November 30th 2002

My mate Jim's been converting his garage into a recording studio for a while now and it's ended up looking very swish indeed. On Nov30th 2002 this was the setting for a 6 hour webcast under the banner "ImageFM" powered by OAS Playout. This was put out onto the web via various cluster servers through Shoutcast.

We (that is me & Jim) tend to use the playout software in a subtly different manner to say the folk at Hastings Rock - there's a lot more jingle inserts/overlapping sequences in the kind of material we produce so whilst the Hastings Rock trial proved the package on a continuous use basis this should more pick up on any quirks of the package with "people trying to be too clever with it by half" as I put it.

Technical Overview

Aside from the general audio connectivity going on, the PC infrastructure supporting this was a good deal more complicated than applied with Hastings Rock. Music was hosted on two different machine - my Linux box and Jim's Win95 box. The studio machine was a 1.5Ghz Athlon running Win98 - in addition to running Playout it was also host to the Winamp Shoutcast DSP and a software audio compressor. The soundcard setup used Jim's Gadget Labs 8 port sound card - 2 channels for OAS Playout and another pair to feed in the mixed output to the compressor before being passed over to Winamp for streaming.

3 Shoutcast streams were produced from the studio machine, to the servers running on my Linux box - 24kbs,64kbs & 192kbs for internal recording purposes. Finally the whole lot had to be routed through a Win2k box serving as a gateway to the internet.

To get the necessary bandwidth we were relying on a number of other people with broadband connectivity to relay out the 2 streams in a Shoutcast 'cluster' arrangement - 2-3 people providing a 64kb cluster and one other to provide the 24kb server. In addition to this, the latest Beta of OAS Playout can provide current track information to Shoutcast servers so a local web server was also being used to provide this information along with cluster server IP addresses - this also needed to be routed out to the 'net.

who's that grinning idiot?

Trial Log

Not the smoothest of starts - initially we had some teething difficulties getting the Shoutcast relay servers clustering reliably, unfortunatly distracting me a fair bit from getting on with my contribution to the webcast. 15 minutes in the studio machine lost visibility to all external Windows workgroups and the machines in them - including my music server and required a reboot. Later on I managed to get OAS Playout locked into playing a jingle repeatdly, this time requiring a restart of the package to rectify the situation. The other significant problem was the software on the Win2k routing us onto the internet also kept crashing causing total disconnect of all Shoutcast relay servers - not doing a lot for listener retention that.

After about the first 2 hours everything was running about as smoothly as it was going to get. Later on the studio machine lost external workgroup connectivity (again) and possibly around the same time OAS Playout stopped updating the Shoutcast track/title information - the two things may be related.... The Win2k routing software was put into a 'self recovery' mode so it restarted again after a crash which smoothed thinks out slightly.

Beyond the issues mentioned so far, Playout also introduced a short sound 'clip' on the end of any sound bites played through the jingle players over main audio - this was somewhat annoying although is clearly only occuring on this particular hardware setup.

strike a pose.....

enough of this fangled gadgetry, back to the vinyl....

From my perspective, OAS Playout performed reasonably well - a significant problem is the Gadget Labs sound card is fairly old and isn't well supported by DirectX, due to lack of hardware buffers, software mixing is used by DirectX which can be problematic - almost certainly the finger of guilt points in that direction wrt to the sound 'clips' occuring on the jingle players. As to the ever repeating jingle I induced earlier, this bears further investigation/testing before this can definitively be blamed on DirectX.

The software audio compressor running on this machine is another unknown element - certainly it floored the machine and in retrospect should probably have been given a dedicated machine to itself.