Upcoming Industry Conferences
Streaming Media West [13-14 November 2018]
Live Streaming Summit [13-14 November 2018]
Streaming Forum [26 February 2019]
Past Conferences
Streaming Media East 2018 [8-9 May 2018]
Live Streaming Summit [8-9 May 2018]
Content Delivery Summit [7 May 2018]
Streaming Forum [27 February 2018]

25 Years of Internet Radio, Part 2
Back when online video was the size of a postage stamp, audio companies were changing the way the world heard radio. Here are reminiscences from RealNetworks and other pioneers.
Learn more about the companies mentioned in this article in the Sourcebook:
{0}

In Part 1 of this article, we spoke with some of the unheralded innovators from the early days of internet radio. In Part 2, we look at the best-known (and arguably most important) company responsible for launching the internet radio revolution—RealNetworks—as well as speak with a few insiders about the state of internet radio today.

No story about internet radio would be complete without a focus on RealNetworks and its audio format, RealAudio. RealNetworks (Real, for short) emerged during the mid-’90s as the web was becoming more consumer-focused and was the primary enabler for the general public to engage with early streaming. In fact, I used RealAudio for my first webcast from the Ministry of Sound in London in October ’96.

For consumer adoption, Real was the key breakthrough technology and for a very simple reason: It had a simple and reliable solution, and you could play with it for free.

While many other technologies were emerging, few had the focus that Real did in terms of getting its technology out and in use as the player for any internet audio download and for streamed internet radio. It had a nice, clean install-and-stream server model which it licensed, and many early broadcasters had relatively high-capacity internet connections, as well as internal local area networks (LANs), so they could get up and running quickly.

RealNetworks: Dave Mallinson and Rob Glaser

I caught up with Dave Mallinson, who was involved with Progressive Networks, which became RealNetworks in 1995. Here are his comments, lightly edited:

In most cases, the early days involved customers finding some old Pentium PC with 8MB to 16MB memory running Windows 95/NT with a sound card in it and installing the RealAudio Encoder software on it. They then connected an audio cable to the mini-jack input on the sound card with the other end often connected to a radio receiver, which played back the broadcast radio station that everyone was listening to. Instead of the radio receiver taking the audio out to speakers or headphones, it went down the cable and into the PC’s sound card, where the encoding software could read the audio signal and prepare it for conversion into a live stream.

Content owners that took their audio seriously would often go for some form of Unix-based solution (Solaris, SPARC, SGI [Silicon Graphics, Inc.], Linux PC) or a Power Mac, as they didn’t want the frequent reboot of Windows! They would also look to receive the audio input prior to broadcast (though this was tricky in most cases, as the teams running the internet broadcasts weren’t seen as core).

The trick was picking the right output encoding profile, as it impacted the audience that had potential to listen; too low bitrate [and] the audio quality was too poor for music and no one (other than the dedicated music follower) would listen. Too high bitrate, and it may [have meant] that the audience that wanted to listen couldn’t as their network wouldn’t allow it. There was no such thing as adaptive bitrate or stream switching—you either could listen or you couldn’t.

Once the content was encoded, the output live stream was often sent over dedicated ISDN connections (or potentially the corporate LAN) to the streaming server that was centrally hosted in a data center somewhere with ‘lots of bandwidth’ (compared to what’s needed today, that’s laughable!) to then distribute to any end user clients that would request it.

Having spoken to Mallinson, I simply had to reach out to an old friend of Streaming Media—Rob Glaser, the founder of RealNetworks—and again asked him for a few reminiscences, as well as a comment on how internet radio has emerged since those early days. Again, his comments have been lightly edited.

When I was 12, I went to summer camp, [and] my camp counselor was a DJ at a station in a college in New York. I was fascinated by his stories and went to see his station and was very excited. I grew up on AM radio in New York—it was a window to the world, a ‘theater of the mind.’ I could get signals from up and down the East Coast. So I loved radio, and met someone who was creating radio.

