Script: Streaming media audio primer

Since its debut in the mid-1990s, streaming media has been considered just a way for entertainment companies to deliver music and videos to consumers. But many analysts and vendors believe the real potential for streaming media is in the enterprise, where plentiful bandwidth can deliver corporate-communication messages, training and sales presentations to a wide audience.

In this primer, we'll explain how the technology works, including the benefits and potential pitfalls when dealing with streaming media.

The key benefit of streaming media is that an end user can begin to view the content almost as soon as he brings up the appropriate player. This is in contrast to the early days of the Web, when users had to wait to download an entire multimedia file before they could listen to or view any of it.

Corporations with geographically dispersed workers can use streaming to broadcast a CEO address to the entire company. For those not able to view it live, they can watch it on-demand at their leisure. Publicly-held companies use streaming to help satisfy government full-disclosure regulations, which require them to make public any comments by executives that could affect the company's stock price. A growing number of companies use streaming media to better explain their products to potential customers.

Currently, the major streaming media formats are RealMedia from RealNetworks, Windows Media from Microsoft, QuickTime from Apple and MPEG from the Moving Picture Experts Group. MPEG comes in four flavors. The best known is probably MPEG3, or MP3, which is used extensively for music files. The newest version, MPEG4, is designed for TV-quality video at the high end and for limited-bandwidth video for wireless handheld devices.

RealNetworks and Microsoft have the most users, but QuickTime and MPEG do have their own niche markets. RealNetworks, Microsoft and QuickTime all use proprietary technology for encoding streams, requiring customers to use the same vendor's server, encoding and player technology. Fortunately, all three offer free versions of their player software, eliminating the cost for end users. MPEG's major benefit is that it is a standard, so an MPEG-compatible player from Company A could receive an MPEG file encoded with software from Company B.

Whether audio or video, the streaming media process takes four basic steps: capturing, editing, encoding and delivering the content. Encoding and delivery have a bigger impact on networks, so we'll cover that first.

Each format can be delivered directly from a standard Web server using the HTTP protocol. However, when using a server built specifically for the format being served, such as RealServer or Windows Media Services, content providers can take advantage of special protocols and features that help maximize the viewer's experience.

RealServer can run on Unix, Linux and Windows-based servers with a 25 simultaneous-stream version of the server available for free. Microsoft offers its unlimited user streaming server free as part of the Windows 2000 platform.

Microsoft uses its Microsoft Media Server (MMS) protocol for delivering between its server platform and proprietary player. Likewise, Apple and RealNetworks use a combination of the Real-time Streaming Protocol (RTSP) and Real Time Protocol (RTP) in their implementations. One caveat with using MMS, RTSP and RTP: Firewalls with restrictive settings will sometimes block this traffic from entering a LAN.

Streaming media can be delivered live or on-demand. In either case, the more bandwidth available for the stream the better the end-user experience. Audio and video packets must be delivered in the right order and at a regular interval or there will be degradation in quality. Network congestion and dropped packets are the equivalent of static in a radio broadcast.

To help reduce degradation caused by network congestion, most streaming media players will buffer content locally before beginning to play. Even live video is buffered, which can cause a time delay of 30 seconds or more between an event and the streamed content. Many content providers using QuickTime deliver streams via progressive download, where the content is partially downloaded before it begins playing.

In the case of live or on-demand, streams can be sent using Unicast or IP Multicast technology. Unicast delivers one stream to each client that requests a piece of content. So if 10 users select a 100K bit/sec stream, the server must send out 1M bit/sec of data. When hundreds of users try to access a stream from a single origin server, the server and its network connection can become overloaded. Unintended denial-of-service attacks are often prevalent in large live broadcasts.

In an IP Multicast environment, a piece of content is "broadcast" across a network using a single stream. Users can "tune" into the broadcast to watch. This saves bandwidth, but requires that the entire network path be IP Multicast enabled. This is possible in a controlled corporate LAN but nearly impossible when streaming across the Internet. For more on this topic, click here.

Distributing streaming media content closer to the end user can help ensure a good experience by limiting the number of network hops between the server and end user. It can also save having to serve from a single source every user requesting a piece of content. Companies can build their own content delivery network with caching products from the likes of Network Appliance and CacheFlow, or the can outsource it to third-party streaming provider such as Akamai or Activate, which have global networks.

In addition to buffering, streams can be compressed or "encoded" to reduce their overall size, by reducing the maximum number of bytes that can be streamed per second.

For example, in its raw form, the primer you're now listening to might be as large as 75 megabytes. If you're listening to this in Windows Media format, we've used a delivery rate of 7K bit/sec, which results in a total file size of roughly 300 kilobytes. If you're using a RealPlayer, you're getting a stream of 32K bit/sec, which means a total file size of about one megabyte. A key caveat is that the greater the compression, the lower the overall quality of the sound or images. A simple audio interview or primer such as this is a much better candidate for heavy compression than, say, a video feed of a sports event.

RealNetworks and Microsoft have adaptive server technologies that allow the delivery speed to be increased or scaled down to compensate for network congestion. As an alternative, some content providers create separate files for each level of bandwidth and then let the end user select the appropriate stream.

With each new release, the encoder creators develop better algorithms for compressing content. Soon, what can only be delivered at 300K bit/sec today will be available at 100K bit/sec.

Streaming media files consume much more disk space and bandwidth than standard HTML Web pages, requiring a robust server platform, with plenty of disk space and a T-1 or better connection to the Internet. For those with a minimal Web infrastructure, using a hosted provider such as Akamai, Activate or Yahoo Broadcast could save money. Most of these providers offer large storage quotas for a minimal monthly cost. Just be careful of how the streams are priced - is it per megabyte served or per stream. If you have all high-bandwidth files being served for minutes at a time, the per-stream option may be cheaper.

Before you can stream, you'll need to create the content. A key part of this is known as capturing. This involves converting an audio or videotape into digital format or using a recording device to create an uncompressed digital version. Using special hardware, a PC can convert the analog signal of a VCR or cassette player into a digital format that can be stored on a hard drive for later manipulation. In the case of the Network World Audio Primers, we use a microphone connected to the standard mic jack on a Windows-based PC running Sonic Foundry's SoundForge to capture the narrator's voice into an uncompressed WAV file. For video projects, we use a digital video camera that hooks directly in to the PC via an IEEE 1394 or Firewire card. Firewire is a standard feature in Macintosh computers and PC users can pick up a Firewire card for under $100.

Once captured, editing tools such as SoundForge or Ace of WAV from Polyhedric Software can be used for audio-only content and CineStream from Discreet, Apple's FinalCut Pro or Sonic Foundry's Vegas Video can used for video content. Ace of WAV can be purchased for as little as $45, while the video editing packages are priced around $500.

Products such as Discreet's Cleaner can be used to encode raw and uncompressed audio and video into all the major formats including Real, QuickTime, Windows Media and MPEG. Companies with high-end post-production equipment can feed content directly into an encoder.

Streaming media can help lower training costs and get a message out to a wide audience with a richer experience. Both Real and Microsoft offer some free versions of their encoder tools that can let a beginner get their feet wet before making big investments in the technology.

For more on streaming, visit our streaming media research page.

Back to the main Streaming Media Audio Primer page

RELATED LINKS