Streaming set-up

[ChiefOperator]

All-

I'm helping a AM talk station set up its streaming (they'll use the Windows Media Encoder). We're currently looking at the processing, especially level control. I'm wondering what others are using for processing.

Are you using outboard hardware, such as a simple dbx or equivalent, or are you using processing software on the encoder machine (or both)? I've heard that some stations are using both together--a hardware compressor, such as a dbx, set for gentle gain riding and then a multiband limiter software program behind it on the encoder machine.

I would be very interested to hear the different set-ups that your stations are using..

Thanks!

 

[frankberry]

Have you attempted to use the station's main audio processor to feed the encoder?
Never done it but it might work for you.

 

[BofH]

I'm using the new Breakaway Broadcast Processor and have nothing but praise for it.

Sounds great and runs on your encoder computer (anything built within the last 2 years should run it fine)

Download a demo here:

http://www.claessonedwards.com

 

[Sgeirk]

We use SOS4 for ad insertion, and it comes with a halfway decent built-in AGC.

On another streaming product in our building, we're actually using an old CRL multiband processor and doing some final limiting in the software for the soundcard. Sounds great.

On another streaming product in our building, we're just using the output of our Omnia 6....with the bandpass filter set below 12k.

Good luck.

 

[gtomas]

Are you using outboard hardware, such as a simple dbx or equivalent, or are you using processing software on the encoder machine (or both)?

You will get best results with a proper streaming/HD broadcast processor or using and HD output from your on-air processor (if it has one). Anything else like studio compressor, limiters, etc will not be as consistent and sounding as good as a broadcast processor, no matter how much "enthusiasm" you put into your setup. There's just too much going on...


Regards,
Goran Tomas

 

[RF Man]

Have you attempted to use the station's main audio processor to feed the encoder?
Never done it but it might work for you.

This is exactly how I am getting audio to my streaming computers. We use the digital outs for air which leaves the analog outputs free to use for this purpose. On my particular model of Omnia they have provided separate volume controls for the analog and digital out which makes it very easy to control the over all level going to your computer.

Windows media encoder has done a great job over all but beware if you are going to use it to stream to your listeners directly it will only handle 5 streams at a time unless you modify the registery.

 

[radiorob2.0]

I'm using an CRL AGC and multiband unit on a music format and it sounds amazing. We use a third party for content distribution. It may go without saying but you might not be able to stream the syndicated programs.

 

[The F Mister]

I'm using a DSPXtreme for streaming..... works quite nice... ;D

At another station we run an OmniaFM for onair but also feed the computer with Windows Encoder and also run omnia AX on it to shape the audio just slightly more for streaming and optimize for listening on "regular pc speaker" most people use. If you have a processor with dual path processing you could skip the AX.

 

[konbaasiang]

Using an on-air processed feed for the stream is generally not recommended, due to the vastly different processing requirements. The on-air signal requires 50 or 75us pre-emphasis clipping, which limits treble headroom and adds distortion products to the sound. It sounds fine on the air, but when encoding a signal processed this way with a low bitrate encoder, several things go wrong:

- High frequency transients helps the encoder have somewhere to hide encoding artifacts. In most on-air signals, there is no such thing as high frequency transients (due to steep pre-emphasis limiting)
- The extra distortion products end up using up bits in the stream which should have been used for audio, resulting in even more encoding artifacts.

It's very important to use the right processing for each task -- there's no way to optimize one signal for air and streaming at the same time.

The best solution is to have a completely separate processor for the stream, and to feed it unprocessed audio - absolutely not on-air-processed audio.

Having a separate processor is even better than using the HD output of an on-air processor, simply because it allows you more control. On-air processors tend to use a common front-end for FM and HD, and this is great for consistency (since receivers will switch between the signals randomly) but the web stream has no such requirements, so one can optimize further without the need to sound exactly like the on-air sound.

I also agree with Goran - a studio compressor/limiter just doesn't cut it. People used them for broadcast processing before 1975, but once the first true broadcast processor (Optimod 8000) was available, there was just no looking back.

///Leif

 

[ChiefOperator]

Folks, thanks for all the great input.. Good information...

konbaasiang-
I understand the advice against using the external processing. However, let me ask, if one is using a software processor (such as Breakaway) for internet streaming, is there a problem with using an outboard box for gentle gain riding BEFORE the software processing. Perhaps I'm wrong, but it seems like some gentle gain riding to control the audio input levels to the software would be beneficial. Is there no advantage (or a disadvantage) to control erratic levels through gentle gain riding BEFORE the audio enters the computer and inputs into the software processing program?

Interesting discussion....

Thanks!

 

[konbaasiang]

Hi Chief!

Using an outboard agc is only advantageous if the outboard one is better than what's already in the processor (whether the processor is software or hardware), except if there is a noisy STL in between.

There's no need to control erratic levels before -- that's the main processor's job! Breakaway in particular has extremely effective gain normalization, and it's the smoothest sounding one personally have ever heard. Never caught by surprise, never pumps (unless you want it to).

However, if you have no choice but to use an external AGC, then you can, but I would then lower the input level to the sound card until the AGC in Breakaway is almost at full gain, so that they don't start fighting each other. Chaining compressors is a very delicate matter, because:

The second compressor's release time must be faster than the first one's attack time!

If it's not, the following will happen:

You surprise the chain by jumping from the fade-out of an 80s track (all compressors released all the way up) to a modern recording that starts suddenly.

