Naim MUSIC SERVER DATABASE DATASHEET

The Naim Hard Disk Player and Music Server Database
Alan Ainslie Naim Audio
alan.ainslie@naimnet.com
Abstract
Music Servers have traditionally stored music in association with a database to allow retrieval of music at an Album level. While this is in accordance with the great majority of commercially available databases it is certainly not the manner in which a normal user selects music from conventional storage.
Music Servers, therefore, generally fall far short of the requirements of serious listeners of Classical Music, Jazz, and Compilation albums.
A better solution had to be developed before Music Servers and Hard Disk players attain ubiquity in the market alongside conventional sources such as CD players.
This paper demonstrates how information at Track level with Naim Extended Metadata can give a user experience beyond the convenience of conventionally selecting a CD from the shelf. Powerful database processing and search mechanisms deliver an intuitive user interface for both the casual user as well as the dedicated musicologist, whether playing music from the Hard Disc or simply playing a CD album in real-time.
1. Background
Traditionally, physical vinyl record and CD collections have been stored on shelves with spines visible. The content of the spines reflects the marketing endeavours of the record company on the one hand and also serves to indicate contents of the single disc or album for selection by the user. Selection of an album quickly offers additional information as soon as the album is removed from the shelf, as track information is readily visible along with other information relating to artists and the specific recording.
In order for a Music Server or Hard Disk Player to match or surpass the convenience of selecting discs from a collection on a shelf, a database internal to the Server has to be constructed to match the ripped disc.
The only unique identification for the disc relates to the properties of the data on the disc: number of tracks, track lengths, total play time etc. Information contained on the CD sleeve is not known to the Server.
The combination of tracks, track times etc. is possibly unique, but not always guaranteed to be so. At the time of ripping the Server will interrogate one or several on­line databases in order to match the disc being ripped with a managed and maintained database. The relevant database entry is downloaded, and together with Cover
Art is used to give the user a local database within which music can be selected for play.
2. Current Situation
The shortcomings of the current state of the art of cataloguing ripped CDs fall into several categories.
Virtual music collections where the data representing the music is stored, but where there is no physical carrier create a new problem for cataloguing. Initially the industry responded with several commercial databases offering cataloguing at an ALBUM level:
ARTIST (SINGLE ARTIST FOR THE ALBUM) ALBUM GENRE
And within the selected album simply:
SONG
will identify any individual track.
This is the situation, supported additionally with Cover Art that has been used by the various free or subscribed databases for a number of years.
ALBUM LEVEL TRACK LEVEL Searchable Searchable Artist Y Song Y Album Y Genre Y Cover Art
While this situation will work for simple Albums where the Album title is unambiguous and the Artist is the same for all tracks, the system immediately fails for compilations, where there are many possible artists individual to each track in the album.
Virtually all Classical music is assembled onto an album as a compilation, and therefore it is implicit that the current situation fails with classical music.
2.1 International releases may be different
Albums may well exist in several versions in different territories. These may have a different track makeup or different Cover Art, even the same tracks but in different order, or differing track lengths due to editing refinements in later releases.
2.2 Gapless playback Albums may be designed for continuous play. The
concept of ‘tracks’ being merely to identify location on the physical disc. E.g. Opera, or Pink Floyd: The Wall.
www.naimaudio.com - 1 - 24
th
April 2008
2.3 Multi Discs
Albums may be part of a multiple disc set
2.4 Data Tracks
Albums may contain material additional to playable music – data tracks etc.
2.5 Metadata isn’t always correct
Integrity of data is paramount. Mistakes in data entry mean that user-submitted databases are not generally reliable.
3. Requirements
The requirements for a reliable database from which to identify ripped CDs are summarised below:
3.1 Correctly identifying and allowing free searching within ARTIST, ALBUM, GENRE, SONG CDs.
3.2 This is adequately covered by the available commercial databases with one important shortcoming. Unless the data is edited or subject to some verification process the shortcoming the incorrect metadata could be displayed, resulting in end-user frustration.
3.3 Correctly identifying and allowing free searching within Compilation CDs, including virtually all Classical and many Jazz CDs.
3.4 There are several types of compilation disc to be found in a typical CD collection, including:
Classical Same composer, various works, various
artists
Classical Same orchestra or artists, various
composers, various types of music
Classical Same type of music, various composers,
various artists.
Sampler Various unrelated tracks, various unrelated
artists.
Sampler Same artists, re-allocation of tracks from
other releases
Sampler Same record label but various artists and
tracks
3.5 Correctly identifying and allowing free searching within multiple disc sets. Providing that there is sufficient cross checking and binding of discs in a multi-disc set then the situation is as above. Multiple discs can be combined into a single playing experience using the server Create Playlist function.
4. The Naim Solution
The solution devised by Naim is creating a database internal to the server populated by data from AMG among other databases, with sophisticated internal processing of the downloaded database entries.
AMG provide data at both Album and Track level. Importantly the database is manually edited: from both submissions from Record Companies and from User Submissions, manual database editing creates a master database, which initially promises to meet the specific requirements of Naim as below.
Additionally the database is configured by Naim to be self-optimising and various user tools as detailed in
section 4 ensure a comfortable user experience. The full amount of AMG metadata is actually downloaded: nothing is discarded. This gives scope for additional functionality in response to market sentiment.
The basic engine for the Naim Hard Disc Player and Music Servers is provided by Digital Fidelity and is based around a powerful XML database application. Further specialised functions ensure that the sound quality achieved is well in advance of industry standards.
5. The Naim Extended Music Database
Correctly creating a database that will be effective for Compilation CDs requires that the search fields operate at both ALBUM level and at TRACK level.
The search fields proposed in the Naim Music Server and Hard Disc Player are: -
ALBUM LEVEL Album Title Album Artist 1 Album Artist 2 Album Genre
TRACK LEVEL Track Titl e Work Track Artist 1 Track Artist 2 Track Artist 3 Track Artist 4
The concept of Artist includes:
Artist or Group or Soloist or Conductor Composer Orchestra or Group
Naim introduces the concept of PEOPLE for all Artists, Composers etc, to avoid users having to manage failed searches due to incorrect recollection of the precise Artist attribute.
There is therefore within The Naim servers and Naim Hard Disc Players the concept of PEOPLE SEARCH.

ALBUM LEVEL

DISPLAY SEARCH TITLE Y Y ARTIST NAME Y Y PERFORMER / ORCHESTRA GENRE Y Y COVER ART Y
Y Y

TRACK LEVEL

TRACK COUNT DISPLAY SEARCH TRACK TITLE Y Y WORK / PIECE / MOVEMENT COMPOSER Y Y PERFORMER / SOLOIST CONDUCTOR Y Y
Y Y
Y Y
PEOPLE SEARCH
PEOPLE SEARCH
www.naimaudio.com - 2 - 24
th
April 2008
Loading...
+ 3 hidden pages