Links from WHEAT:NEWS Volume 4, No. 1


WHEAT:NEWS, Wheatstone’s monthly newsletter, publishes a list of links of general interest in each issue. If you don’t receive WHEAT:NEWS and would like to, simply click the banner above and you’ll be led to a page where you can subscribe. We don’t sell or give away your contact info and we don’t SPAM, and we make it easy to unsubscribe at any time.

Volume 4, No. 1 published the following links:

Tips for Streaming

Jeff Keith

Jeff Keith

If you’re dealing with streamed audio, we have two words for you: bit triage. Yes, you’re going to have to give up some audio details in order to get programming to fit the bitstream. But the secret is to process the audio so the codec doesn’t deep-six the details that matter. Try these tips from someone who lives and breathes audio processing: Jeff Keith, our Senior Product Design Engineer for audio processing.

  • Set the peak audio level going into the codec at 3dB or 6dB below 0 dBFS. This extra headroom is needed for the decoding process. The lower your target bitrate, the more important this extra headroom is, without which the decoder can easily overrun 0 dBFS and cause distortion.
  • Roll off the highs starting at 12kHz, even lower if you’re streaming talk or at a low bit rate. Don’t waste your bits on frequencies most of your listeners won’t hear anyway, especially since coding them can cost you the frequencies that matter more.
  • Adjust stream processing so that listeners don’t have to chase the volume control. Volume levels should be consistent for a mobile listener who’s likely to be in an area where there’s traffic or other background noise, yet not overdone for the office PC listener who has a smaller sound field.
  • Go for linear audio as your source. How can you tell if it’s full linear? Linear audio takes up around 10 megabytes/minute (CD standard: 16-bit, 44.1kHz, stereo), or about 50 megabytes for a five-minute file. Audio files that are smaller than this probably are bit reduced and therefore not linear.
  • Never, ever stream MP3 songs. MP3 songs have already been through the codec wringer and have lost a great deal of audio information as it is. Subjecting whatever’s left to another round of coding forces the streaming codec to remove even more audio, which can bring up the noise floor and cause that unwelcome “swishing” sound typical of cascaded codecs.
  • Leave the aggressive processing for on-air. The more aggressive the stream processing, the more artifacts you throw at the codec and the fewer opportunities it has to get the sound right.
  • Leave the loudness war there, too. The buffering time at the decoder can make A vs. B loudness comparisons difficult, if not impossible, which renders the loudness versus quality debate irrelevant.
  • Forget about streaming straight from the FM audio processor. Don’t go there. Tap audio from the studio playout system instead and use dedicated processing for the streams. The last thing you want is to force the codec to throw bits at pre-emphasized highs; this is just another waste of bits that no one will hear.
  • Target for using at least 96kbps for stereo and 48kbps for mono. If you can ratchet up the bitrate, go for it. The quality will be that much better.
  • Ease up on using too much stereo enhancement on music, and always use mono for sports or talk. A specialized algorithm in Wheatstone Aura8-IP and VP-8 audio processors automatically reduces the L-R on a frequency sensitive basis, so the codec can assign those bits to audio content that is more important to perceived quality.
  • Banish clipping altogether. There’s a reason why Wheatstone VP-8 and Aura8-IP audio processors have no clipping preset for streaming applications and why we use look-ahead limiting to manage audio peaks instead. Signal clipping generates both harmonic and intermodulation distortion components, which the codec can’t discern from audio but tries to code anyway.
  • A little bit of multiband processing can make up for a multitude of codec sins and can make the audio a lot more pleasant to the ear. Start with presets in your Wheatstone VP-8 or Aura8-IP processor and make minor adjustments as you listen. The VBMS feature in our processors is helpful at enhancing bass while simultaneously reducing bass distortion, improving bass feel and mid and high-end detail and clarity.
  • Always adjust streaming processing by listening to the output of the codec, not the output of the processor. There’s a lot going on in the processor’s codec algorithms and what it needs to do to make the audio sound great at the output of a listener’s decoder probably won’t sound very good at the output of the streaming processor. Only by listening to the end result – the output of a decoder or player – will you know for certain which adjustments actually sound good, and which ones do not.

