Your audio journey – Part 36 – Compression for internet radio

Following the compression series of the audio journey where I looked at compression for different scenarios with the single band compressor and compression using the multiband compressor I received a question through the website asking me about setting up compression for internet radio that would achieve a sound like an FM station.

It’s worth noting that FM and DAB radio stations have a lot of powerful processing in their chain. The Playout system (RCS Zetta, Myriad, Genesys), microphones, IRN feed, ISDN, telephone feeds all go through the desk and they are all monitored by the presenter through the desk to make sure the levels are correct. The microphone will go through local processing in terms of EQ and compression but the rest of the feeds will be natural.

Once the sound leaves the desk it then goes for processing. This will usually be an automatic leveller, multiband compressor and limiter and that combination can be set in many different ways which is why Kiss FM UK sounds different to Capital FM, and why Radio 2 sounds different to Magic. Stations are generally quite secretive about the way their processing is set to achieve that recognisable station sound that each station has. You need only hear the same song on Kiss FM UK and Capital FM UK to hear that song sound slightly different on both stations due to the way they have their processing set.

The most common processor is the Orban Optimod. You just need to look at the specifications here – https://www.orban.com/specifications-optimod8700i – to see how powerful a unit it is and the different tasks it undertakes to get the signal to sound “perfect” to station management when it then hits the transmitter for radios to pick up.

Internet Radio is run on a much smaller scale and budget. There are however two solutions to levelling out your audio through compression.

One would be hardware solution. To achieve this you would need a processor such as the dbx 166xs. You would run your desk output into the dbx166xs, set the processing as desired and then take the output from that to an audio interface. This audio interface could then connect back to the playout PC if it’s powerful enough and has enough processing and memory headroom and the stream could then be sent using the input signal from the audio interface. If the playout PC is already near maxed out then put the interface into a second PC and stream from there.

The second option is a software solution. Within a software solution there may be different options of how to achieve your compression.

Software such as PlayoutOne, popular with Internet Radio stations, comes with a basic single band compressor which import to note doesn’t compress the master output but compresses each player. The reason for this is that PlayoutONE has four players plus a stream player, all of which can be sent to different outputs for your mixer. It would therefore be impossible to compress a final mix without removing the option of having each player being able to be sent to a different output. (Note the ability to send to different outputs depends upon the sound card available to PlayoutOne within your play out PC.)

In order to achieve a good, level of sound you need to set your threshold, ratio, attack and release before then adding back make up or output gain. In the video I work through the possible settings using Adobe Audition’s single band compressor because the preview window in Audition allows me to demonstrate the effect the compression settings will have on the levels.

You’ll see a perfectly level sound can be achieved by setting the threshold, ratio and attack but a level sound set that way can compromise the quality of your final audio sent to the host for streaming.

It’s for that reason that I recommend using a lower ratio and a longer release time. The release time is how quick the compressor stops working when the audio level passes back beneath the threshold. The compression is only applied at the set ratio to sound which passes above the threshold, but by extending the release time to be longer you are basically running the compressor full time meaning it will hold the audio at a more constant level without too much degradation of the audio quality.

You’ll see in the video that while we set an aggressive ratio with a higher threshold and achieved a good constant level we did this at massive cost to the quality. We also set a lower threshold, therefore affecting more audio, combined with a lower ratio which, looking at the waveform, provided a fairly level sound but also affected the quality.

Only by extending the release time of the compressor could we get something which was fairly level and still sounded ok.

I demonstrated all this within Audition but the settings I’ve used could be transferred to your software if your play out software has a compressor built in.

On the compressor in Audition, and PlayoutONE, there is an output gain, also known as make up gain. Compression, when set aggressively, will reduce the overall volume and this means you need to make back up what you’ve taken away by adding a certain amount of db to the compressed sound.

Some compressors also have an input gain. This means that you can boost the audio before putting it through the compressor and then setting the remaining settings of the compressor the way we did in the example on the video. Be careful as too much input gain will result in a harsh distorted sound.

Another option is to compress your audio within your stream software. It may be that your stream software allows for plugins to be used in the chain and will process these before outputting your stream to the host for others to then listen in to. A good processer here could be from Waves Audio, the L3 Ultramaximiser.

One option I didn’t cover in the video would be routing your audio through an excellent piece of software called StereoTool. There are VST and Winamp / DSP versions which are essentially a plugin that your stream software could use to send the audio through before sending it to the host. This would be the easiest way to do this.

There is also a stand alone version of StereoTool available. In order to use this you’ll need to take the signal that was going to the streamer and run it through StereoTool before then configuring your streamer to take the output of StereoTool and send that out. This is achieved by using software such as Virtual Audio Cable (PC) or Loopback from Rogue Amoeba (Mac).

Configuring Virtual Audio Cable or Loopback can seem daunting but what these utilities do is allow you to route your audio around within your PC virtually without having to plug cables into line out and in. In a simple set up you can take the output from your play out software and send it to Virtual Audio Cable or Loopback, send that to StereoTool and then take the output from StereoTool to your stream software.

There are plenty of solutions available. Maybe the easiest would be to send your audio through a hardware solution but then you’re into the cost of the processor and interface needed to get the sound processed and back to a PC for streaming. Software compression does work but can be a little more tricky to set up.

You may want to spend careful time re-watching the compression part of the Audio Journey to get an understanding of compression and certainly watching how the compressor can be set up badly and how it can be set up well before attempting to change the processing of your internet radio stream.