Mark Grimshaw; Louisa Yong; Ian Dobie | "Boom Booom Net Radio" |
Abstract.
Internet radio is one of the growth areas of the Internet but, as this
article will show, is fraught with difficulties and frustration for both the
modestly-funded broadcaster (bitcaster) and the listener. The article will illustrate some
of these problems by means of a short case study of an existing Internet radio station;
Boom Booom Net Radio (www.bmbient.demon.co.uk).
Whilst necessity dictates some use of technology-related terminology, wherever possible we
have endeavoured to keep such jargon to a minimum and to either explain it in the text or
to provide further explanation in the appended glossary.
Technical Background:
In early 1998, the Faculty of Media, Music and Performance of Salford University bought and installed an entry-level web server (Silicon Graphics Challenge S) to complement the main university web server and to provide services to faculty members that were not available through the latter. A major service that we wished to offer was the possibility of both audio and video streaming of files. As opposed to only being able to listen to an audio/video clip once it has completed its download from the Internet, streaming technology enables the end-user to listen to the clip as it downloads regardless of whether they wish to listen to a static file or a live bitcast. To this end, the Faculty purchased Progressive Networks (www.realaudio.com) Real Audio Server which allows for multiple simultaneous listeners. Assuming both the server (bitcaster) and client (end-user) are connected to the Internet through their desktop computers, two other pieces of free software from Progressive Networks are required for the bitcast; the player and the encoder. The player is used by the end-user to listen to the real audio bitcast (or static file); the encoder requires more explanation.
The encoder is usually installed on a desktop computer (in our case we
utilise Wintel PCs) and takes audio that is input to the computers sound card, or is
played from the computers CD player or is a file on one of the computers disk
drives. It then encodes that audio in one of several Real Audio formats (the process with
video is much the same). During the encoding process much of the audio information is
compressed or discarded in order to reduce the bandwidth of the resulting audio file and
consequently the quality of the audio suffers drastically. This bandwidth (speed) can
range from about 300Kbps (Kilobits/second) down to 8Kbps and there are various options
(working mainly on frequency) that attempt to optimise the result for voice only, music
and voice, or music only.
Consider this, in order to bitcast compact disc quality stereo music, the
network (including receiving devices such as modems) through which it is sent must have a
bandwidth that is never less than approximately 1.4Mbps (Megabits/second). The fastest
publicly available modems have speeds of 56Kbps and many modem users will be using modems
that are of lower speeds than this (such as 28.8Kbps). This is the first problem, one of
bottlenecks in the network: your transmission speed is only at least as good as the
lowest bandwidth point in the network. In other words, the bitcaster cannot expect
their modem-based listeners to have much success when they attempt to bitcast CD-quality
music. The newest generation of Real Audio Encoders and Servers are able to encode
simultaneously at different bandwidths and sense what bandwidth the player is listening on
in order to deliver a more reliable stream. In order to accommodate modem users, Boom
Booom Net Radio encodes at approximately 22Kbps.
For the Real Audio bitcast to be available to listeners, it must be sent
via a network to the Real Audio Server. When bitcasts take place from within the Faculty,
a stable ATM (Asynchronous Transfer Mode) network of 155Mbps is used to reach the server,
although the encoding PCs network card has a bandwidth of 10Mbps. For location
bitcasts (see below) a Demon network is used, limited by the 56Kbps modem attached to the
PC, hence the encoding compression required must not only take account of end-users
networks but also the encoders network.
Naturally, as were talking of networks (specifically the Internet),
the Real Audio player does not connect directly to the Real Audio Server (nor the encoder
to the server); the audio signal must travel through several computers - 'routing through
hosts' in Internet parlance. Each of these host-to-host connections (or hops) will add
time overheads to the signal (in addition to the computer processing time necessary to
encode the audio signal before it can be routed to the server) and may traverse wide
geographical areas. Consequently, there can be a considerable delay between the encoding
computer and the listeners computer.