Sample Rates and Interoperability

What would you say if someone tried to sell you a new car that didn’t run on gas or battery? Exactly.

We thought the same thing, which is why we designed our AoIP with selectable sample rates. The WheatNet-IP Intelligent Network has selectable audio sample rates of 44.1kHz or 48kHz, which is critical to music creation, storage and replay now as well as under any proposed interoperability standard going forward.

Sure, it would have been easier to insist broadcasters go with one sample rate.

cd_digital_audio_logo2_28351But, locking broadcasters into sampling at 48kHz, for example, would mean that every bit of audio originated on a CD would have to be sample-rate converted in order to pass through the system. As it is, 44.1kHz is the Red Book audio CD base sample rate, and therefore the sample rate already used for most radio music libraries. An AoIP system that requires the majority of radio broadcasters to convert their entire CD libraries to 48kHz sampling just seems cruel and an unnecessary loss of audio quality.

Besides, sample rate flexibility is in the true spirit of studio interoperability.

Wheatstone is a proponent of studio interoperability as well as an active member of organizations such as the AES X192 task force.

Q & A: Interactive Digital Networking

Q: I’ve been hearing a lot about the interactive digital network as it relates to television audio consoles. Is this a new digital console feature offered by Wheatstone?

Paul Picard Headshot

Paul Picard

A: Sort of. What you are referring to is a new approach to console design. Whereas some consoles have fixed functions, we’ve designed our Dimension One and other television audio consoles to provide these functions on an interactive basis through the network. There is no fixed number of inputs and no limits on the type of audio the console can accept. This design removes these restrictions so that any audio source can connect to any fader, and any type of audio can connect to any input – even MADI (we have a bridge for that, too). Signal processing power is not limited to the size of the console or the console’s ability to dissipate heat, either. A Dimension One console can deliver 1,024 channels of digital signal processing and support 72 faders, each 5.1 capable. Broadcasters say it’s just a better all-around design than what you’d find with fixed-function consoles.

Paul Picard is a Technical Support Engineer with Wheatstone Corporation.

Game On for IP in Stadiums

IMG College StudiosWhen Studio Operations Manager Ben Blevins at IMG College, a huge player in sportscasting with 2,100 affiliates and 51 college sports networks, says he’s doing some interesting IP experiments in the field, we take notice.

He is easily one of our IP super-jocks with more than 75 BLADEs, dozens of Wheatstone surfaces, and all the WheatNet-IP AoIP networking you could ever want in a 46-studio sports complex located in Winston Salem, North Carolina.

He says he can’t imagine returning to the days of punch-block routing and analog consoles, when making changes on the fly was nearly impossible. “Even something as simple as hot levels meant we had to reset the entire studio,” he says.

Blevins tells us he likes the simplicity and clean sound of the system, so much so that he’s been experimenting with an all-IP audio feed from the stadium to the WheatNet-IP BLADEs at the studios. He’s tried it on a few smaller venues using a Comrex Access unit at the stadium to deliver linear AES/EBU digital into a WheatNet IP BLADE at the studio. He also has a Tieline Genie unit with optional WheatNet-IP card available, and this audio codec unit also can connect remotely into the WheatNet-IP network similar to a BLADE.

The system gives him full access to the WheatNet-IP studio network from anywhere there’s an internet connection, complete with all logic control for creating salvos and macros, starting and stopping GPI-connected devices, and controlling faders. Another plus is that he can change around the studios in a flash, so that if there’s a basketball game going on at a college in the afternoon and a football game at the same venue that evening, setup is a breeze.

He says he’d like to have that same ease of use everywhere, and that’s why he’s now working with other sports networks and organizations, including the NFL, to establish an IP standard at major sports stadiums.

If all goes as planned, sports stadiums could begin offering IP connectivity as an alternative to ISDN connectivity starting next game season.

Links from WHEAT:NEWS Vol. 3, No. 15


WHEAT:NEWS, Wheatstone’s monthly newsletter, publishes a list of links of general interest in each issue. If you don’t receive WHEAT:NEWS and would like to, simply click the banner above and you’ll be led to a page where you can subscribe. We don’t sell or give away your contact info and we don’t SPAM, and we make it easy to unsubscribe at any time.

