As it turns out, 300 kbp/s or greater. At least in critical listening environments according to the paper titled Perceived Audio Quality of Realistic FM and DAB+ Radio Broadcasting Systems (.pdf) published by the Journal of the Audio Engineering Society. This work was done by group in Sweden and made various observations with different program material and listening subjects. Each person was given a sample of analog FM audio to listen to, then they listened to various audio selections which were using bit reduction algorithms (AKA CODEC or Compression) and graded each one. The methodology is very thorough and there is little left for subjective interpretation.
In less critical listening environments, bit rates of 160-192 kbp/s will work.
I made a chart and added HD Radio’s proprietary CODEC HDC, which is similar to, but not compatible with AAC:
||Bit Rate (kbp/s)
|HD Radio FM; HD1 channel*
||HDC (similar to AAC)
||96 – 144
|HD Radio FM; HD2 channel*
|HD Radio FM; HD3 channel*
|HD Radio AM*
||32 – 128
||MPEG II, Dolby digital
||192 – 256
||PCM, DTS, Dolby digital
||MPEG I,II,III, WMA, AAC, etc
||32-320, 128 typical
||128 – 256
||96 – 320
||64 – 256
**PCM: uncompressed data
This is the composite Mean Basic Audio Quality and 95% confidence intervals for system across all excerpts:
Over the years, we have simply become accustomed to and now accept low quality audio from mp3 files being played over cheap computer speakers or through cheap ear buds. Does this make it right? In our drive to take something good and make it better, perhaps it should be, you know: Better.
Special thanks to Trevor from Surrey Electronics Limited.
A little blast from the past. This was found in a transmitter manual at one of the sites we take care of:
CCA Optomod 8000
I thought I would scan it and make it available here. As luck would have it, there is also a corresponding piece of equipment to go along with it. I had never seen a “CCA Optomod” (.pdf) before I was working at one of the radio stations in Trenton, Florida. This unit was rescued from under a pile of garbage out in the lawn shed. It was full of mud was nests and mouse droppings. Needless to say, it required a bit of TLC to return it to operation. I replaced the electrolytics, cleaned it up and ran some audio through it. It is probably as good as the day it left the factory. Bob Orban made some really good stuff in his day.
CCA Optomod 8000
The original Optomod 8000 was an evolutionary design that made FM radio processing what it is today. The idea of combining broadband limiter, AGC and stereo generator in one box was a radical departure from the norm. The audio limiter functioned as a 15 KHz low pass filter and broadband AGC.
Orban Optomod 8000 audio limiter block diagram
The stereo generator used very modest amounts of composite clipping to reduce overshoot and transients. Many people disparage composite clippers. If done correctly, it is transparent to the listener and increases perceived loudness by stripping off modulation product that is non-productive.
Orban Optomod 8000 Stereo Generator block diagram
Some thirty five or so years later, there are still many of these units in service in various stations around the world.
Update: Apparently this is quite interesting to a number of people. I have rescanned the manual, properly compressed it and which you may find it here.
Found this manual at one of the older transmitter sites:
Raytheon RL10 limiting amplifier manual cover
Entire manual is available for your reading pleasure here: Raytheon RL10 limiting amplifier
As this is an older design than either the Gates Sta level or the Collins 26U, it may not be as useful to tube audio enthusiasts.
Raytheon RL-10 Schematic diagram
The main issue with the Gates and Collins unit is the GE 6386 remote cutoff triode used, which were great tubes, but very difficult to come by these days. This design calls for a 1612 or 6L7, which is a pentagrid amplifier. Feedback is provided by the screen of the following stage, a 6SJ7GT. Anyway, perhaps it will give somebody some idea on how to make a good tube compressor limiter.
The finished product:
SAS Rubicon console, WAJZ Albany, NY
This is the finished product from an earlier post. Currently, it it the studio for WAJZ in Albany, but that is not permanent. The SAS studio goes together fairly quickly, as most of the trunking between the TOC and studio is done over the SAS data channel.
The studio monitors (Tanoy Reveal) are set on little posts under the computer screens. I like this set up as the DJ’s are less likely to rock the house if they decide to crank up the volume on their favorite tune. I am also kind of digging the lack of a table top equipment pod. That takes up a lot of counter top space and always seems to be in the way. There are two CD players rack mounted below the counter (lower left), which are almost never used.
