It seems everyone is interested in streaming media these days. Many events have recently been broadcast on the Internet though what has come to be called Webcasting. There have been interesting events from the field of medicine, including the birth of a baby (http://www.abcnews. com/sections/living/DailyNews/netmom980610.html), and open heart surgery (http://207.87.15. 72/liveevents/openheart/f_main.htm). NASA generated quite a crowd with the Mars Pathfinder Mission (http://mpfwww.jpl.nasa.gov/). Rock concerts, auto races, and various marketing events, like the launch of Windows 98 have drawn a lot of viewers. Several celebrities, including the actor best known as Captain Kirk, have recently discovered the Internet as yet another outlet for their marketing and personal promotion efforts. More and more, the web is being seen as an additional channel for the broadcast of events captured in digital audio and video.

According to Multimedia Research Group, Inc. (1998) of Sunnyvale, CA, and Fuji Keizai USA, approximately 4,000 Web sites offered video clips in 1996. That number tripled to 12,000 in 1997 and is expected to triple each year for at least the next three years. While this rate of growth is impressive, those 12,000 sites represent less than 1% of the sites on the Web.

Three trends are converging that will speed the adoption of these robust forms of multimedia communications on the web. First, compression techniques and algorithms continue to improve, enabling increasing audio and video quality at decreasing bit rates. At the same time, networking technologies are developing at a dramatic pace, with nationwide initiatives like the Next Generation Internet (NGI) and the Internet2 (I2). People involved with these collaborative efforts are predicting bit rates 100 times faster that today’s Internet to be widely available by 2001, with some users experiencing bit rates 1000 times faster (Grush, 1998). Finally, the digital capture and editing of audio, still graphics, and video (cameras, hardware, software, etc) is now within the reach of the mass market, both in terms of price and ease-of-use. The quantity and quality of multimedia created is about to increase dramatically. The convergence of these three trends will create additional capacity for high quality audio and video communications on the Internet, and simultaneously, additional demand.

Why would educators, trainers, and instructional designers be interested in this technology? Research has shown that methods of communication utilizing multiple sensory channels increase both learning and retention. We simply learn and remember more of what we see and hear than we do of what we read, or see, or hear alone. While computer aided instruction (CAI) has long benefited from this principle (Paden and Barr, 1980), the Web was unable to support the bit rates necessary to enable these kinds of media. Video, and to a lesser extent, audio files were simply too large to transmit via the Web.

Streaming media technologies promise to extend these multimedia communications channels to the Web. Streaming is a technology that allows the client and server to communicate and determine the amount of buffer needed so that a file can begin playing before it is completely downloaded, and finish downloading as it plays. This method of timing the delivery of data simulates real-time delivery of multimedia. Some of the predictions for the growth and availability of network bandwidth may make this technology may appear to be stopgap in nature, but in the absence of unlimited bandwidth some form of digital compression will always be necessary for the long-distance transmission of audio and video. Software that measures and compensates for the variability in network connections will be necessary well into the future.

At the time of this writing, both Microsoft and RealNetworks have Webcasting solutions that let you encode a live video stream (with audio) on one PC and hand that stream off to another PC (a media/Web server) for consumption by anyone visiting the site hosted by that Web server. In other words, as long as a user has the proper client software installed correctly, she can come to your site with Netscape Navigator or Internet Explorer 4 and see a live or previously captured audio and or video event embedded in a Web page. All you need to provide these live or archived events to your learners is a video camera, a video capture card, and two PCs--a dedicated capture machine, and a streaming media server/Web server. Figure 1 below is the recommended configuration, but a single machine can serve to encode either live or stored content, and a separate single machine can act as both NetShow and Web server.

Figure 1: Recommended NetShow Configuration

As recently as last year, streaming media was a very complex proposition. There were several companies with competing products all of which provided similar results (Johnson, 1997). Today, users still have to choose a file format, but the formats are distilled to just a couple of major choices. More importantly, the quality of the streaming media products available has significantly improved. The highly fragmented streaming-media market that included products from Vosaic, Xing, Vivo, VDONet, and VXtreme among others in 1997, has matured into a race between Microsoft’s NetShow Services and RealNetworks’ G2 technology (Ozer, 1998).

From the users’ perspective, it’s an easy choice. There is a rapidly growing body of streaming-media content available using each of these products. Microsoft’s and RealNetworks’ products can both automatically download playback components for new technologies as they emerge. RealNetworks uses this feature to support AVI, WAV, MIDI, and MPEG-1, but not NetShow files. Media Player supports more formats, including files created using RealVideo through version 3.0, but Media Player can’t play RealSystem 5.0 or G2 files. Web surfers should download and install both players since many sites post their media in only one or the other format.

For a developer, choosing a development platform is a more difficult decision. As I approached this decision, I identified four criteria as important: The availability, reliability, and compatibility of the necessary client technology; the relative quality of the product produced; the probable longevity of the platform; and the cost of the total (client/server) solution.

The competing firms have adequately addressed the first criterion. Each of the competing products is readily, almost ubiquitously, available. In my testing, when properly installed and configured, both were equally reliable. Neither or the client ‘players’ conflicted with the software installed on the test machine.