So in ninth grade I started a station at school. We couldn’t really sort out a transmission license, so we set up the station on the intercom and ran it a few hours a day 5 days a week. We called it WECS, and that experience remained with me and was always in my mind.

I started at Microsoft in 1983 and was there for 10 years. I was program manager for numerous projects such as early versions of Word and Access. I learned a lot about networking while looking after some network products, and then in the last 3 years, I worked on multimedia and consumer systems, focusing on the media and creation side of the sector.

Then, having worked in numerous pieces of the puzzle I decided to step away—initially as a leave—and then in summer 1993 I was invited by my friend Mitch Kapor (who was one of the founders of Lotus) to an Electronic Frontier Foundation [EFF] meeting. There I met Dave Farber from the University of Pennsylvania (he worked on a lot of the early network protocols), and he showed me a pre-beta of NCSA Mosaic.

I saw that and saw the future.

I returned to Seattle, worked out a way to get a dual-bonded ISDN line at home— from Halcyon—and started to observe everything going on with the explosion on the web. I remember seeing two new websites a week become two a day become overwhelming growth.

I thought that the silo approach wasn’t going to scale. The web, plus DNS [Domain Name System] and URLs, was creating something incredibly scalable.

IMG tags appeared, making the web look better, and I thought, ‘What if we can do the same for multimedia?’

But I also thought, ‘It will take years if we start on video’ (because of the bandwidth). Given dial-up adoption was typically around 14.4Kbps, I thought voice-quality audio would be a good starting point. So in April 1995, we made an on-demand voice coding solution. We knew we wanted to add music and live in due course, but the MVP [minimal viable profit] was on-demand voice.

We hit the timing brilliantly—we were the first solution to offer reliable streaming. We almost launched it supporting TCP [Transmission Control Protocol], but we tried it over long distances, and we hit windows sizing issues. So we tried UDP [User Datagram Protocol], and we needed to build our own control protocol, and we made it tolerate 20 percent packet loss. Internet World and NAB were coming up (on the same week), Mitch and his friends at EFF were keen on free [distribution], and so we launched at both with a free download of the player and encoder and a limited download of the server.

We also launched with ABC radio network—headline news on the hour—using a satellite dish delivering the headlines on-demand. And we did a similar show with National Public Radio.

So when we launched, we launched with useful content to people. Every time you launch a platform, there are hundreds of possibilities for what you can offer, but launching with a useful offering made all the difference!

In the first week, our teams of 15 to 20 folks in Seattle were trying to support people all over the world, and yet because we didn’t over-hype it and because the product worked, we survived this explosive take-off.

So about 5 months later, we streamed a major baseball game, which was our first major live [event]. But we had a bug, where the player would crash after about 2.5 hours, which was a pain because the game lasted 2 hours and 45 minutes! So anyone who had joined at the beginning hit this bug! However, folks had seen a good show to that point, and we fixed the bug promptly after the game, but it was a good example of how you can never test everything!

What are my thoughts now, some 20-plus years later, about how internet radio has transformed the broadcast landscape?

It made everything global. Everyone could be a broadcaster, and everyone could listen. It took radio out of the niche. It has dramatically exploded the enthusiasm for broadcasting.

Then there was all the commercial disruption. It used to be that you could lock the market for a given territory. Internet (and satellite) has changed that forever. That’s great if you take advantage, and yet on the flip side, the commercial broadcasters have found it a major challenge.

And social media has created a feedback loop of echo chambers. For me, internet radio exposes you to so many choices. I lament the fact that modern social media distribution systems restrict things as a mindshare matter (not a technical matter). Somehow the social networks have tribalized us, which is a shame, since the proliferation of internet radio offers us so much more choice than we had before.