It is time, once again, to replace some very old Pacific Recorders BMXII consoles. The Pacific Recorders consoles were very expensive when new, but after 30 years of continuous use, have more than paid for themselves. The replacement console of choice for this installation is a SAS Rubicon. I have installed these units elsewhere and they are the modern equivalent of the PRE BMX.
The heart of the Rubicon system is the 32KD router. Routed audio systems can save a lot of time and effort in a large studio facility installation. Not having to run and terminate multiple analog and digital trunk cables between rack room and studio is a huge deal in a six or ten studio installation project.
The SAS 32KD router and Rubicon console system uses a serial TDM buss to communicate and transport audio around. This is a simpler system than packet switched IP data. Basically, the console surface is a very large, fancy computer control interface. Here are some pictures of the start of the project:
New Studio room, furniture installed
This is the view from the entry door. The furniture was placed last week and the counter top cut in for the console. The furniture is made by Studio Technology. The pile of yet to be installed equipment:
New studio equipment to be installed
For monitors, we are using the Tanoy 602p near field monitor placed on the table top above the computer screens. This studio will not have a turret. Turrets used to be necessary to hold things like cart machines and CD players. These days the CD players are used so infrequently that it was decided to put them in the side rack under the counter top. Turrets also take up a lot of counter top space that can be put to better use.
New studio punch blocks
Punch blocks and power connections. The red outlets are isolated ground UPS type, the back outlets are feed by the emergency generator power panel. All electric wiring is inside of metal conduit. The punch blocks are the inputs to the SAS RIO link unit, one 16 pair analog audio cable and ten category 5e shielded cables. The cat 5e is used for computer and TDM data buss to the router.
New Studio Rubicon console
The SAS Rubicon console cut into the counter top and protected by plastic sheets.
Rack room with 32KD routers. This facility has 9 studios total plus a news room with three work areas.
SAS 32KD router on line
The SAS 32KD router. All audio from the automation systems, satellite feeds and other sources is connected directly to these units. This unit is on line for other studios that have already been converted to the SAS gear.
The imminent demise of ISDN has been talked about for some time. There now appears to be a date attached which makes it semi-sort of official. As of May 18, 2013, Verizon will no longer accept orders for new ISDN lines. They will also not make any changes to existing lines and will start charging more for the service.
Taking the place of ISDN will be a variety of Ethernet/IP based audio transmission methods. As technology evolves, this makes sense. The quality of ISDN and the bidirectional nature was a vast improvement over the old system 5/7/10/15 KHz point to point analog lines. The one downside, ISDN equipment was expensive and the service was expensive to install and operate.
High speed internet is available in almost every business and venue. Many times, there is no cost to access it and equipment is relatively inexpensive. Depending on the equipment, CODEC, and speed, it can sound almost as good as ISDN. For those opposed to using the public network due to reliability issues, there is always frame relay.
Time moves on, so buy your IP CODECS now.
Sound hound, making us look smarter than we should
For all of us that work at radio stations but are not programmers, Sound Hound great app for those “WTF is the name of that song?” moments. As I get older, this seems to happen more and more. These are either senior moments or I am just not keeping up with the new music today. Probably a little of both.
To use the application, one can play, sing or hum the song in question and if Sound Hound can match the audio to a known song, it will return the song title, version if more than one and artist to your mobile device. It will also provide links to lyrics, chart information, artist concert dates and Youtube videos, which is pretty cool.
It comes as a free version with banner ads. For those of us that hate banner ads, a paid version is available as well.
I fooled around with this for a while playing songs from youtube club videos. If the audio is not too distorted (some of those club videos are pretty bad), it will work. It will also pick out live performances (and include venue and date if available), club mixes, etc.
Best of all, it makes me look like a genius to my kids. Any help I can get in that department is most welcome.
FM and AM broadcast radio processing has gone through many iterations. At first, the main processing function was to limit the input audio to a transmitter and prevent over modulation. This was a particular problem with early tube type AM transmitters, where over modulation could create power supply overloads and kill the carrier while engineers scrambled around resetting things and hopefully pressing various buttons to get the transmitter back on the air.