Compressor 1, which is slower than Compressor 2, will attack very slowly.
Compressor 2 will attack quickly and make up for the difference, quickly bringing the audio in line.
Compressor 1 then has no idea what Compressor 2 is doing, and will continue downwards at moderate speed. This is good, it's what it's designed for, but:
If Compressor 2 then do not release as fast as Compressor 1 attacks, you will have a gain-ride *downwards*, quieter than before the loud surprise happened in the first place!

I've heard things like this on the air occasionally, and can only shake my head...

Properly tuned though, 2 or even 3 compressors in series can work VERY well. Breakaway exploits this in certain presets, for example French Kiss. When you surprise it, and see three different colours on the AGC bar, that's three wideband compressors in series all working together! (in Breakaway Personal, the bar is just one colour, so you don't see which stage is doing what, but in BBP you see everything).

///Leif

 

[ChiefOperator]

Okay, I get it now.... Thank you!

Speaking of compressors in series, that is the design of the Very Nice Compressor from FMR Audio. When the compressor is in "Super Nice" mode, audio passes through three internal compressors, each one adds just a little compression. I've listened to the box and it's very transparent.

http://www.fmraudio.com/RNC1773.HTM

Thanks again for the explanation--it cleared up my questions.

 

[dtube1]

Speaking of compressors in series, that is the design of the Very Nice Compressor from FMR Audio. When the compressor is in "Super Nice" mode, audio passes through three internal compressors, each one adds just a little compression. I've listened to the box and it's very transparent.
http://www.fmraudio.com/RNC1773.HTM

The FMR is one of the most pleasent-sounding compressors I've heard; and its an absolute steal for what they charge. But, it runs unbalanced i/o, which might or might-not, be a problem for your streaming needs, depending on the soundcard. Plus, as its a compressor, its activity will be completely dependant on the input levels. If you are running an active AGC ahead of it, it will work more consistently. As it is a wide-band unit, it will also hole-punch on transients. The FMR Really Nice Levelling Amp (RNLA), might be a better choice for an inline streaming app. Again though, it is also unbalanced.
-D

 

[ChiefEngineer]

Feeding the stream with an over the air feed results in stable levels and bypasses the preemphasis issues.

And you can always listen from anywhere to check the station's transmitter.

 

[Sgeirk]

I disagree with this statement, in that: it's a known issue that heavy bit-reduced codec's do not react well with most clipped audio. And that's precisely what happens at the mpx output of 90% of all processors.

Also, if your transmitter goes off the air, wouldn't you really rather prefer to at LEAST be able to have people turn to a stream, if given the option? Just sayin'.

 

[RadioMan203]

I know when I had my Optimod 6200 DAB on my former oldies stream and I reduced the bandwidth feeding the encoder that the stream fidelity improved tremendously. I would think it's because the codec isn't processing information that will be thrown away.

 

[gtomas]

I know when I had my Optimod 6200 DAB on my former oldies stream and I reduced the bandwidth feeding the encoder that the stream fidelity improved tremendously. I would think it's because the codec isn't processing information that will be thrown away.

Actually, it's because you are removing part of the spectrum, therefore you are reducing the amount of information codec has to deal with. It also happens to be part of the spectrum (high frequencies) where codecs exhibit most artifacts, so by removing them you are very effectively reducing the artifacts.

I disagree with this statement, in that: it's a known issue that heavy bit-reduced codec's do not react well with most clipped audio. And that's precisely what happens at the mpx output of 90% of all processors.

Which is the biggest issue.

But there is also another aspect - by using off-air signal you are feeding the encoder with FM processed audio. That audio has been pre-emphasized by the processor and high frequencies that have been boosted up to 17dB were then compressed, limited and clipped back down to 0dB. All of that causes the high end density to be increased, in comparison to non pre-emphasized processing. That increased density puts extra burden right on the codecs weakest point - the high frequencies. Interestingly, sometimes that sounds better as it helps mask the artifacts. Most of the time however, it seems to increase the artifacts


Regards,
Goran Tomas

 

[The F Mister]

But there is also another aspect - by using off-air signal you are feeding the encoder with FM processed audio. That audio has been pre-emphasized by the processor and high frequencies that have been boosted up to 17dB were then compressed, limited and clipped back down to 0dB. All of that causes the high end density to be increased, in comparison to non pre-emphasized processing. That increased density puts extra burden right on the codecs weakest point - the high frequencies. Interestingly, sometimes that sounds better as it helps mask the artifacts. Most of the time however, it seems to increase the artifacts


Regards,
Goran Tomas

Some might even like these artifact to some degree. It's like clipping, some don't like it at all, some just a bit and some like heavy clipping with IM distortion to certain extend. Most people here value the quality we can achieve these days and also in the past but do the new generations care? My generation don't buy CD's/DVD's. We download our music in MP3. I'm afraid that our ears are becoming less sensitive to these artifacts. If I know my audience isn't going for the quality but for the content do I care to put so much effort in trying to mask these artifacts? As a engineering point of view yes but...... what does the rest think?

 

[konbaasiang]

I believe quality still separates the winners from the rest... Even though people may not be consciously aware, it tends to affect TSL. Quality will not make someone listen to a station playing music they don't like, but a lack thereof may make someone tune out without even knowing why.

///Leif

 

[The F Mister]

I agree with you on this.