Volume 3, No. 15 published the following links:

  • Here’s a pretty good wrap-up of the recent Arbitron conference.  (Radio Info)
  • Music to code and engineer by? Here are some suggestions. (Mashable)
  • In case you haven’t heard, here’s the scoop on Harris Broadcast. (Broadcast Engineering)
  • If you’re a joiner, you might want to check this out.  (Google Plus)
  • The Conclave, a nonprofit group dedicated to improving the quality of broadcasting through education, has some interesting webinars and a learning conference scheduled for 2013.

Dealing with Today’s Over-Processed Source Material

Jeff Keith

Jeff Keith

You know what a sine wave looks like – and it sure as heck isn’t square. Which, frankly, is more than we can say for a lot of the source CDs that you have to somehow pass as quality on-air programming these days.

No doubt about it. The quality of CDs and other source material has been on the decline in recent years. Recording studios used to master at a 20 dB peak-to-average ratio. These days, it’s closer to 6 dB. So, what can you do? Plenty, according to Wheatstone’s Broadcast Audio Perfectionist, Jeff Keith.

  • Keep a level head. About 18 to 20 dB headroom throughout the air chain should do, whether the studio is analog, digital or a mixture of both. This will give you enough peak headroom above average operating levels to keep distortion in check when those source levels run hot. And they will.
  • Set the audio gains within your air chain so that each piece of equipment reaches its clipping point at the same time. This gives you optimum dynamic range and you might even gain a little sound depth and detail along the way.
  • Balance left/right channel levels, differences of which can generate artifacts in the stereo sound field.  Channel imbalances are especially problematic for mono material because even slight level differences in live or recorded voice, for example, can create additional L-R information that ends up as distortion in the stereo field.
  • Use bit-rate reduction and codecs sparingly. This is doubly important now that source material is more densely compressed, which makes it more difficult for codecs to determine and therefore remove only the insignificant audio that our ears might not notice. One codec in the air chain is probably necessary. But running the result through another codec is just asking for it.
  • Go for linear instead of bit-reduced music storage. A lossy codec at a 4:1 compression ratio takes away almost three-quarters of the audio waveform. Don’t go there.  The cost of hard drive storage went from around $100/gigabyte to $0.01/gigabyte in recent years, so use those gigs for linear audio storage.
  • Resist the temptation to air or stream low bit-rate MP3 files from the internet, no matter how new and hot the song. Enough said.
  • Use multiband EQ to cover a multitude of mastering sins. Applying selective EQ to hotly compressed cuts can improve the sense of dynamic range and make it easier for any codec in your air chain to determine which bits are critical to the ear and which can be removed without your ear noticing. Remember that the songs cut today and those that were cut in the ‘80s are decibels apart in just about every way, so a little equalizing in just the right places can get them to meet somewhere in the middle for a more consistent on-air sound.
  • Back off on limiting and clipping. You don’t need as much as you did in the old days.  Your friends in the recording studio have done plenty of that already.
  • Eliminate troubles. If it’s bass you’re missing, there’s a good chance you’re trying to make up for it with EQ in the on-air processor. If your bass settings seem like a lot, they most likely are and there is trouble elsewhere. Find the culprit through a process of elimination. Jeff recalls one broadcaster who called to say he had the low-end EQ cranked up as much as he could get it, but his station still sounded thin on the air. “I suggested he listen to just the output of the console and sure enough, the bass was there. It turned out that he had a processor purchased in the ‘80s sitting upstream in the rack getting nice and toasty. Over time, the coupling capacitors kept getting smaller in value.  Unfortunately, when the capacitance goes down, the reactance goes up and the bass goes away. Problem solved,” says Jeff.
  • Check frequency response. If something doesn’t sound right, start at the source and analyze the frequency response of the entire air chain. Stations with all analog gear should be able to achieve a frequency response flatness of better than +/- 1 dB between 30 Hz and 15 kHz. Stations with all digital air chains will typically measure an order of magnitude flatter (+/- 0.1 dB) response and those with a mix of analog and digital gear will typically see numbers somewhere in between. These days, you don’t need expensive test gear to do this. Check out Jeff’s article on everyday studio and software tools you can use to proof the air chain.
  • Do an audio proof.  Seriously. (See above.) If you have to change equipment settings often, you need to do an audio proof. If you have distribution amplifiers that haven’t been benched in years, bench them and proof them. The same goes for mic processors, audio consoles, and any and everything that passes audio. Most of the analog equipment is suspect, but keep in mind that DSP gear has to bring in the signal somehow, and analog components age in DSP equipment same as they do in analog gear.
  • Go IP all the way. Ditch the sound cards, and use software drivers for automation-to-AoIP communication when possible. Our AoIP BLADEs use 24-bit audio and taking advantage of this capability can help retain a competitive and dynamic sound.