In addition, and although the Internet structure attempts to compensate
for this, various hosts in the route may have connection and/or congestion problems which
in extreme cases can lead to the Real Audio player experiencing drop-outs which are
manifested by either silence or wide-band noise.
As an example, in April 1998, the Facultys Music Department bitcast
at 22Kbps a conference on Popular Music & Technology and one of the authors of this
article was working at a university in Durban South Africa attempting to listen to the
conference. The conference was held on a Friday and a Saturday; on the Friday, reception
was so poor as to be worthless due to network congestion, mainly caused by the number of
users utilising Natal Universitys Dual ISDN 112Kbps network. On the Saturday (when
presumably students have better things to do), the reception was excellent. In both cases,
the delay was around 5 seconds.
Fig. 1 shows the results of a ping from a desktop PC in Salford University
to Natal Universitys web server (www.und.ac.za)
while Fig. 2 shows the results of a traceroute between the same two computers
(both taken after the event):
Fig. 1
Pinging Raven.und.ac.za [146.230.128.50] with 48 data bytes
Statistics for www.und.ac.za
5 packets transmitted, 3 packets received, 40% packet loss
round-trip (ms) min/avg/max = 2203/2409/2663
Fig. 2
traceroute to Raven.und.ac.za (146.230.128.50), 30 hops max, 40 byte packets
1 gw-adelphi.g-ming.net.uk (146.87.216.1) 2 ms 2 ms 3 ms
2 gw-mcc.g-ming.net.uk (194.66.23.1) 10 ms 14 ms 11 ms
3 manchester-core.ja.net (146.97.253.117) 24 ms 21 ms 14 ms
4 external-gw.ja.net (146.97.253.57) 26 ms 29 ms 33 ms
5 tglobe-gw.ja.net (193.63.94.85) 27 ms 28 ms 35 ms
6 ppt-gw.ja.net (193.62.157.6) 130 ms 136 ms 122 ms
7 gin-ppt-bb2.Teleglobe.net (207.45.215.165) 168 ms * 152 ms
8 gin-mtt-core1.Teleglobe.net (207.45.223.42) 170 ms 188 ms 295 ms
9 gin-nyy-core1.Teleglobe.net (207.45.223.62) 187 ms 181 ms *
10 gin-maee-bb1.Teleglobe.net (207.45.223.118) 253 ms 193 ms *
11 mae-east.att.net (192.41.177.3) 183 ms 265 ms 189 ms
12 * * gr1-h31.wswdc.ip.att.net (192.205.31.173) 213 ms
13 br2-a350s5.wswdc.ip.att.net (192.205.31.166) 198 ms 198 ms *
14 12.127.9.126 (12.127.9.126) 240 ms 194 ms 185 ms
15 ar1-a300s1.n54ny.ip.att.net (12.127.0.5) 197 ms 281 ms *
16 * 12.127.241.70 (12.127.241.70) 201 ms *
17 * * *
18 168.209.254.220 (168.209.254.220) 4994 ms * 4636 ms
19 * * 168.209.0.85 (168.209.0.85) 5377 ms
20 * * *
21 168.209.3.129 (168.209.3.129) 5133 ms * *
22 * * *
23 Gw-Uninet1.und.ac.za (146.230.196.1) 6835 ms * *
24 * * *
25 * * *
26 * * *
27 * * *
28 * * *
29 * * *
30 * * Raven.und.ac.za (146.230.128.50) 7607 ms
As will be noticed, both figures show that the connections were not always
successful (which may or may not manifest itself in problems for the Real Audio player
were this to be an audio signal we were attempting to bitcast), although the structure of
the Internet usually means that we will eventually reach the destination host. Each
asterisk in Fig. 2 indicates a failure to reach the host - there are usually three
attempts per host. Fig. 2 also shows that for an Internet signal to travel from the UK to
South Africa it must first travel to the USA where routers such as Teleglobe and AT&T
pass it to the final destination.
