From: Silvia Pfeiffer <Silvia.Pfeiffer@csiro.au> To: lwn@lwn.net Subject: Announcement: MPEG audio analysis tools Date: Mon, 18 Mar 2002 19:26:43 +1100 We are proud to announce the latest version of our frequency-domain audio analysis tools: (MPEG) Maaate Version 0.2.1 and bewdy 0.2.0a http://www.cmis.csiro.au/maaate/ (MPEG) Maaate: ============== Maaate is a C++ frequency domain audio analysis toolkit. http://www.cmis.csiro.au/maaate/docs/index.html Maaate is a set of libraries supporting frequency-domain based feature analysis. Although it currently only works on MPEG compressed audio files in the compressed domain, the interface to plug-in other file formats (uncompressed or compressed) exists. Maaate also contains some 30 audio feature extraction, segmentation and classification algorithms. It's all published under the GNU GPL. Maaate has now reached a third and quite stable development level. Download it from: http://www.cmis.csiro.au/maaate/docs/download.html Highlights of Maaate version 0.2.1: - a generic frequency-domain based analysis framework - a specific MPEG compressed (frequency) domain analysis library - 30 analysis modules covering * energy features (usable for loudness statistics), * bandwidth features (usable for speech/music segmentation), * spectral statistics (usable for sound type distinction), * silence statistics (usable for speech/music segmentation), * noise features (usable for explosion or crowd cheer detection), * generic segmentation algorithms (thresholds on features) and * some generic statistical algorithms (histogram, variance). - extensive documentation on all modules is also available online (http://www.cmis.csiro.au/maaate/docs/modules.html) Bewdy : ======= Play around with the compressed audio content of MP3 files! http://www.cmis.csiro.au/maaate/bewdy/index.html Bewdy is a program to visualise frequency domain audio feature analysis and segmentation algorithms. For example, it can display the spectral content of an MPEG audio file in the compressed domain and allows you to play around with analysis algorithms on that data. It uses the (MPEG) Maaate libraries for that purpose and can thus load up any analysis module that is implemented for Maaate. The pictures at http://www.cmis.csiro.au/maaate/snapshots.html and at http://www.cmis.csiro.au/maaate/bewdy/docs/handbook.html show off more of bewdy. Download bewdy from: http://www.cmis.csiro.au/maaate/bewdy/index.html Download Maaate from: http://www.cmis.csiro.au/maaate/docs/download.html About CSIRO audio analysis research: ==================================== Here at CSIRO, an Australian government research organisation, we perform research into the analysis of music and sound. We create software to unravel the structure and texture of recorded audio, and we develop systems which make use of the information extracted. Examples of such systems include searching for music features and indexing soundtracks. Much of our work is available as open source software. See: http://www.cmis.csiro.au/audio/ --- Silvia Pfeiffer <Silvia.Pfeiffer@csiro.au>