Over the years, processors incorporated not just limiting, but compression, gating, equalization, clipping and so on all in an effort to keep ahead or at least abreast of the station across town.
Today, broadcast air chain processors come in all shapes and flavors. In addition to that, internet streaming stations have their own unique set of issues to deal with. The top of the line Telos Omina or Orban Optomod systems are great, however, they can set one back a pretty large sum of money. Enter then, the Stereo Tool PC based software processing program.
Stereo Tool sofware screen shot
The first difference between, say the Omina and Stereo Tool is the end user decides the hardware and basic operating system. The second difference is Stereo Tool comes with a free trial. Then there is the price difference, which ranges from about $48.00 US for the basic version, to $161.00 US for the basic FM version and finally $269.00 US for the full version (actual prices are in Euros, which will fluctuate day to day and the credit card company will likely charge an exchange fee). Add to that a medium speed (2 Ghz) Intel Pentium4 or better computer, 1 Gb or more of RAM, good sound card and it all comes out to a reasonably priced audio processor.
Here are some of the specific features for broadcasting:
- Hiss Removal Filter
- FM Hiss Removal Filter
- Automatic Gain Control (AGC)
- 10-band multiband compressor / limiter / clipper
- Phasing error (AZIMUTH) correction filter
- Stereo booster
- Bass booster
- Final limiter
- Distortion masking Loudness filter
- Lowpass filter
- FM pre-emphasis filter
- FM stereo encoder
- FM RDS encoder
- Composite limiter
Much more info at the Stereo Tool website.
The idea of PC based audio processing is new and interesting to most of us. The designer and owner of Stereo Tool, Hans van Zutphen, was nice enough to answer a few questions I posed to him via email:
PT: What prompted you to write audio processing software?
HvZ: Since I was very little I’ve always wanted to have my own radio station. I remember playing with walkie-talkies and trying to receive their sound on a real radio when I was about 8 or 9. I never really did anything with it until I found out in 2001 that you could easily start a webradio station – I actually found out because I was listening to a pirate station in my car which turned out to have a stream; within a week my own station was online.
Of course I needed a bit of processing for it, and I wrote some command line tools – a singleband compressor, a stereo to mono convertor that didn’t cause any loss of audio (I was broadcasting hard trance on a mono 56 kbit/s stream, and this was the only way to get a decent sound out of it), and some time later a multiband compressor.
In 2004 I left the company I worked for (ASML, they make machines to make computer chips, customers are companies like Intel, AMD etc.) to start working for Philips Healthcare, where I was going to work on image processing for X-Ray systems. I had 2 months of ‘spare time’ between those jobs, and I wanted to learn to program in Visual C++, so I decided to a GUI around my command line tools, and make a Winamp plugin out of it. I called it ‘Radio Tool’. I never really planned to do anything with it, it was just an exercise project.
About a year later I came across the Winamp site again and I saw that you could upload plugins. So I uploaded my program, now renamed to ‘Stereo Tool’ because a Google search for “Radio Tool” gave far too many hits. Within a week there were over 1000 downloads and a while later it surpassed 90,000. At that point I decided to create a new version, Stereo Tool 2.0.
For quite a while this remained a hobby project, I occasionally worked on it for a few months and then I wouldn’t look at it for months. But at some point I was approached by someone people who worked at a “real” (FM) Dutch radio stations who asked for some extra features – he couldn’t get the audio loud enough, and that’s how I got into clipping. Things started to get better, I learned more and more about processing, the number of downloads increased and people became more and more enthusiastic about it. At some point, after reading something about how an FM stereo signal looks, I thought it might be possible to output a stereo signal with a 192 kHz sound card, so I bought one and did some tests and it worked that same night, and within a few weeks I added RDS.
PT: Do you know, approximately, how many stations (AM/FM/internet) Stereo Tool is being used on?
HvZ: FM: About 500, ranging from small community and pirate stations up to large nation-wide stations which run Stereo Tool at a dozen transmitter sites. Streaming: Not sure, but definitely over 1,000, probably a lot more.
PT: I have read through the forums on your site, Stereo Tool looks like a very complete processing system. Any plans for new features, future upgrades, etc?