Quality is a difficult construct to measure. Although I perceived some quality advantages, the experts report that the products cannot be differentiated on audio and video quality. "Though one technology may prove superior in limited experiments, tested across a range of video clips and bandwidths both offer equivalent audio and video quality" (Ozer, 1998, p. 36).

RealNetworks’ recently debuted G2 product introduces Synchronized Multimedia Integration Language (SMIL). SMIL is an XML-based language that defines how to create integrated presentations combining linked text, audio, video, still images, and Flash animations. It also enables pans, zooms, fades, and transitions between these elements. Microsoft supports streaming JPEG images synchronized with audio but without special effects. NetShow’s key strength in this area is the ability to produce streaming content from PowerPoint presentations ‘annotated’ with audio.

While RealNetworks’ basic audio and video encoders are easier to use than Microsoft’s, RealNetworks has yet to deliver easy-to-use SMIL encoding tools, leaving SMIL content creation the providence of experts. Still, for those developers seeking the most compelling presentation of video at 28.8 Kbps or the integration of diverse forms of streaming media, G2 with SMIL is currently better qualified.

From a broad field of competitors a year ago, two products remain. Microsoft’s NetShow Services and RealNetworks’ G2 technology. Many people have wasted time and effort creating educational multimedia content that, due to obsolescence of the development platform, is now essentially unusable by the majority of web clients. The probability that media developed using the chosen platform will remain valuable over time is an important consideration. Both of these companies have demonstrated a commitment to the continued advancement of their streaming-media platforms. RealNetwork was an early leader in this field, and maintains a loyal following. There are over 200,000 hours of live content broadcast weekly using RealNetworks software, and more than 31 million players have been downloaded. Microsoft’s popularity is demonstrated by the adoption of the phrase often used in the past to describe hardware purchasing decisions, "No one was ever fired for recommending IBM" has become, "No one was ever fired for recommending Microsoft."

The advantage in the cost of the total client-server solution clearly belongs to Microsoft. Free is always good--NetShow is free. The client software, Windows Media Player, is free and replaces the Media Player installed with the Windows 9X or NT operating System. The NetShow Tools software, used to create and edit the streaming media files, is free. Even the NetShow server software is free, assuming that you are already running Windows NT Server 4.0. If not, the retail cost of NT Server 4.0 is $385 from beyond.com.

The RealNetworks solution is not free. The limited client, RealPlayer G2, is free, but the complete client software costs $29. The full-featured content production package, RealProducer G2, costs $149.95.

Serving media using the RealNetworks products can be costly. The Classroom Server Package, designed and licensed specifically for educational use is $1,895 for 50 concurrent users. You’ll probably want to add free server upgrades and phone support - the first year is $495; subsequent years are $795. It comes with a 50-seat license for the RealPlayer Plus software. If you think you’ll need to support more than 50 concurrent users, RealNetworks has the following options; Internet Solution I, 100 Concurrent Users, $5,995, with support and upgrades at $2,395, total $8,390; Internet Solution II, 200 Concurrent Users, $11,195, with support and upgrades at $4,475, total $15,670; Internet Solution III, 400 Concurrent Users, $21,595, with support and upgrades at $8,635, total $30,230.

A colleague of mine pointed out that the $29 cost of the RealPlayer Plus is hardly a burden for most web surfers, so I asked him to fund it’s installation on just the 1,200 machines my students typically use. Placed in the context of growing demand and stagnant budgets, even the extremely reasonable cost of the RealPlayer Plus is a significant disadvantage.

For many Web site developers, and most distance learning content-creation efforts, NetShow’s free tools for streaming audio and video should suffice. It is my opinion that RealNetworks’ current advantages, the ease of use of their encoder and the use of SMIL to coordinate multimedia streaming, are likely to be short-lived. I expect the Microsoft product line to incorporate the most popular of these features within the year. In fact, the beta version of a new product created to address some of these concerns, Windows On-Demand Producer, became available after I finished researching this article (Windows Media, 1998). Using the criteria described here, I chose Microsoft’s NetShow Services.

Having made the choice, let’s examine the types of streaming media one can create using NetShow Services. While others use more and different classifications, I contend that there are only three basic types of files you can produce using any of these products; Video, Audio, and Synchronized Audio. Video and audio files may be streamed either live, as the event occurs, or from an archived file saved to the NetShow server. The process of creating synchronized audio files makes them suitable only for archived streaming.

Video files are the newest, and get all the attention. From a basic perspective, video files include both an audio signal and a series of images played back as rapidly as possible. While new compression and decompression (codec) techniques included with streaming technologies make these files much better than was possible even a year ago, with greatly enhanced audio and much better video quality, the resulting video experience is still no where near VHS tape. The typical streaming video is small (160X120), and the frame rate is typically between 8-15 frames per second (fps). While video compression techniques have improved, and we can now make tradeoffs between the clarity of the images displayed and the frame rate, it is the audio quality of these video files has experienced the most dramatic improvement. Software developers have come to realize what many of us already knew. In the complete multimedia experience, audio is often more important than video.

