Identification of Technology: Streaming Audio  (Top)

Streaming audio is a technology which has revolutionized how Internet users define "multimedia."  Prior to 1995, for a person surfing the web to hear audio, he or she had to download a huge audio file before being able to hear even a few minutes of speech or music.  The transfer ratio of an audio file is typically 5:1, meaning for every minute of audio, the user has to wait 5 minutes for that one-minute segment to download.  To hear a 30-minute audio program, the listener would have to wait over 2 hours, a delay which is not only unreasonable, but expensive if Internet access is charged by the hour.

Even if a person was willing to wait for the amount of time needed to download the file, any technical difficulty during transfer would mean that the entire file would be lost.  Also, the amount of available hard drive space would also pose a constraint, since audio files are so large.  All of these factors inhibited the widespread usage and availability of a true multimedia experience.

In 1995,  Progressive Networks (later renamed Real Networks) introduced a new technology which solved many of the existing problems with audio.  Streaming audio enables the listener to hear sound in real time: the user is listening to the sound file as it is being transmitted instead of having to wait for the file to download first.  Since the sound data is sent in "packets" over the network, each set of data is being heard immediately as it sent. The user does not have to store the information on the hard drive or risk more than a waiting a substantial amount of time only to lose the file; even if a packet of information is lost, the rest of the audio file is intact and can still be played back by the listener.
 

Identification of the Technology: Streaming Video (Top)

Streaming video is a technology which has expanded upon the strengths and opportunities presented by streaming audio. Only a few years earlier, video over the Internet was virtually impossible, due to the size of video files and the download time required. The emergence of video streaming technologies reshaped multimedia usage in much the same way as audio streaming.

As in audio, video streaming compresses the data and distributes it in small packets across Internet lines where it is decompressed by a user's PC. The video player on a  PC can continually request video data from a server, so that the user is watching video footage as it is being downloaded.

As recently as a year ago, streaming video still faced quality problems and was not usable for the majority of Internet users.  However today, while the video quality is still not optimal at lower modem speeds, streaming video packages are leveraging increasingly sophisticated technology to improve the quality problems of the past.
 
How They Work (Top)

Streaming audio uses codec (coder/decoder) technology to compress the audio file for distribution across the Internet. Once the original audio file has been digitized, it is encoded using a vendor-specific algorithm which compresses the data and allows efficient transmission over the network.

The audio data is then sent across Internet lines using either UDP, TCP, or IP Multicasting as a protocol to communicate the information.  The main players in the audio marketplace all rely on a different protocol, each which has its own benefits and limitations.

As the data is received, the browser uses a helper application to play back the file, buffering the first portion of the data. Buffering allocates a portion of memory on the user's hard drive to store a few packets of video or audio information.  As each buffer is played, new data is received from the server. This way, the player always can get information from the buffer instead of waiting to receive data from the server.

Streaming technology allows the user to skip forward or back to any portion of the file, since the transmission is "bidirectional": in addition to the server transmitting information to the player, the player can also transmit information to the server, requesting a specific audio packet of data.

Streaming video products use a process identical to that of audio compression: the encoder software uses a proprietary algorithm to create a compact form of the original file. The coder shrinks video by replacing the original frames with more compact versions by using mathematical algorithms such as wavelet, fractal, and discrete cosine transform (DCT). Encoders are also based on standards such as MPEG-1 and H.263. The effectiveness of the encoding process can be judged by video output quality, such as the level of artifacts, or objects introduced to the picture by compression.  Decoders, or players, decompress and play the video.

Most streaming video technologies use a server to ensure efficient video distribution to clients. Servers transmit video through a variety of network protocols including UDP, TCP, and HTTP. As in audio, the underlying protocol affects the overall playback quality of the video.

When low bandwidth limits the client's ability to receive data, the vendor's use of a scalable technology will ensure an unbroken audio stream by scaling the amount of video data transmitted. The simplest technique is stream thinning, in which the server doesn't transmit every frame to the client.  Encoding with a wavelet algorithm also lets you send less information per frame, losing image detail but preserving the frame rate.