HvZ: Yes. I’m currently working on a new multiband compressor. The multiband compressor in Stereo Tool is still based on the code that I wrote in 2001 for my webradio station, which in turn was based on an even older version that I had used on 8-bit audio. It also has far too many bands. Because of this, the multiband compressor is currently the weak spot of Stereo Tool. In the last weeks I have made a new singleband compressor that sound a lot better, it actually outperforms other compressors I have tested, and I expect great results for the new multiband compressor, which will also have less bands. Something else that I’ve been planning for a long time is a composite clipper, which will add 1-2 dB of extra loudness and especially better highs. Stereo Tool can already be louder with good audio quality than nearly any hardware box on the market (see for example this video, Radio 538 uses an Orban 8600 http://www.youtube.com/watch?v=4VpfcqUPQys – unfortunately due to the mpeg compression it’s a bit difficult to compare but listen for distortion ) – but there’s always room for improvement.
PT: What are the advantages of a PC software based processor vs. a hardware based (e.g. Omni or Optomod)?
HvZ: Ah, good question. Not sure if it’s the right question… With processing, a lot of things come down to taste, and there are several stations that have replaced their hardware processing by Stereo Tool not because it’s software and PC based but because they preferred the audio that comes out of it. Stereo Tool is also one of only 2 processors that contain a declipper (the other one is the Omnia 9, I licensed my declipper to them). For a demo of the declipper see: http://www.youtube.com/watch?v=oqOljvx9KaM
Also, Stereo Tool contains a stereo and RDS coder, most other processors don’t, so instead of having a whole bunch of devices everything can be done in a single PC, which also results in a better quality. Recently I added a new feature that enables synchronizing multiple FM transmitter signals that all connect to a simple Shoutcast stream (video: http://www.youtube.com/watch?v=GYQ5CYs0ZX8 ), so you also don’t need any streaming hardware anymore.
Of course there’s the price. A hardware box that gives “similar” quality (of course every processor sounds different, and it’s a matter of taste, so it’s difficult to compare, but I’m assuming that things like low volume levels, gain riding, distortion and lack of clarity in the highs are bad) easily costs $10,000 or more. And you can always easily upgrade to new versions. If you already have a PC with enough spare processing power you don’t need to buy anything.
I know that some people at radio stations are ‘afraid’ of using a PC in their processing path, but based on feedback I get from the stations that run my software it’s completely stable – and of course if a PC does break, you can replace it with any fast enough PC you have lying around – you just need to put the proper sound card in.
But for development, the advantages are huge. If you use DSP’s, it’s usually a lot of work to even make a very small change. When I worked at Philips Healthcare, the image processing that had been done – without much changes – on DSP’s for many years was being converted to PC’s because of speed of development and price of hardware. Once the conversion was finished, the development speed increased dramatically and 2 years later the image quality had improved beyond anything that was imaginable with DSP’s. PC’s get faster every year, and you don’t have to do anything for that – for the same price the processing power that you can buy roughly doubles every 1.5 years, and if you pay more you can get even more. If you use DSP’s, you have to do a lot of work yourself, you cannot just ‘buy a faster DSP’. Testing things is very easy, I can write some code that does something new, post it on my forum and I’ll have feedback from users the next morning – with DSP’s that’s a LOT more difficult and it takes a lot more time. I’ve learned by now that everyone hears things in a different way, and occasionally there are groups of people who hear something they find very annoying while many other people (often including myself) don’t hear anything wrong with it at all. Especially in cases like this it’s really great to be able to quickly send new versions to several people all around the world for testing.
PT: Are there any particular sound cards that work better with Stereo Tool?
HvZ: Yes. For the best results, use the Marian Trace Alpha, with the ESI Juli@ as second-best choice (it needs calibration).
Thank you very much, Hans, for the interesting insight.
Checkout the videos, especially the declipper video, which is quite amazing. That will cleanup all but the most ham handed DJ mistakes.
PC based audio processing software is a great solution stations on a limited budget that cannot afford high end air chain processors. There are many LPFM’s, Part 15 stations and others that can get great sounding audio and RDS for a very reasonable price. Currently, the AM settings do not allow asymmetrical modulation, which is more of a US thing. There is some talk of adding it in a later update.