Wheatstone at FEMA and elsewhere in DC!

Just ahead of Sandy and again a week before the presidential election, our own Paul Picard traveled to Washington, D.C., to commission WheatNet-IP routing and IP-12 consoles for FEMA, as well as for another customer in the District. (Shhh.) Just a short jaunt from the White House, the FEMA Media Center is using a Wheatstone routing system to equip several studios and a briefing room. We’re told the installation was put to good use during FEMA’s response to Sandy.

Q & A: Ethernet Switch Selection

Q:  Can WheatNet-IP work with any switch? What are the main switch considerations I should be thinking about for AoIP?

Kelly Parker

A:  There are literally hundreds of Ethernet switches on the market today. Chances are you have a compatible switch in your facility already. The key choice you face is whether to use managed or unmanaged switches. The amount of Ethernet audio hardware you deploy ultimately drives this decision. Keep in mind, most gigabit switches will work with WheatNet-IP to a degree. Unmanaged switches are relatively inexpensive, so in a small system this may be an acceptable cost versus performance compromise. As your system grows, so do the chances of exceeding the capability of inexpen­sive unmanaged switches. Careful consideration of current and future needs should be given when choosing switch hardware.  My white paper, Network Design Considerations for Ethernet Audio, should help you wade through the options. 

Tee Time

There’s something you need to know about Tee Thomas, aside from the fact that he’s Ramar Communications’ Chief Engineer for three FMs, an AM and ten TV stations in Lubbock, Texas.

He’s a bit of an eccentric when it comes to sound quality, even by Wheatstone standards. He’s known to squeeze every bit of quality from audio. We can’t tell you some of the cool audio techniques he revealed to us (we promised), but we can tell you that he installed the WheatNet-IP AoIP Intelligent Network earlier this year using about 15 BLADEs, and elected to use software drivers for his BSI OpX automation system instead of soundcards. No 16-bit, 44.1 kHz sampling for his stations. It’s straight up 24-bit, 44.1 kHz sampling all the way. And by way, we mean WheatNet-IP AoIP routing into a fiber optic link to the transmitter site.

As a result, he says, the sound is more dynamic and open and loud. “The WheatNet really improved the quality of broadcast because while most of the digital audio equipment is 16-bit, 44.1 kHz (sample), the BLADEs are 24-bit, 44.1 kHz (sample) so you have so much more bit depth and the audio quality is so much more superior than standard audio,” explains Tee, who admits he’s gone “nuts with the Wheatstone stuff.”

He has two E-1 digital consoles and a smattering of digital boards that run off the WheatNet-IP. He has mic processors built into most of them, many used as pre-processors for streaming, which Ramar Communications does a fair amount of these days. No surprise, then, that he has a couple of BLADEs with audio processing built in for each of the 16 streams feeding the stations’ programming on the web. The Aura8-IP BLADEs take up two rack spaces for all 16 streams. Sixteen streams of multiband processing in two rack units? It still amazes us.

Audio programming comes out of a server in the back room, then goes through a network cable to the studios, and then to the fiber optic link out to the transmitter. It’s 100 percent IP all down the line.

Tee says he noticed the difference in quality immediately after installing the BLADEs.

We’re not surprised.