By the early 2000s, Microsoft was competing hard with Real and had established the Windows Media ecosystem. As domestic bandwidth increased, its attention was dominated by battles for video eyeballs, and so its innovation in the internet radio space slowed down. This gave space for Icecast and SHOUTcast to demonstrate a wider cross-platform flexibility than both Real and Windows, and this suited quirky early personal digital assistants (PDAs, such as the Palm Pilot) that could connect to IP and had MP3 or open source decode codecs. It was in this era that internet radio and live audio streaming began to go wireless, and eventually cellular.

Yes, Real and Windows were quick onto those mobile platforms too, but in that time, SHOUTcast and Icecast had also established themselves. Not least because PDAs and palmtops quickly added MP3 player feature sets, meaning the step to enable live streaming was small. In 2001, I demonstrated a live Icecast radio stream (of a friend DJ’ing drum and bass back in Brighton, U.K.) to the U.K. prime minister’s office using an HP Jornada and a Nokia phone connected on general packet radio service via infrared. So even 6 years before the iPhone, mobile internet radio was just about possible!

While the Windows-based WinPlay3 predated it, the multiplatform Winamp revolutionized things when it was introduced in 1997. Not only was it widely adopted because of its ability to instantly decode and play back an MP3 on-the-fly, but Winamp has a simple plugin architecture, and there was a plugin that—to this day—allows anyone to stream to a SHOUTcast server. SHOUTcast servers are free and easy to set up, and you can always announce your SHOUTcast stream on the shoutcast.com service. SHOUTcast itself does not, however, actually distribute the streams, but the aggregation of links to channels on shoutcast.com offers a simple route for the consumer to discover streams. This made SHOUTcast a perfect platform for ultra-niche streamers producing shows for very specialist audiences. SHOUTcast is still up and running despite several changes in ownership. I checked while writing this article, and there are today more than 88,000 live streams available on shoutcast.com. Indeed, shoutcast.com is often thought of less as a software and technology provider and more as one of many places to promote a stream.

Behind the scenes, there are a number of other active audio CDNs, including Triton Digital, which has been integrating with multiple ad insertion companies, and Pandora, which itself recently acquired Adswizz. Along with Apple Radio and Spotify, those are certainly currently the significant players in this space today. But they are not, by any means, the only technology providers active.

SharpStream: Dane Streeter

An episode of smadvancedforum.com with Dane Streeter and Paul Riismandel turned me on to writing this article. I caught up with Streeter to find out more about how his company SharpStream is snapping at Triton’s heels. His comments, lightly edited, appear below:

During my 12 years in the audio space, the marketplace has seen significant and steady growth. In 2007, consumption of streaming audio was touched by approximately 15 percent of market users; fast-forward to today, and that number is over 55 percent and growing every day. To put that into CDN context, SharpStream’s live audio distribution has grown from 2,000 concurrent listeners in ’07 to in excess of 150,000 today. The growth of on-demand audio (podcasts and listen again services) has arguably happened at a faster rate than its live counterpart; however, this seems to be dependent on the format and listening preferences of any given territory. The Swedes enjoy podcasts over live broadcast, yet the UK is the opposite. I suspect the difference in corporate and social cultures goes at least some of the way to explaining the listening habits in any given territory.

Just as with video, the industry has seen the rise and fall of a variety of different core encoding and distribution technologies. Where once upon a time Windows Media Audio was the de facto option for audio consumption online, the launch of mobile technology forced the industry toward codecs and protocols supported on mobile browsers. It goes without saying that the most significant of these changes came about during the first few years after the launch of the iPhone.

Related Articles
In the latest installment of our do-it-yourself series, we'll look at how to set up an Icecast server on an AWS cloud instance.
We're putting together a deep dive into the past, present, and future of internet radio, and we need your help. Take our internet radio survey and enter a drawing for a $50 Amazon gift card.
Can you believe it's been a quarter of a century since the first internet radio broadcast? In this first installment of a two-part series, we talk to some of the trailblazers who started a revolution.