Video streaming is implemented today using helper applications that work with your browser to detect multimedia hyperlinks.  Streaming technology begins playback by buffering the first part of the files.  Although  buffering results in a brief delay of usually 5 to 30 seconds, the waiting time is minimal when compared with the download of an entire video or audio file.  When the initial buffer is loaded, the video or audio clip begins to play and continues as the rest of the file is downloaded into the buffer simultaneously.
 

Limitations of Streaming Technology (Top)
 
Audio and video quality is influenced by a variety of factors. The smoothness of playback can be judged to some degree by the amount of "packet loss." Since both audio and video information is streamed over the Internet in small packets,  a lost packet means that the set of data contained in that packet needs to be re-sent by the server, causing small gaps in the file being played back to the user.  The amount of packet and overall quality of sound is determined by three factors: connection speed, the underlying transmission protocol, and the compression algorithm used by the vendor.

Connection Speed (Top)

Transmission Protocols (Top)

The transmission mechanism used by the vendor will also determine the smoothness of sound during playback.  The TCP protocol is based on the architectural protocol used on the Internet and is the most common protocol used to transmit large volumes of information with a guaranteed packet delivery. The packet size of the transmission  is larger than UDP or IP Multicasting, so it is more efficient with significant amounts of data than the other protocols. It also has "flow control" mechanisms which manage the utilization of  resources evenly across Internet users.

This is particularly important in the case of corporate intranets, where dense files can cause short-term data spikes that would interrupt other network functions.  Furthermore, the processing requirements of audio and video streaming can prevent the server from performing other network-related functions unless there is a mechanism to control these resources.

Because the packet sizes are bigger in TCP than in other protocols, a lost packet of data will result in a larger "hole" during playback. Also, the larger packet size means that retransmission from the server will take longer than it would with other protocols.

Unlike TCP, UDP transmits small packets of data extremely quickly.  Because of the small packet size, transmission is highly efficient.  However, since the protocol has no means of dealing with packet loss, the user's application has to resolve the gaps during playback.

In addition, since the protocol is not intended for large volumes of data, using the protocol for dense transmissions can quickly saturate the network.  Since there are not flow control systems to limit the use of available bandwidth, it is a transmission standard best used for single, point-to-point communications.

IP Multicasting also has its pros and cons. With this protocol, all IP data is delivered in the most efficient and reliable way to a large number of users.  IP works by having the server send the audio stream only once to all computers using the most effective means to transmit the data.  It delivers the best results when the same audio stream is sent to many users simultaneously.  Below is a summary of the most effective protocol for individual applications.

A collaboration between MCI Communications Corp. and Progressive Networks Inc. (www.realnetwork.com) landed Progressive's audio and video streaming technology into MCI's Internet backbone.  RealNetwork is a cost-effective, high-quality service for business - including broadcasters, content providers and large corporations - to broadcast "live" and "on-demand" RealAudio and RealVideo programming to large audiences.  RealNetwork leverages the expertise of MCI and RealNetworks by creating a mass broadcast infrastructure for large-scale delivery of multimedia content over the Internet.  As an industry first, RealNetwork offers the most complete and cost effective audio and video streaming available today.  In terms of quality, Progressive Networks has developed sophisticated algorithms which actually help correct packet losses.  The programs predicts what the lost packet information was and then produces audio similar to the lost fragments.  By "filling in the gaps," this technology makes it easier to listen to audio files even with high packet loss rates.
 

In another major industry play, Microsoft acquired VXtreme Inc., an Internet video and audio streaming start-up and simultaneously released an update of its NetShow streaming media server.  The news came just two weeks after Microsoft announced a licensing and standards development deal with and minority investment in VXtreme's chief competitor, Progressive Networks, Inc.  The purchase of VXtreme fits with Microsoft's recent work on an Active Streaming Format (.AVI) on the web.  The announcement also coincided with Microsoft's launch of its NetShow 2.0 streaming media server.  Microsoft also holds a seat on the board of still another streaming video vendor, VDOnet, which is focusing on providing video the corporate and broadband network market.

With the acquisition of VXtreme, Microsoft acquired additional technology that would allow it to take its product line and extend it by authoring once and then delivering across a range of bit rates and bandwidths to different customers.

