snapclient

Snapcast client for ESP32

Synchronous Multiroom audio streaming client for Snapcast ported to ESP32

Feature list

Description

I have continued the work from @badaix, @bridadan and @jorgenkraghjakobsen towards a ESP32 Snapcast client. Currently it support basic features like multiroom sync, network controlled volume and mute. For now it supports FLAC, OPUS, PCM 16bit audio streams with sample rates up to 48Khz maybe more, I didn’t test.

Please check out the task list and feel free to fill in.

I dropped the usage of ADF completely but copied stripped down, needed components to this project (using ESP-ADF v2.6). This was necessary because ADF was using flac in closed source precompiled library which made it impossible to get good results for multiroom syncing.

Codebase

The codebase is split into components and build on ESP-IDF v5.1.1. I still have some refactoring on the todo list as the concept has started to settle and allow for new features can be added in a structured manner. In the code you will find parts that are only partly related features and still not on the task list. Also there is a lot of code clean up needed.

Components

The snapclient functionanlity are implemented in a task included in main - but should be refactored to a component at some point.

I did my own syncing implementation which is different than @jorgenkraghjakobsen’s approach in the original repository, at least regarding syncing itself. I tried to replicate the behaivior of how badaix did it for his original snapclients.

The snapclient frontend handles communication with the server and after successfull hello hand shake it dispatches packages from the server. Normally these packages contain messages in the following order:

Each WIRE_CHUNK of audio data comes with a timestamp in server time and clients can use information from TIME and SERVER_SETTING messages to determine when playback has to be started. We handle this using a buffer with a length that compensate for for playback-delay, network jitter and DAC to speaker (determined through SERVER_SETTING).

In this implementation I have separated the sync task to a backend on the other end of a freeRTOS queue. Now the front end just needs to pass on the decoded audio data to the queue with the server timestamp and chunk size. The backend reads timestamps and waits until the audio chunk has the correct playback-delay to be written to the DAC amplifer speaker through i2s DMA. When the backend pipeline is in sync, any offset get rolled in by micro tuning the APLL on the ESP. No sample manipulation needed.

Hardware

You will need an ESP32 or ESP32-S2 and an I2S DAC. We recommend using a Lyrat board. For pinout see the config options.

-   ESP pinout                         MA12070P
------------------------------------------------------
                          ->            I2S_BCK        Audio Clock 3.072 MHz
                          ->            I2S_WS         Frame Word Select or L/R
                          ->            GND            Ground
                          ->            I2S_DI         Audio data 24bits LSB first
                          ->            MCLK           Master clk connect to I2S_BCK
                          ->            I2C_SCL        I2C clock
                          ->            I2C_SDA        I2C Data
                          ->            GND            Ground
                          ->            NENABLE        Amplifier Enable active low
                          ->            NMUTE          Amplifier Mute active low

Installation

Clone this repo:

git clone https://github.com/CarlosDerSeher/snapclient
cd snapclient

Update third party code (opus, flac and esp-dsp):

git submodule update --init

ESP-IDF environnement configuration

Snapcast ESP Configuration

Frist copy one of the template sdkconfig files and rename it to sdkconfig

cp sdkconfig_lyrat_v4.3 sdkconfig

then configure your platform:

idf.py menuconfig

Configure to match your setup

Compile and flash

idf.py build flash monitor

Merge bin to flash at 0x0 with web.esphome.io

esptool.py --chip esp32  merge_bin -o merged.bin --flash_size 4MB --flash_freq 80m 0x1000 build/bootloader/bootloader.bin 0x8000 build/partition_table/partition-table.bin 0xd000 build/ota_data_initial.bin 0x10000 build/snapclient.bin 0x370000 build/storage.bin

Test

Setup a snapcast server on your network

On a linux box:

Clone snapcast build and start the server

./snapserver

Pipe some audio to the snapcast server fifo

mplayer http://ice1.somafm.com/secretagent-128-aac -ao pcm:file=/tmp/snapfifo -af format=s16LE -srate 48000

Test the server config on other knowen platform

./snapclient  from the snapcast repo

Android : snapclient from the app play store

Contribute

You are very welcome to help and provide Pull Requests to the project.

We strongly suggest you activate pre-commit hooks in this git repository before starting to hack and make commits.

Assuming you have pre-commit installed on your machine (using pip install pre-commit or, on a debian-like system, sudo apt install pre-commit), type:

:~/snapclient$ pre-commit install
pre-commit installed at .git/hooks/pre-commit

Then on every git commit, a few sanity/formatting checks will be performed.

Task list

Minor task

This site is open source. Improve this page.