Original README content further below
This repository is a port of the original Opus code for the ESP32. It provides a component.mk makefile with the necessary configuration to compile and use the Opus audio codec on the ESP32. It also removes the original build system of Opus to not clutter the filetree and confuse the ESP32 build system.
This repository can be added as submodule to any valid ESP32 component directory.
The easiest way to do this is to create a component/ directory in the root
of an ESP project, using git submodule add to clone this repo into there,
and to ensure that this new component is recognized via make list-components.
Note: Currently the component.mk file might only work for the older build system of the ESP-IDF, however creating a config file for the newer build system should be fairly trivial.
After adding this component to the build configuration, the "opus.h" file can be included. All fixed point Opus decoder and encoder methods are available, consult the Opus documentation for examples.
Opus is a fairly heavy codec, and it might not always work smoothly on the ESP32. The following things have to be taken into account:
- Encoding and Decoding heavily utilize Stack via
alloca. It is recommended to have at least 30kB of Stack reserved for the FreeRTOS thread that is running the Opus encoding/decoding (the stack is freed after the Opus call, so a single thread can encode and decode). - Decoding data on the ESP32 is fairly lightweight, and will consume about 30% of the CPU time at 240MHz on a single core, even at 48kHz samplerate. This makes it fairly usable.
- Encoding however is much more intensive. 16kHz samplerate at encoding complexity 1 will use 70% CPU. 24kHz samplerate at anything above complexity 4 will not be able to run at realtime. 48kHz can only run at realtime with complexity 1, and even then it uses 85% CPU. It still sounds understandable at 16kHz if you just need basic VoIP or voice recognition though, but don't expect to encode high quality music.
Additionally, the WiFi of the ESP32 is very capable of high throughputs, so the Samplerate is rarely an issue. What I have found to be problematic is the unstable latency of the network. The ESP will receive in "chunks" of a few packets, then pause for about 300ms, give or take. If you want to use Opus to stream Audio, make sure to have a bigger buffer of one or two seconds worth of audio on the receiving end, so that these lagspikes can be smoothed out. Once you do that though it's quite easy to have a very nice sounding audio stream.
Todo: Add a class that provides an easy to use ESP32 MQTT Audio stream. It'll need to handle decoding the data, providing it to the XasWorks audio core (which will need a slight rework, oh well), and keeping the buffer in synch by varying playback speed a bit (clock drift and stuff).
If I do get around to doing this, it'll be in my XasWorks repository, so check it out.
Original readme:
Opus is a codec for interactive speech and audio transmission over the Internet.
Opus can handle a wide range of interactive audio applications, including Voice over IP, videoconferencing, in-game chat, and even remote live music performances. It can scale from low bit-rate narrowband speech to very high quality stereo music.
Opus, when coupled with an appropriate container format, is also suitable for non-realtime stored-file applications such as music distribution, game soundtracks, portable music players, jukeboxes, and other applications that have historically used high latency formats such as MP3, AAC, or Vorbis.
Opus is specified by IETF RFC 6716:
https://tools.ietf.org/html/rfc6716
The Opus format and this implementation of it are subject to the royalty- free patent and copyright licenses specified in the file COPYING.
This package implements a shared library for encoding and decoding raw Opus bitstreams. Raw Opus bitstreams should be used over RTP according to https://tools.ietf.org/html/rfc7587
The package also includes a number of test tools used for testing the correct operation of the library. The bitstreams read/written by these tools should not be used for Opus file distribution: They include additional debugging data and cannot support seeking.
Opus stored in files should use the Ogg encapsulation for Opus which is described at: https://tools.ietf.org/html/rfc7845
An opus-tools package is available which provides encoding and decoding of Ogg encapsulated Opus files and includes a number of useful features.
Opus-tools can be found at: https://git.xiph.org/?p=opus-tools.git or on the main Opus website: https://opus-codec.org/
Todo: Add testing according to the Opus test system. This is currently not implemented on the ESP32 variant of libopus
Note: Fixed point is used for the ESP32. Floating point is disabled via a compiler define. The ESP32's FPU is way too slow for this.
This implementation uses floating-point by default but can be compiled to use only fixed-point arithmetic by setting --enable-fixed-point (if using autoconf) or by defining the FIXED_POINT macro (if building manually). The fixed point implementation has somewhat lower audio quality and is slower on platforms with fast FPUs, it is normally only used in embedded environments.
The implementation can be compiled with either a C89 or a C99 compiler. While it does not rely on any undefined behavior as defined by C89 or C99, it relies on common implementation-defined behavior for two's complement architectures:
o Right shifts of negative values are consistent with two's complement arithmetic, so that a>>b is equivalent to floor(a/(2^b)),
o For conversion to a signed integer of N bits, the value is reduced modulo 2^N to be within range of the type,
o The result of integer division of a negative value is truncated towards zero, and
o The compiler provides a 64-bit integer type (a C99 requirement which is supported by most C89 compilers).