I2S wiring:

Default capture channel is right. You can change pins/channel in local_env.ini.

Use either the PlatformIO VS Code extension or the CLI.

cp local_env.ini.example local_env.ini

Edit local_env.ini if needed for:

• I2S pins
• location/timezone
• deep sleep behavior
• optional Wi-Fi seed credentials

Wi-Fi credentials do not need to be hardcoded; the device can be configured from the setup web page.

pio run --target upload

pio device monitor -b 115200

On first boot, if no saved Wi-Fi credentials exist, the ESP32 starts a setup access point.

Connect to:

ESP32-Audio-Setup-XXXXXX

Then open:

http://192.168.4.1/

Save your Wi-Fi credentials. The page and serial log will show the device IP after it connects.

http://<device-ip>/

From the UI you can:

• See stream status
• Play the HTTP stream
• Change audio profile
• Change HPF / gain conversion settings
• Restart audio
• Manage Wi-Fi
• View logs and system status

Recommended URLs:

RTSP is TCP/interleaved only. If a client first tries UDP, the firmware rejects that setup and waits for a TCP RTSP setup.

Only one stream client can be active at a time.

Start with:

• URL: rtsp://<device-ip>:8554/audio/
• Audio profile: Stability — 24 kHz if Wi-Fi is marginal
• Audio profile: Quality — 48 kHz if the stream is stable

The profile can be changed from the Audio page. After changing profile, click Restart Audio.

Build:

pio run

Upload:

pio run --target upload

Monitor:

pio device monitor -b 115200

Test RTSP with ffplay:

ffplay -rtsp_transport tcp rtsp://<device-ip>:8554/audio/

Test HTTP WAV:

ffplay http://<device-ip>:81/stream.wav

Detailed information has been moved out of this README:

• Streaming details
• Configuration reference
• Web UI and API reference
• Architecture notes

This project was influenced by BirdNET-Go ESP32 microphone work from the community, especially:

• birdnetgo-esp32-rtsp-mic
• birdnetgo-m5stack-atom-echo-rtsp-mic