VXtreme, which was founded two years ago, has products that include Web Theater Server, a streaming video and audio server, LiveStation, for delivering live video over the Web; Producer, Web video development tools, and Web Theater Client and Client Gold, plug-ins to Netscape's Navigator and Microsoft's Internet Explorer.

Other major video streaming players include:

- CVideoNow
- Microsoft NetShow
- RealVideo
- StreamWorks
- TrueStream
- VDOLive
- VivoActive
- Vosaic MediaSuite

Recent Applications (Top)

Also, with the recent teaming up of Progressive and MCI to create RealNetwork, the industry has a service that operates much like a cable network by leasing "channels" to content providers.  MSNBC is taking a slightly different tact with Business Video, a Web-based service that offers both live and searchable archives of video programming.

Corporations have traditionally implemented streaming video on their intranets using MPEG-1, a popular but unwieldy compression standard primarily intended for CD-ROMs. The streaming-video technologies in this roundup, on the other hand, produce much smaller data rates. Consider this: An MPEG-1 video stream has a bandwidth of 1.2 megabits per second (Mbps). That's more than 10 percent of the available bandwidth on a 10-Mbps Ethernet. With the same bandwidth you could send 19 video files at 64 Kbps or 9 video files at 128 Kbps, reaching many more users without costly hardware upgrades.
 

  
KCRW             Liquid Audio         National Public Radio
Los Angeles

In terms of strictly audio streaming products, there are a range of new products which may challenge the way consumers think of music purchases.  For instance, a firm called Liquid Audio has taken to the Web with an innovative idea for selling audio CD's.  Imagine a world where you can instantly listen to the newest songs or old classics on-demand and in the privacy of your own home.  You can move backwards or forwards through the tracks, read music reviews on the various artists and engage in chat with other consumers.  Then, once you have decided which CD to buy, you simply pay for what you like and download a copyrighted, CD-quality version directly to your personal computer.  The downloaded file will contain all the songs and information of a physical CD, as well as some possible "bonus" items such as multimedia add-ins.  This business model is precisely what Liquid Sky employs to sell its audio CD's over the Web today.

While some companies are reinventing the sales channel using streaming audio technology, other companies like California-based KCRW are using it to augment  programming.   KCRW has found that the Web is a perfect space to expand and augment shows such as its famous "Hollywood Wrap" segments.  By leveraging the new audio technology they are able to increase scope and scale of market-an important factor for a company which derives most of its income from advertising revenues.  Other well known radio shows like National Public Radio (NPR) are using audio streaming technology to extend their scope by offering custom newscasts throughout the day to a global audience.  By utilizing this new technology then, NPR is able to realize a larger market then it has ever enjoyed using traditional radio station channels.

Audio streaming technology appears to be one of the few multimedia add-ins that actually is both profitable and popular based on the success of such products as Progressive Network's RealAudio.  The primary reasons for the successful entry of streaming audio appears to related to quality and scope.  While streaming video struggles with issues of speed of transmission, streaming audio can accommodate data-transfer rates at the snail-like pace of 9.6Kbps.  And in terms of quality, unlike video which is prone to poor clarity and timing differences, streaming audio products have successfully developed algorithms which self-correct for lapses and imperfections.  As a result, streaming audio is able to meet a larger global market of existing Web users with a high level of clarity.  Therefore, the adoption of this technology by both users and consumers continues to grow despite the lagging success of video.

Access to video content is prevalent on the Internet.  Similarly, the cost and availability of video streaming packages does not present a barrier to adoption. Individuals and companies alike can simply purchase or download the software packages from the major vendors in the marketplace.  However, the adoption of video streaming technology has not been as rapid as many providers have hoped.

Present limitations, such as bandwidth speed, have hindered companies from investing in video streaming at the level anticipated. In addition, integrating video and audio content into the web page itself has proved to be a challenge for many buyers. The cost of preparing and delivering multimedia to a broad, cross-platform audience has also been an obstacle preventing widespread adoption.

On a more basic level, the incompatibility of many plug-ins and players have prevented some from reaping the benefits of the latest video technology.  However, this factor may soon prevent less of an obstacle with the attempt by many industry players to create a common standard.
 