Boom Booom Net Radio:
Boom Booom is a collective that runs events once a month at a sited venue
in Manchester. It is run by a group of DJ's, sound engineers and media practitioners who
combine to form a safe and interesting mixed-media experience for the discerning
club-goer. The venue is filled with hand-made décor, video projections, lightshows,
strange paraphernalia, and loud, pumping dance music. Boom Booom has been on the Internet
for some time, maintaining its own web-site in a modest fashion for several years,
providing information on bands and DJ's, forthcoming events and a large section which
provides links to other members of the Netted Musical Underground. Since then, however,
Boom Booom's Internet presence has grown with regular forays into bitcasting. This
happened at a time when Salford University's Music Department had access to its own web
server, as well as purchasing a Real Audio server. Some members of the Boom Booom team
worked at Salford University and were instrumental in setting up the Real Audio server
software on their web server.
This process allowed Boom Booom to extend their scope from monthly nights into twice
weekly Internet broadcasts. With a laptop Wintel PC and a 56Kbps modem, the people
involved would meet at each other's houses on Sunday evenings and the DJ's would practise
their art whilst the world listened - we hoped.
Although we experienced network connection problems from time to time, all in all this
proved to be a technical success. The Wednesday broadcasts were done through Salford
University's ATM broadband network (although still encoding at approximately 22Kbps),
which allowed for much greater stability in the path between the encoder and the server,
creating a rock-solid output stream. We slowly watched an audience build up over time, via
a Java applet (a web browser application), that allows the Real Audio server
administrators to monitor the number of current listeners, and via various text-file logs
that the server stores for all listeners.
The next stage was to introduce Internet Relay Chat to the process. IRC allows
participants to chat in real-time within a text-based interface on a particular chat
channel. In this way, listeners had the opportunity to give feedback to the DJ's and chat
with other listeners or members of the Boom Booom team. This interaction between the music
producers and the consumers meant that it was possible for the producers to establish a
rapport with an 'audience', and for the audience to feel more involved in the whole
musical experience. It was, in short, a way of creating more of a community atmosphere
than is often possible with conventional media such as radio and TV. In addition, the
audience chatting with us on IRC were also able to provide feedback on the
technical/network issues, relaying to us any problems they were experiencing as they
happened.
A further step was taken when Boom Booom started bitcasting the live events. The
emphasis here was slightly different as they were 8-hour broadcasts, and we wanted to try
and capture some of the atmosphere of the club. We had two rooms, each with different
music, and took an audio feed from each room into a mixer and then into the PC. This
allowed us to choose which room's music to broadcast at any particular moment.
Additionally, we took a feed from the video mixer into a video capture card in the PC.
Both the audio and video inputs were encoded as a live file, sent down the telephone line
to the server at the university, and from there it was available to the rest of the world.
This enabled people to watch the video projections that were being used on-site, whilst
they listened to the live music.
Although the live events attracted quite a few listeners, it became clear that an
Internet audience can be frustratingly elusive. We started publicising the live events, as
well as the twice-weekly broadcasts, submitting them to web-based channel listings and
submissions engines such as Timecast (www.timecast.com)
, Audionet (www.audionet.com) and Yahoo (www.yahoo.com). This proved useful in helping to build up
the listener base.
Another development was Java chat as opposed to IRC. It became clear that not enough
people were actually taking the time to chat with us and we decided that this was because
it was too much effort to download a separate IRC client, install it, log on, find the
channel, and then start chatting, in addition to using a web browser and Real Audio player
on the same computer. The solution was to use a Java applet to allow us to embed a chat
engine into the actual web page, which meant that the listener only had to load the web
browser in order to be able to log on to the chat channel. Thus, more people were willing
to join in the online chat.
