Mozzi  version v1.1.0
sound synthesis library for Arduino

This example demonstrates the Sample class.

Example playing samples encoded with Huffman compression.
Demonstrates the SampleHuffman class.
SampleHuffman, most of this explanation, and the script are adapted from "audioout",
an Arduino sketch by Thomas Grill, 2011 http//
Huffman decoding is used on sample differentials,
saving 50-70% of space for 8 bit data, depending on the sample rate.
This implementation just plays back one sample each time next() is called, with no
speed or other adjustments. It's slow, so it's likely you will only be able to play one sound at a time.
Audio data, Huffman decoder table, sample rate and bit depth are defined
in a sounddata.h header file. This file can be generated for a sound file with the
accompanying Python script, in Mozzi/extras/python/
Invoke with:
python --sndfile=arduinosnd.wav --hdrfile=sounddata.h --bits=8 --name=soundtablename
You can resample and dither your audio file with SOX,
e.g. to 8 bits depth @ Mozzi's 16384 Hz sample rate:
sox fullglory.wav -b 8 -r 16384 arduinosnd.wav
Alternatively you can export a sound from Audacity, which seems to have less noticeable or no dithering,
using Project Rate 16384 Hz and these output options:
Other uncompressed files, Header: WAV(Microsoft), Encoding: Unsigned 8 bit PCM
The header file contains two lengthy arrays:
One is "SOUNDDATA" which must fit into Flash RAM (available in total: 32k for ATMega328)
The other is "HUFFMAN" which must also fit into Flash RAM
Audio output on digital pin 9 on a Uno or similar, or
DAC/A14 on Teensy 3.1, or
check the README or
Mozzi help/discussion/announcements:!forum/mozzi-users
Tim Barrass 2013, CC by-nc-sa.
#include <MozziGuts.h>
#include <SampleHuffman.h>
#include "umpah_huff.h"
void setup() {
void updateControl(){
AudioOutput_t updateAudio(){
return MonoOutput::from8Bit(;
void loop() {