Publishers of technology books, eBooks, and videos for creative people

Home > Articles > Digital Audio, Video > Audio

  • Print
  • + Share This
From the author of

Streaming Architecture

The audio on the site is authored on a combination of Windows and Macintosh computers using the streaming formats covered in this book (RealMedia, QuickTime, Windows Media, and MP3). The following sections outline many of the tools and processes used for digitizing, encoding, and publishing.

Formats Used

Wherever possible, multiple formats are used for streaming audio on the site. For the 45-second song clips, 56Kbps dial-up modem users are provided with 32Kbps streams in QuickTime, Windows Media, and MP3, as well as a RealMedia SureStream for up to 64Kbps. Full-length tracks are provided as four multiple bit rate MP3 downloads (24Kbps, 96Kbps, 128Kbps, 192Kbps) and a low bit rate RealMedia streaming file. Live webcasts are typically provided in the format that the partner company prefers. In recent years, the site's webmaster has partnered with QuickTime, although previous webcasts have included RealMedia.

Encoders

When possible, any Web site that features music should offer low bit rate on-demand clips (segments) of as many songs as possible. Joe Satriani's discography includes approximately 150 songs, and http://www.satriani.com offers a 45-second clip of each one, authored as on-demand files for Windows Media, QuickTime, RealMedia, and MP3. 150 clips times the four formats equals 600 individual clips. Encoding these files is a potentially time-consuming task. All 600 sound clips were created using a combination of Xing's AudioCatalyst 2.1, Discreet's Cleaner 5.1, and custom Unix script programs.

First, each CD is put into a Windows machine running AudioCatalyst, and then the first 45 seconds of each track is selected to be ripped into an uncompressed WAV file. Certain tracks that begin more slowly (as well as live tracks with applause at the beginning) are set manually for a more appropriate 45-second section.

Before the ripping process begins, AudioCatalyst looks up album and song information through queries to the CD database service CDDB (http://www.cddb.com) and then configures to save each clip into individual files. Each file is named by song title and track number and placed in a folder named by the album title (handled by AudioCatalyst's useful naming configuration).

After the ripping process, Cleaner 5.1 compresses the extracted excerpts into MP3, QuickTime, RealMedia, and Windows Media on-demand streaming files. Special settings are created for each format desired (see "Formats Used" earlier in this chapter) to set bit rates, fade-ins/outs and author/title metadata and, for formats that had the option, the Comment/URL http://www.satriani.com/.

Because Cleaner 5.1 can only easily output files from each batch into one directory (without manually setting a new destination for each file), files are batch encoded for each album's collection of songs. Each album's folder of uncompressed WAV excerpt files is dragged into a new Cleaner batch window. The files are then selected and the destination directory for the encoded files is set to a new output folder named for the album. With all songs still selected, Cleaner's Advanced Settings option is chosen and all of the desired format settings are selected. After applying the settings, Start is clicked and Cleaner encodes the multiple files, one for each format. When complete, the process is repeated for each album.

When all the tracks have been encoded, a folder with subfolders for each album containing each clip encoded into four different formats is uploaded to the streaming server. A custom Unix script program is used to generate metafiles for the RealMedia, Windows Media, and MP3 files. Metafiles for the RealMedia and Windows Media clips include song title and album title information. This information is retrieved from a simple database of album information by cross-referencing the album name (which is also the name of the folder containing the encoded file) and track number (also in the encoded file's filename). Another custom script is used to insert the song and album information into ID3 tags for the MP3 files. The sound clips are then linked into each album's page within the discography section of the Web site.

Servers and Bandwidth

All on-demand streaming is served as part of the Web site hosting. Because the majority of audio content on the site is short song excerpts, overall bandwidth use is light.

However, when full songs are offered as part of an exclusive promotion for a new release, download popularity temporarily generates a considerable amount of additional bandwidth. Deals are typically set up with the Web site's ISP (ISP Networks in recent years) to trade bandwidth for a logo and link on the site.

Live special event webcasts also bring in too many listeners to handle using the regular day-to-day streaming servers. Typically, partnerships with other companies are created to stream these webcasts. Most recently, Apple has assisted in the encoding side, while Apple's streaming partner, Akamai, has handled the serving. Certainly, Satriani's popularity and major music label representation make it possible to form these partnerships. Nonetheless, the streaming audio industry is young enough that some organizations are willing to set aside immediate profit by using these showcases to increase consumer awareness of and demand for the technology.

  • + Share This
  • 🔖 Save To Your Account