Sonic Annotator is a utility for batch feature extraction from audio files. Sonic Annotator (formerly known as "Runner") is a command line program for batch extraction of audio features from multiple audio files. The basic idea is to abstract out the process of doing the feature extraction from the extraction methods, by using Vamp plugins for feature extraction. The output formats are RDF, CSV, and others; new output formats can be added with a modest amount of programming work.

Sonic Annotator is conceived of as having two main uses within the OMRAS2 project. First, it is useful for creating publishable feature data within the semantic web using the Music Ontology and Audio Features Ontology. Secondly, it can be used to created feature lists for use with AudioDb.

Though the official name of this program is Sonic Annotator, the working name during development was Runner, and the command-line examples in this document reflect the working name in some places. These may change in future.

Sonic Annotator requires several different things to work.
1. First of all, it needs a collection of audio files. Typically these will be on the local disk, often within the same file system directory, although Sonic Annotator can also retrieve and process files from the internet given a list of URLs.
2. While not strictly necessary, each Vamp plugin should have an RDF file supplied with it, which contains metadata that Sonic Annotator can use to enhance the descriptive data it outputs. The user should not have to concern themselves with this, but it may be a concern if you write your own Vamp plugins.
3. An input file is also required that explains the necessary features to be extracted, and the settings that each extractor requires. This can be in RDF format, or in XML. See A Usage Example below for more information.
4. Output may be in CSV, RDF or certain other formats. The user can specify whether output is written to a single big file, to one output file for each input audio file, or to one output for each transform and input.

RDF, or Resource Description Format, is a simple way of expressing the relationship between two pieces of information. It is becoming an important element in the construction of web-based ontologies, especially with regards to the semantic web.

Sonic Annotator is called from the command line as follows:
Usage:
runner [-mr] -t trans.xml [...] -w <writer> [...] <audio> [...]
runner [-mr] -T trans.txt [...] -w <writer> [...] <audio> [...]
runner -s <transform>
runner [-lh]

Where <audio> is an audio file or URL to use as input: either a local file path, local "file://" URL, or remote "" or "ftp://" URL.
The following audio file extensions are recognised: aiff, au, avr, caf, flac, htk, iff, mat, mp3, ogg, paf, pvf, raw, sd2, sds, sf, voc, w64, wav, xi.

Playlist files in M3U format are also supported.

The license of this software is Freeware, you can free download and free use this audio utility software.