Von: David Hirst via cec conference
Datum: Fri, 29 Jul 2016
Betreff: Re: [cec-c] Database
Sound Databases
Ways of categorising and organising sound files on computer usually
start with organising sounds into directories or folder according to
some kind of categorisation. This traditional system is a hierarchical
system, with sounds stored in different levels of classification from
general to specific. This system starts to break down when a large
number of files and/or a large number of attributes makes finding
something, by searching manually, long and laborious. Using database
techniques means “tagged” files can be saved anywhere and then retrieved
using semi-automated search techniques.
Describing, tagging and categorising sound files
Describing, tagging and categorising sound files can be though of as two
main types:
Metadata tagging – Describing the file with attributes like Title,
Date, Composer, Artist, Genre, Sampling rate, bit rate, etc. Often
this type of data can be stored within the sound file, e.g. mp3,
WAV. This is usually a combination of data within the sound file
standard, and “hand-curated” data that is entered by hand by a
person. It is the sort of data used by music sales companies like
iTunes, which combines this information with music listener
information, in order to market music. Sometimes composers, sound
designers, and librarians want to add extra descriptors and
categories that must sit outside the sound file itself. Attributes
like: where, when, and what was recorded. Plus descriptions of
physical attributes of the sound (long/short, high/low), semantic
references (dog bark), and onomatopoeic references (whooshing sound)
etc etc. Metadata tagging produces a textual list of attributes.
Tagging using physical data derived from the sound – Usually this is
achieved by mathematical analysis of the sound file to produce
specific data from the sound file like: event onset times, spectral
centroid, spectral flux, pitch tracking, loudness, number of zero
crossings, etc, etc. The aim, in this case, is to reduce the sample
rate data to a specific (smaller) “vector” of attributes. Such
vectors can lead to the “finger-printing” of a file which can be
used to find similar files etc. e.g. Shazam. Physical signal
analysis produces a vector of attributes in text and numbers.
Regarding Metadata Programs
Paul Virostek has two articles on the creativefieldrecording.com site:
This one is an intro to Metadata:
And this one reviews various sound metadata apps:
Program Links
Here is a list of the reviewed metadata programs, and more. Most of the
commercial products have entry level and pro versions for people on
different budgets. Paul Virostek’s review concludes with suggestions for
people on various budgets:
Full-Featured Metadata Apps
Soundminer (Win/Mac) – from $199.00 to $899.00 Soundminer HD is a cut
down version, see store.soundminer.com/mac-win-hd-product.html
And read their Metadata white paper here:
http://www.soundminer.com/current/MI_Whitepaper.pdf
Basehead (Win/Mac) – http://www.baseheadinc.com
Library Monkey (Mac) – http://www.monkey-tools.com/products/
AudioFinder (Mac) – http://www.icedaudio.com/site/
ifoundasound – has been discontinued, but still has a free version
online for Windows users: ifoundasound.com/?page_id=6
Get soundly replaces ifoundasound and works with online libraries like
freesound.org
your local sound effects, see getsoundly.com
Apple iTunes (Win/Mac) Free – can be shoe-horned to include search and
metadata functions.
Simpler Metadata Editing Apps
Digital Confidence’s MetadataTouch (Win) $60 $80 –
http://www.digitalconfidence.com/MetadataTouch.html
TwistedWave (Mac) $80 – twistedwave.com/mac.html
Sound Devices Wave Agent (Win/Mac) Free with registration –
http://www.sounddevices.com/support/downloads/wave-agent
BWF MetaEdit (Win/Mac) – (free) bwfmetaedit.sourceforge.net
Splat Sound FX Database (Win) – (free) gersic.com/splat/
Limit Point Software’s Commenteer (Mac) –
http://www.limitpointstore.com/products/commenteer/
(Or purchase from the Apple App Store – $20)
Read-Only Metadata Apps
Snapper $79 – is like a quick look tool for Macs:
http://www.audioease.com/snapper/
but could be used in conjunction with Commenteer and the Finder to
classify and find files.
Other Apps not in the Review Article
Sonicwire Mutant (Win/Mac) – Free Version sonicwire.com/mutant
SoundLib (Win/Mac) – $15 http://www.rootsolutions.de/soundlib/index.htm
FindSounds Palette (Win) – Various Versions $20+
http://www.findsounds.com/palette.html
RyAudio Sample Librarian (Win) – 40€ http://www.ryaudio.com/
74mph Solutions WavSniper (Win) – $49.95
http://www.74mph.com/wavsniper/index.html
Revel Breezer (Win) – Free Express Edition / Home Edition 14.90€ /
Professional Edition 49.90€ http://www.revel-software.it/
Rapid Evolution (Win) http://www.mixshare.com/re3preview.html
– helps find key & bpm information
MusicBee (Win) getmusicbee.com/
– audio player with many customizable fields. (Free or donate.)
Make Your Own
Filemaker Pro (Mac) – a general database creation program. ($329 USD)
View its “Content Management” starter tool on this page:
http://www.filemaker.com/solutions/starter-solutions.html
OpenOffice (Multi-platform) – Free. Use the database design feature
“Base” to create your own database:
http://www.openoffice.org/product/index.html
Research Apps
Psychology researchers:
PRAAT – can perform various kinds of signal analysis and allow you to
annotate various points in a file using the “Text Grid” feature.
http://www.fon.hum.uva.nl/praat/
Education or Linguistic researchers, or anyone wanting to annotate video
or audio recordings, or do qualitative research:
NVIVO – Expensive but comprehensive. http://www.qsrinternational.com/product
Atlas TI – Also expensive but comprehensive. atlasti.com
Sonic Annotator is a batch tool for feature extraction and annotation of
audio files using Vamp plugins. vamp-plugins.org/sonic-annotator/
can work in association with:
Sonic Visualiser is an application for viewing and analysing the
contents of music audio files. http://www.sonicvisualiser.org
EAnalysis – is a similar program by Pierre Couprie.
For the many and varied Music Information Retrieval (MIR) applications,
see the “analysis” entries in lists like:
Two notables are:
Marsyas (Music Analysis, Retrieval and Synthesis for Audio Signals), and
MIR Toolbox – Functions written in Matlab dedicated to the extraction of
musical features from audio files.
Dr David Hirst
Jul 2016
/__/
Associate Professor David Hirst, PhD
Honorary Principal Fellow
Melbourne Conservatorium of Music
The University of Melbourne
Building 141, Gate 12, Royal Parade,
Parkville, Victoria 3010 Australia
Ph: 0413 325 001
Email: d.hirst@unimelb.edu.au
Compositions and research: davidhirstmusic.wordpress.com/about/
/The Faculty of VCA and MCM respectfully acknowledges the Boonwurung and
Wurundjeri people of the Eastern Kulin nation./