Future Developments (Top)

The future landscape of streaming technology will include a more complete synthesis of video and audio applications.  Along those lines, the industry direction will more than likely be determined by the major players in the industry: RealNetworks and Microsoft.  RealNetwork will leverage itself in the industry as the premiere technological breakthrough in Internet broadcasting.  MCI and RealNetworks will offer a service that will for the first time deliver audio and video to large audiences over the Internet by combining MCI's Internet backbone with RealNetwork's leading multimedia streaming technology.  (Prior to the formation of RealNetworks, Progressive's RealAudio held a 90% share of the Internet streaming audio market).

Microsoft, as previously discussed, has dealt itself into the video streaming game from several angles.  Through its acquisition of VXtreme, it will be moving toward incorporating several streaming technologies into its present line of products.  Microsoft also hopes to establish the industry standard file format for video streaming with its Active Streaming Format, while simultaneously building support for its NetShow client.  Microsoft believes that end users will be able to use NetShow as their single client-side player to view just about any form of streaming content on the Web.  However, if Microsoft is wrong in its predictions, it will still benefit via its minority investment in Progressive, which is now RealNetworks.

Last December, many major hi-tech industry players banded together to support the development of a common standard for streaming media. These companies, which include Hewlett-Packard, IBM, Sun Microsystems, Cisco, and Apple, are supporting Real Time Streaming Protocol.  RTSP is an open standard for delivering real-time multimedia over the web and industry observers believe it may become as prevalent as the HTTP protocol is over the web.  The promotion by Microsoft of its own standard may end up delaying the adoption of the RTSP protocol as an industry standard.

RealNetworks has also done some things to ensure its "cooperative competitiveness" with Microsoft.  RealNetworks will be able to support Microsoft's ASF in the next release of RealVideo and RealAudio players.

Although RealNetworks and Microsoft are cohabiting, Oracle Corp. has a different agenda and is unlikely to be integrated.  Oracle's Video Encoding Standard (VES), a video-server application programming interface that will likely go head-to-head with Microsoft's ASF.  By adopting VES, video encoders will be able to store digital video directly on a video server, independent of any other type of encoding standard.
 
Audio and video are poised to move onto the Internet the same way images did because of better coding and compression schemes, and increased bandwidth and processing power at the desktop.

Bibliography (Top)

Gonzalez, Sean   "Sound Foundations: Audio on the Web"  PC Magazine Online  March 26, 1996
Frentzen, Jeffrey  "Sound-enabled Web Sites Strike Chord"  PC Week Online  January 29, 1996
Ozer, Jan  "Audio On Demand" Computer Shopper Online January 1996
Ozer, Jan  "Sound Blasts the Web" PC Magazine Online March 26, 1996
Moeller, Michael "Real Audio Makes Real-Time Waves"  PC Week Online October 30, 1995
"Standards for Streaming Media"  (author unknown) PC Magazine Online  December 17, 1996
Memom, Farhan   "Audio Streaming Technology Flows Online"  Interactive Week December 2, 1996
O'Brien, Jim  "Hot Off the Wire: Audio On-Demand Primes the Web for Profit" Computer Shopper  July 1995
Dziatkiewicz, Mark, America's Network, "System promises real-time video mail," 11/15/95.
Glascock, Steve, Computer Reseller News, "Streaming audio, video on the Net spurs alliances," 8/11/97.
Jones, Chris, InfoWorld, "Director plug-ins to allow live audio, video streaming on Web," 3/18/96.
Kirkpatrick, Keith, Computer Shopper, "Microsoft Unites Streaming Video Players Behind ASF," 11/97.
PC Magazine, "Streaming Video," www.zdnet.com/pcmag/features/stream/, 10/7/97.
Plain, Steve, Computer Shopper, "Streamlining," 12/96.
RealNetworks, www.realnetwork.com, 10/25/97.
Senna, Jeff, InfoWorld, "VivoActive offers server-independent video and audio streaming," 8/4/97.
Tedesco, Richard, Broadcasting & Cable, "RealVideo debuts this week," 2/10/97.
Ubois, Jeff, Midrange Systems, "Radio and TV on the 'Net," 3/29/96.
Wirbel, Loring, Electronic Engineering Times, "Design - Computers & Software," 10/5/92.