With all these processes, problems arose at the encoding end due to the amount of
processing that we required of the PC. Simultaneous encoding, running of the Java server
monitor, the Java chat and sending and receiving via the modem, all on one PC is
precarious, to say the least, and computer crashes at Boom Booom events are rather
frequent, leading to frustrated consumers and producers. The solution would have been to
have two computers attached via modems to different telephone lines and dedicate one to
just the encoding of the audio and video signals. Unfortunately, although many venues may
have more than one telephone line, they naturally prefer to reserve one for its proper
purpose.
Fig. 3 shows the results of a modem-based PC traceroute through to the Real
Audio server while Fig. 4. shows the results of a modem-based traceroute from
the encoding PC to the Real Audio server (both were taken during an actual broadcast on a
Sunday between 19.00 and 21.00 GMT):
Fig. 3
Tracing route to hexie.memtech.salford.ac.uk [146.87.216.80] over a maximum of 30 hops:
1 313 ms 150 ms 145 ms gyle-69.access.demon.net [194.159.254.69]
2 270 ms 365 ms 273 ms tele-core-1-ge025.router.demon.net [194.159.254.100]
3 336 ms 235 ms 157 ms tele-border-1-fxp0.router.demon.net [194.159.252.252]
4 157 ms 160 ms 160 ms gw.linx.ja.net [195.66.224.15]
5 158 ms 187 ms 159 ms south-east-gw.ja.net [128.86.1.50]
6 169 ms 165 ms 179 ms manchester-core.ja.net [146.97.253.62]
7 195 ms 180 ms 178 ms gw-mcc.g-ming.net.uk [146.97.253.118]
8 178 ms 173 ms 174 ms gw-peel.g-ming.net.uk [194.66.23.5]
9 174 ms 175 ms 174 ms hexie.memtech.salford.ac.uk [146.87.216.80]
Fig. 4
Tracing route to hexie.memtech.salford.ac.uk [146.87.216.80] over a maximum of 30 hops:
1 2892 ms 133 ms 162 ms gyle-87.access.demon.net [194.159.254.87]
2 2608 ms 2447 ms 1511 ms tele-core-1-ge025.router.demon.net [194.159.254.100]
3 155 ms 140 ms 147 ms tele-border-1-fxp0.router.demon.net [194.159.252.252]
4 146 ms 146 ms 1816 ms gw.linx.ja.net [195.66.224.15]
5 3016 ms 2902 ms 2932 ms south-east-gw.ja.net [128.86.1.50]
6 2716 ms 636 ms 153 ms manchester-core.ja.net [146.97.253.62]
7 2838 ms 2549 ms 1502 ms gw-mcc.g-ming.net.uk [146.97.253.118]
8 1918 ms 2070 ms 1789 ms gw-peel.g-ming.net.uk [194.66.23.5]
9 2922 ms 3044 ms * hexie.memtech.salford.ac.uk [146.87.216.80]
10 978 ms 1246 ms 1173 ms hexie.memtech.salford.ac.uk [146.87.216.80]
Conclusions:
Boom Booom bitcasting utilises modest technology, especially when making
use of modems for the live bitcasts, and the following conclusions may not hold for other
bitcasters. However, regardless of the organisation and level of funding, public
bitcasters have no control over possible bottlenecks between server and end-user.
It is quite obvious that theory often exceeds practice. This is a result
of the modest technology that is utilised by Boom Booom and the desire to provide a full
variety of services to their audience; hence the frequent computer crashes at live events.
Despite frustrations caused by network congestion, drop-out, computer
crashes and so on, audiences are remarkably patient. They perhaps understand that
bitcasting technology is still in its infancy, and it may be that there is a novelty value
attached to such bitcasts. Likewise, bitcasting - as in Boom Boooms case - seems to
have found or created an audience for events that would not normally be broadcast on other
forms of media. Additionally, providing chat pages, as a form of interaction between
consumers and producers, helps create a loyal audience who feel a part of the events.
It may seem that the problems enumerated here far outweigh the advantages.
