Sphinx is a speaker-independent large vocabulary continuous speech recognizer released under a BSD style license.

It is also a collection of open source resources and tools that allows developers and researchers to build speech recognition systems.

Sphinx-4 is a state-of-the-art speech recognition system written entirely in the Java programming language.

It was created via a joint collaboration between the Sphinx group at Carnegie Mellon University, Sun Microsystems Laboratories, Mitsubishi Electric Research Labs (MERL), and Hewlett Packard (HP), with contributions from the University of California at Santa Cruz (UCSC) and the Massachusetts Institute of Technology (MIT).

Sphinx-4 started out as a port of Sphinx-3 to the Java programming language, but evolved into a recognizer designed to be much more flexible than Sphinx-3, thus becoming an excellent platform for speech research.

CMU Sphinx Group is releasing are a set of reasonably mature, world-class speech components that provide a basic level of technology to anyone interested in creating speech-using applications without the once-prohibitive initial investment cost in research and development; the same components are open to peer review by all researchers in the field, and are used for linguistic research as well.

Note however that Sphinx is not a final product. Those with a certain level of expertise can achieve great results with the versions of Sphinx available here, but a naive user will certainly need further help. In other words, the software available here is not meant for users with no experience in speech, but for expert users.

Here are some key features of "CMU Sphinx 4":

· Live mode and batch mode speech recognizers, capable of recognizing discrete and continuous speech.
· Generalized pluggable front end architecture. Includes pluggable implementations of preemphasis, Hamming window, FFT, Mel frequency filter bank, discrete cosine transform, cepstral mean normalization, and feature extraction of cepstra, delta cepstra, double delta cepstra features.
· Generalized pluggable language model architecture. Includes pluggable language model support for ASCII and binary versions of unigram, bigram, trigram, Java Speech API Grammar Format (JSGF), and ARPA-format FST grammars.
· Generalized acoustic model architecture. Includes pluggable support for Sphinx-3 acoustic models.
· Generalized search management. Includes pluggable support for breadth first and word pruning searches.
· Utilities for post-processing recognition results, including obtaining confidence scores, generating lattices and embedding ECMAScript into JSGF tags.
· Standalone tools. Includes tools for displaying waveforms and spectrograms and generating features from audio

Requirements:

· Java 2 SDK, Standard Edition 5.0 or later
· Ant 1.6.0 or later
· Subversion (svn), but only if you want to interact directly with the svn tree (which is recommended).

What's New in This Release: [ read full changelog ]

· Large arbitrary-order language models
· Simplified and reworked model loading code
· Raw configuration and and demos
· HTK model loader
· Many code optimizations
· JSAPI-independent JSGF parser
· Noise filtering components
· Lattice rescoring
· Server-based language model