Digital Audio
Jon Dunn
Digital Library Program
Indiana University
June 23, 2003

Why digital files?
Why digital?
Quality does not degrade from generation to generation
Why files?
Network delivery possible
New forms of interactivity
Digital media formats become obsolete quickly; migration essential
Format migration hard to manage for physical media; must be automated to be sustainable

Digital Audio
Sampled waveform
Pulse code modulation, or PCM
PCM characteristics which determine quality
Sample rate (hertz)
Determines frequency range reproduced
Sample rate = 2 * frequency range (Nyquist)
Sample size (bits)
Larger sample size = less quantization noise
Number of channels

Digital Audio

File formats for PCM audio
Two main formats (actually many more)
WAV (Microsoft/IBM)
AIFF (Apple)
All essentially interchangable
Header + sample data
No true standard (i.e. ISO standards process) in common use
WAV is defacto standard

Audio compression
Uncompressed audio is too big
CD quality: 700 MB for one hour
Nearly all online audio is compressed
Lossy compression using 'psychoacoustic' techniques
Based on understanding of human hearing and audio processing
Standard and proprietary formats

Compressed Audio: Standards
Layers 1, 2, and 3 (‘MP3’)
Typically 128-384 kilobits/second (for stereo)
Layers 1, 2, and 3 and AAC (Advanced Audio Coding)
Typically 96-384 kilobits/second (for stereo)
Improved AAC
Synthetic audio
Low bit rate parametric audio

Compressed Audio: Proprietary Formats
QDesign Music
Used with Apple QuickTime
Used with RealSystem
Windows Media Audio
Used with Windows Media

Internet Audio Delivery Methods
Web server (HTTP)
Streaming server
RealNetworks Helix server
Microsoft Windows Media Services
Apple QuickTime Streaming server
Encrypted/tethered download
Commercial services

Bandwidth Requirements
Audio formats:
CD quality audio: 1400 kilobits/second (kbps)
MPEG audio: 96 - 384 kbps (typical)
RealAudio: 24+ kbps
Network technologies:
Ethernet LAN: 10/100/1000 megabits/second (Mbps)
Cable modem: 512 kbps - 3 Mbps
DSL: 128 kbps - 1.5 Mbps
ISDN: 56-128 kbps
Telephone modem: 56 kbps

Equipment for digitization
Computer workstation
Audio source equipment
Analog-to-digital converter
Sound card in PC or external box
Audio editing, compression software

Digitization: Two Approaches
Expensive equipment, software
Highly-trained technicians
Better-than-CD-quality file formats
Commodity equipment, software
Student workers
CD-quality files (sort of)

Indiana University Music Library Digitizing Lab

Indiana University School of Music
Located on Bloomington campus
140 faculty, 1400 students
Strong programs in both performance and academic (music theory, history, education, composition) areas
Cook Music Library
Large collection
320,000 printed volumes
96,000 scores
218,000 performance parts
130,000 sound recordings
New facility (1996)
60,000 sq. ft., 130 student computer workstations

Digital library of music sound recordings (~8000 complete titles)
Some musical scores (~200 titles) - experimental
Used primarily for course reserves…
Listening assignments for students
In-class presentation of examples
…but not just reserves
Access from limited locations
Music Library, Main Library, faculty offices/studios, some classrooms

Simon Center

Music Library Workstations

Slide 18

Slide 19

Slide 20

Slide 21

Simple approach:
Audio e-reserves
Digitize selections
Streaming server
Your university may already provide a streaming service
Or MP3 download
Authentication based on course
Web interface / database
Possibly use existing e-reserves system, LMS reserves module, or course management system
More info:
Google search on "audio reserves"
DAL-L Listserv: Digital Audio in Libraries

Four-year research project
Started October 1, 2000
Funding from NSF and NEH through Digital Libraries Phase 2 (DLI2) program
Large interdisciplinary team of investigators
Faculty: Music, Information Science, Law, Computer Science
Librarians and technologists: Libraries, University Information Technology Services
Bloomington and Indianapolis campuses

The Variations2 System
Integrated access to music in all formats
Digital audio recordings
Score images
Score notation
Multiple task-appropriate user interfaces
Supports research in metadata, usability, copyright, music instruction, and computer networks
Staged development

Variations2 Data Model

Variations2 Data Model
Appropriate metadata elements attached to each entity
Can import and map MARC records
Closely related to FRBR
International Federation of Library Associations Report on the Functional Requirements of Bibliographic Records, 1997

Variations2 Data Model: Example

Slide 28

Work Structure: Example

Work Structure: Example

Work Structure: Example

Slide 32

Slide 33

Slide 34

Slide 35

Slide 36

Intellectual property
Complexity of music rights
Lack of interest by labels (with exceptions)
č No market
Fair use
Limit by location
Limit by user
Limit by amount
Will things change?
MLA "Copyright for Music Librarians" site:

Commercial services: consumer
Subscription services
Pressplay, Rhapsody, Musicnet on AOL
Monthly subscription: $9.95
Streaming, tethered downloads
Charge for CD burning/MP3 download (varies)
Individual subscription only, no library plans
Pay-per-download services
Apple iTunes Music Store
99c/track for download
Can be burned, loaded into iPod, but limitations
No library plans
Data model problems for classical music

Commercial services: library
Classical Music Library (
10,000+ recordings
Mainly obscure labels and performances
Reference materials
Images, biographies (source?)
Hosted or local options
Priced per simultaneous user
Starting at $995 for 3 users
Naxos record label
Details not yet available

On the horizon: Music IR
Content-based music information retrieval
Analysis, searching
"Query by humming"
More information:
4th International Conference on Music Information Retrieval, Baltimore, MD, October 26-30, 2003

Thank you!
More details:
Contact information:
Jon Dunn -