Whilst it is true that modestly-funded groups such as Boom Booom will encounter most if
not all of these problems, larger organisations, like major Radio and TV networks, are
able to overcome them with more sophisticated and expensive technology. For the smaller
organisations though, bitcasting can provide valuable access to a global audience, for a
relatively low cost, in what is, still, a largely unregulated arena.
Glossary:
ATM | Asynchronous Transfer Mode. A broadband computer network with a bandwidth of 155Mbps. |
Bandwidth | The maximum number of bits per second that can be transmitted or received over a network. |
Bit | Contraction of binary digit. The smallest unit of digital data existing as either a 1 bit or a 0 bit (binary) |
Bitcast | An Internet broadcast in real-time analogous to a radio or TV transmission. |
Byte | A group of 8 bits. One character (for example a) is usually encoded within a byte. |
Client | Usually referring to a computer within a network that accesses data on another computer although, more correctly, the software on that host which accesses that data. |
Demon | An ISP (see below). |
Downloading | The process of getting a file from another computer via a computer network. |
Host | A computer participating in a network. |
Internet | A global collection of interconnected computer networks. |
IP address | Internet Protocol address. A unique network address for a host. Analogous to a postal address. The IP address of hexie.memtech.salford.ac.uk is 146.87.216.80. |
ISDN | Integrated Services Digital Network. A computer network with a bandwidth of 64Kbps. |
ISP | Internet Service Provider. An organisation that provides Internet access to other organisations or individuals. |
Java applet | Computer software written in Java code and that can be interpreted by a java-capable web browser on most modern PCs. |
Kbps | Kilobits (1000 bits) per second. A measurement of a computer networks (or part thereof) bandwidth. |
live file | Refers to the data in a bitcast as opposed to a static file that is stored on a computer. |
Mbps | Megabits (1000000 bits) per second. A measurement of a computer networks (or part thereof) bandwidth. |
Network | A collection of interconnected computers. The Internet is a network. |
Ping | Software that performs a quick check on the viability of a computer network and the status or accessibility via the network of the destination host. |
Server | Usually referring to a computer within a network that provides data although, more correctly, is the software on that host that provides or serves that data to a client. |
Static file | A computer file such as an audio file that is stored on a computer or removable media such as a floppy disk. |
Streaming | The process of either bitcasting or playing a media static file as it downloads. |
Traceroute | Software that displays information relating to the network path between two hosts on a computer network. |
Wintel | A contraction of Windows and Intel and referring to a PC that has a Windows operating system and an Intel (or similar) processor. |
Mark Grimshaw Mark Grimshaw
teaches Music Technology & Studio Production in the Music Department of Salford
University. With a BMus(Hons) from Natal University, South Africa and a MSc Music
Technology from York University, he worked as a sound engineer in Italy before taking up
his current position. He writes articles on music technology and African music in addition
to recording and production work. He can be contacted at: m.grimshaw@music.salford.ac.uk
Louisa Yong has been a freelance webpages designer since
1992 when she was still a music student in the University of Hong Kong. After she
completed the Music Technology MSc degree at the University of York in 1996, she received
a 3-year PhD studentship from the University of Salford to research and develop music
applications on the Internet. Louisa's current research project is on the Composers
Experimental Online Suite (ComeXos). The aims of ComeXos are to provide a graphical
interface for accessing the Composers Desktop Project (CDP), a sound manipulation package,
via the Internet in order to create and share audio samples. The research also
investigates the economic, sociological and educational impacts of such a project. She can
be contacted at: c.yong@music.salford.ac.uk
Ian Dobie graduated
from Salford University in 1998, having specialised in Music Technology and Studio
Production, and is currently doing a PhD at Salford University. He is undertaking research
into the creative, social, technical and commercial aspects of broadcasting and streaming
audio on the Internet. He can be contacted at i.m.dobie@music.salford.ac.uk
![]() |
![]() |
![]() |
© 1999-2007 International Journal of Urban Labour and Leisure |