Sound for live event streaming - processing?

[revbobuk]

There's a really good guide to using the ATEM software effectively

which I've recommended to many folk.

[TomHoward]

So far, what I've got is that Waves Playlist Rider plugin is about the most effective quick and easy tool. I've downloaded the demo and processed the audio off the stream in Logic with it, I haven't tried it in OBS yet but I do see a lot of info about Waves not working well in OBS and OBS only supporting older VST versions.

The processed and unprocessed waveforms:

Due to Logic's funny waveform zooming the unprocessed audio does not sound as bad as it looks and it certainly improves the quiet sections or reduces a sudden burst of applause to even out the stream quite a lot.

The only problem is it does 'pump' slightly in the songs, if there is a sudden loud section it does reduce, but I think if it were combined with a limiter / top end compression first to take that real top end off the applause and loud song sections, and Playlist Rider were used to bring out the low sections, it really does work quite well.

 

It has a 'short term loudness' and a 'long term loudness' and provides +-18dB of gain automatically according to the long term loudness. I'm finding set to a target of -14dB it's pretty much equally reducing the loud parts and bringing out the low parts.

Edited December 13, 2020 by TomHoward

[richardash1981]

There are multiple different problems to potentially solve here, but I remember that if you are happy with the local levels, but want to reduce the overall dynamic range in a non-obvious way, the EBU R128 suite of documents has some suggested compression approaches. Despite being a broadcast standard document, it's perfectly readable and freely available! https://tech.ebu.ch/docs/r/r128.pdf The one you want might be https://tech.ebu.ch/...ch/tech3343.pdf page 16 but I'm not sure - they have revised and re-issued since I read them first. I think what was recommended was a low threshold, low ratio approach like J Pearce described "1.1:1 compression across all the mics with the threshold right down at -60dB"

At the other end, dealing with applause (apart from moving the microphones away from the clappers and towards the performers!) need a fast-acting limiter with the threshold set above the performance level.

 

Could you route the pre-recorded audio sections into the stream using an aux send, and so level independent of the live playback of them? This should mean you can pre-set the level of those parts of the stream to not hit the fast compressor/limiter threshold.

[TomHoward]

Could you route the pre-recorded audio sections into the stream using an aux send, and so level independent of the live playback of them? This should mean you can pre-set the level of those parts of the stream to not hit the fast compressor/limiter threshold.

To be honest, we are doing at the moment, although the other way round - we are running speakers in the room off an Aux bus, with playback and some mics routed to it, and some mics routed only to stream.

Thanks for the document links, I will also try those compression settings on my dodgy stream recording and see what difference it makes.