Audio files have been available longer. They are inherently smaller in size, and compression techniques for audio have been found that work very effectively for each of the different types of audio. A symphony orchestra has more variability and requires a much greater dynamic range than does a single person speaking. Codecs have been developed to address these differences.

Synchronized Audio files are not really new in concept. I suspect that most of us have experienced an annotated slide show using a 35mm slide projector and a cassette tape. With the advent of client-side processing in web browsers using scripting languages like JAVA or VBScript, these types of media presentations became possible on the web. In effect, an audio file is streamed to a client and embedded in that audio file are commands that cause other types of media to be downloaded and displayed on the client machine. These images or text messages can be embedded in the stream and rendered in the MediaPlayer or they can be "URL flips," or commands that cause an image or html coded page to load to a target frame in the client browser.

While perhaps not perceived as state of the art, Synchronized Audio files have several attractive characteristics for distance education and training environments. In terms of the necessary bandwidth, illustrated audio presentations are a step up from streaming audio and at least one step below streaming video. If you anticipate that your target audience will have lower bandwidth capabilities and your content will benefit from illustration, then synchronized audio may be a viable solution. Even when you can determine that your learners have access to higher levels of bandwidth, you may choose to use higher-quality, less frequent, still images rather than poorer-quality video. If you remember that a second of broadcast quality video is really a collection of 30 images shown in a rapid sequence, then the question becomes, do you really need all those pictures?

This is not a rhetorical question. One of the fundamental principles of instructional media design may be obvious, but I think it bears repeating. Each aspect of instructional media (graphics, sound, video) can aid learning, it can be neutral, or it can detract from learning. Although it is very easy to become enamoured with the gee-whiz aspects of motion video at 30 fps, the focus must be its effect on learning. In my experience, and in most of the settings in which it’s employed and evaluated, visual motion by itself has little effect on learning (Reiber, 1994). In some specific cases, such as when the underlying instructional objective includes motion (e.g. Newton’s Laws of Motion), or when used to direct the attention of the learner, putting visuals in motion does contribute to learning and retention. I suggest that you use motion only when you believe it will contribute to your learning objectives, and not just because it’s technologically possible.

Once you’ve selected an appropriate delivery media for the transmission of your instructional material, it’s time to get the hardware and software configured to create your streaming file.

There are two primary components of NetShow Media Services. NetShow Tools includes the software necessary to create and edit ASF files, and NetShow Server adds a service to Windows NT Server 4.0 that serves, or streams, these ASF files.

Setting up NetShow Tools: 1. I chose a Windows95 machine to accomplish the encoding function. This happened to be my laptop machine, a Gateway Solo 9100 (PII 266/128MB of RAM). The recommended minimum hardware configuration for Win95 is a P133/32MB RAM. I also installed the software on a Dell Optiplex GX1 (PII 400/192MB of RAM) desktop computer running Windows NT. When not encoding ASF files, these machines do all the others things they’ve normally done.

2. The laptop shipped with a videocapture card (one of the reasons I chose it). I installed the Microsoft recommended Winnov VO Video Capture Card in the desktop machine. This card is getting very popular in live streaming environments and is giving the other recommended cards (Intel Smart Video Recorder III, Osprey-100) some serious competition at a lower price.

3. I installed the NetShow Tools software, downloaded from the Microsoft Web site, on both machines. Following this installation, I connected my camcorder to each machine, then ran the encoder from the Start/Programs/NetShow Services/NetShow Encoder menu. In both cases, the Encoder found the video source without incident.

Configuring the NetShow Server: 1. Select an NT Server 4.0 machine as the NetShow Services box. Win95 does not support the NetShow Server Services. The NT 4.0 server in my office (PII400/128MB of RAM) also had IIS 4 installed and had been running as a Web server. Microsoft recommends running separate machines for these two services (IIS and NetShow), but supports running both on the same machine, and I’ve not experienced and difficulties thus far. I recommend locating the NetShow server and the Encoder machines on the same LAN when possible.

2. Install the NetShow Services software by executing the NSSEVER.EXE installer, downloaded from the Microsoft Web site. Following the installation, select Start/Programs/NetShow Services/NetShow Administrator.

3a. If you want to serve archived ASF files, you’re practically finished. Place the encoded ASF files in the proper "root" directory (default is C:\asfroot) and they’re ready.

3b. If you want to serve or broadcast live ASF streams, there is another step. Create a Broadcast Unicast Publishing Point (easy to do via a NetShow Administrator wizard) linked to your NetShow Encoder machine. This publishing point remains active after you close the NetShow Administrator. The wizard also creates HTML code for use on your Web server. Performing these three basic steps engages the NetShow Services machine in the archived or live Webcasting operation.

NetShow Services ships with the tools listed in Table 1 below. The NetShow Tools are installed on the machine that takes the audio and video signals, digitizes and compresses them, and provides them to the NetShow Server for distribution. Using these tools, either alone or in combinations, you create and or convert streaming content (NetShow Services, 1998).

Table 1

NetShow Tools