Aside from everything else, we have been working at WSBS, Great Barrington, MA installing a new Audioarts Air-4 console. WSBS is a small AM station (860 KHz, 2,500 watts day, 4 watts night) serving the Great Barrington area. They also have a 35 watt FM translator (W231AK) on 94.1 MHz which is highly directional. During the day, the AM station has a much better signal than the translator. After dark, the translator covers the down town area fairly well. WSBS has been on the air since December 24th, 1957 (Happy 55th anniversary!), broadcasting from a non-directional tower just east of town on US Route 7.
The format could be termed full service, in the old tradition. Music, professional sports, local news, network news and weather with coverage of special events like election night and so on. The station does local very well, and as such, is profitable and has a great community presence.
WSBS control room console
The air studio console was this rather tired out Broadcast Audio unit from the early 1980′s. It had certainly served its station well, but change was in the air, so to speak. Actually, we were getting worried about continuing to service this unit, as parts had become scarce about ten years ago.
New WSBS control room console
Thus, we moved the air studio to the production room temporarily and removed all the old equipment and furniture. We installed an Audioarts AIR-4, which is a pretty cool little console. The AIR-4 has four built in microphone preamps, a telco mix minus feed, two program busses selectable VU meters and so on. The control room rebuild project included a new counter top, adding extra microphones, headphone amplifiers, cleaning up wiring rat’s nests, installing new monitor antennas, rewiring a good bit of the rack room and so forth.
It was a little more involved than we first thought, however, it came out pretty well:
WSBS Great Barrington, MA control room
The carpenter will be back next week, after Christmas to install the sides on the studio furniture under the counter top. It is a small operation in a small market in Western Massachusetts, but they have a real, live station staff including two news reporters. Hey, what a concept! To be honest with you, it is a joy all its own to work at a real radio station, if only for a short while.
There is a lot of buzz about converged technologies and what not. I have always been a wee bit leery of bleeding edge technology, lots of money and time can be wasted there. Incompatibility between different manufactures equipment and protocols can cause major heartburn in all equipment life stages. See also: VHS vs Betamax. Thus, when many disparate standards are homogenized into one acceptable system for everyone, we all benefit and technology moves forward.
Audio over IP (AOIP) is moving into the general acceptance of broadcasters as a reliable system for studio construction. As with anything, there are pluses and minus to this development: First of all, packet switched data is more efficient and flexible than circuit switched data. For the purposes of clarity, an AES3 signal within a broadcast facility going from one piece of equipment to another can be defined as circuit switched data. Once the data is segmented, packetized and framed, it can be sent anywhere, over any LAN or WAN. This allows for greater connectivity between facilities and greatly increased delivery methods and redundancy.
The downsides are increased complexity in transmission, greater reliance software and delicate operating systems to process audio into data and deliver it, and Quality of Service (QoS) issues. Additionally, there are many different AOIP protocols and applications currently in use. To date, this is the current list AOIP standards that are used by various manufactures:
- Wheatnet – Wheatstone, inc
- Livewire – Telos
- Ravenna – ALC Networkx (Open source)
- Dante – Audinate
- CobraNet – Peak Audio
- EtherSound – Digigram
- N/ACIP – EBU
- Q LAN – QSC Audio Products
- AVB – IEEE, AVnu
Each system has different characteristics. A Livewire system will not talk with a Wheatnet system and so forth. This is because of differences in the transport layer encoding schemes. Some use UDP, some use RTP, some use a propriety transport protocol, and some may even use TCP (remember the 7 layer OSI model). It would be similar to having an analog Wheatstone console unable to send audio to an analog Optimod which would be unable to modulate a BE transmitter.
AES X192 is an effort by the Audio Engineering Society to set an Audio over IP interoperability standard. This is the direction that studio audio equipment is moving and indeed, broadcasting in general.
The X192 project endeavors to identify the region of intersection between these technologies and to define an interoperability mode within that region. The initiative will focus on defining how existing protocols may be used to create an interoperable system. No new protocols will be developed to achieve this. Developing interoperability is therefore a relatively small investment with potentially huge return for users, audio equipment manufacturers and network equipment providers.
Eventually, broadcast audio consoles will plug into a WAN and be able to source audio from all over the place, not just the local physical studio structure. This lends itself to the evolving wired or wireless IP delivery method in place of the current terrestrial radio broadcasting currently used. As such, I will be diving into the fascinating world of AOIP more in future posts.