Skip to content

sippy/libg722

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

122 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

libg722

Build & Test

Introduction

The G.722 module is a bit exact implementation of the ITU G.722 specification for all three specified bit rates - 64000bps, 56000bps and 48000bps. It passes the ITU tests.

To allow fast and flexible interworking with narrow band telephony, the encoder and decoder support an option for the linear audio to be an 8k samples/second stream. In this mode the codec is considerably faster, and still fully compatible with wideband terminals using G.722.

History

The code was originally written by Milton Anderson milton@thumper.bellcore.com from BELLCORE, and has been modified by the Chengxiang Lu and Alex Hauptmann from the Speech Group, School of Computer Science, Carnegie Mellon University, to be fairly fast and efficient, while retaining high fidelity.

Steve Underwood steveu@coppice.org improved the code a lot later on and contributed it into several popular open source projects.

Himanshu Soni sonihimanshu@gmail.com has adjusted some coefficients to avoid over/under-flows in the decoder.

Phil Schatzmann phil.schatzmann@gmail.com has added cmake-glue and Arduino support.

Librarized by Sippy Software, Inc.

Build and Install library:

MacOS & Linux

git clone https://github.com/sippy/libg722.git
cmake -B libg722/build -S libg722
make -C libg722/build clean all test install

Note for macOS users: The library will be installed to ~/Library/libg722 by default. If you prefer a different location, you can specify it with:

cmake -B libg722/build -S libg722 -DCMAKE_INSTALL_PREFIX=/your/preferred/path

After installation, you may need to add the library path to your environment:

export DYLD_LIBRARY_PATH="$HOME/Library/libg722/lib:$DYLD_LIBRARY_PATH"

iOS

git clone https://github.com/sippy/libg722.git
cmake -B libg722/build-ios-device -S libg722 \
  -DCMAKE_SYSTEM_NAME=iOS \
  -DCMAKE_OSX_ARCHITECTURES=arm64 \
  -DCMAKE_OSX_SYSROOT=iphoneos \
  -DCMAKE_OSX_DEPLOYMENT_TARGET=12.0
make -C libg722/build-ios-device

Install Python module from PyPy:

pip install G722

Build and Install Python module from source code:

git clone https://github.com/sippy/libg722.git
pip install libg722/

This builds the core package (G722) with no hard dependency on NumPy. To install NumPy decode support on demand:

pip install "G722[numpy]"

or:

pip install G722-numpy

LIBG722_BUILD_MODE controls build profile for the main G722 extension:

  • production: build with optimization (-O2, or /O2 on Windows).
  • debug: build with -g3 -O0.
  • auto (default): if . is a git repository and git is available, run git diff v{version} -- .; build in debug mode when it differs, otherwise production. If . is not a git repository or git is unavailable, use production.

LIBG722_PACKAGE_VARIANT controls which Python distribution is built from this repository:

  • core (default): builds/publishes G722.
  • numpy-addon: builds/publishes G722-numpy from python/G722_numpy_mod.c.

G722(sample_rate, bit_rate, use_numpy=None) accepts an optional use_numpy flag:

  • True: return NumPy arrays from decode() (raises if G722-numpy is not installed).
  • False: return Python array('h') from decode().
  • omitted: auto-detect the optional backend and use NumPy when available, otherwise array('h').

Pull library into your Docker container:

ARG BASE_IMAGE=debian:sid-slim
ARG LIBG722_IMAGE=sippylabs/libg722:latest-debian_sid-slim
FROM ${LIBG722_IMAGE} AS libg722
FROM ${BASE_IMAGE} AS build
COPY --from=libg722 /usr/local/lib/libg722.* /usr/local/lib/
COPY --from=libg722 /usr/local/include/g722* /usr/local/include/

License

This code is mostly Public Domain. Library test code is under BSD 2-clause license.

See LICENSE for copyright details and each individual file for specifics.

About

The ITU G.722 codec, Copyright (C) 2005 Steve Underwood, Copyright (c) CMU 1993 Computer Science, Speech Group Chengxiang Lu and Alex Hauptmann

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors