The Telos Alliance

  • Home
  • Blog
  • Contact Us
  • Get NOW!
  • eNews

OVERPROCESSED CDs: WHAT YOU CAN DO ABOUT IT

Details

Category: Omnia
Created on Wednesday, 09 May 2012 17:35

OVERPROCESSED CDs: WHAT YOU CAN DO ABOUT IT

By Leif Claesson

Disclaimer: I have no ties or inside information with the music industry. All opinions are my own, and are  in no way an official stand point of Telos, Omnia, Axia or Linear Acoustic. My opinions are based entirely on listening to a lot of music over a long period of time.

We've all heard about loudness wars on the airwaves, but you may not be aware of the loudness war in the music industry, or at least not about how far it has progressed.

Nobody wants to release a song that is quieter than other contemporary releases. Few want to just match the loudness, and most seem to want to release their music louder than other contemporary releases. The reasons are not clearly known. Perhaps there is a misconception that a loud release will cut through better on the radio. Perhaps it's human nature. Maybe it will be a bigger hit if it's louder.

Radio is partly to blame as we've had our own loudness wars ever since the first distortion-cancelled clipper was invented in the 1970s, but enough is enough; it has gotten out of hand.

Frank Foti and Robert Orban collaborated on an article about the subject a few years ago. Anyone in the industry knows that when those two actually agree on something, we'd better perk up our ears.  Alas, nobody in the recording industry did.

Allow me to illustrate the progression of the loudness war in waveform (oscilloscope) pictures, with a few random but representative selections.

First, here is a kick drum from the original CD release of Dire Straits "Money for Nothing" from 1985. The lines are at 0dB full scale. Notice how they weren't worried about maximizing the volume at all -- the industry had just gained 30dB signal to noise ratio due to the transition from Vinyl to CD (reducing background noise by a factor of a few thousand to one). They had headroom which they could use -- or not! It sounded fantastic, and still does today, provided you turn your volume control up.

 

1985-money-for-nothing

[1985-money-for-nothing]

Next, we have Alanis Morissette's "Ironic" from 1995. Here, the waveform is maximized without being modified. 1995 was (depending on your perspective) a relatively long time ago. In technology terms, it's eons ago. GSM phones were just coming out, computers counted their hard drives in megabytes (not terabytes), and people generally connected to the internet with 14.4 Kbps modems. As far back as that was, that's when we ran out of headroom on CDs.

 

1995-ironic

[1995-ironic]

This recording is sonically unaffected by the loudness war, but to go any further, we'll have to do something about those pesky peaks.

Next comes Toni Braxton's "You're Making Me High" from 1996, an early loudness war casualty. Although quiet by today's standards, the kick drum is audibly distorted, and the vocals along with it.

 

1996-makin-me-high

[1996-makin-me-high]

Then came 1999, with Red Hot Chili Peppers and their Californication album. This was the first album where I personally took notice just from hearing it in the background at a party. "Did you blow your speakers or something?"

 

1999-scar-tissue

[1999-scar-tissue]

Although it doesn't look all that clipped by today's standards, it sounded horrific because of how it was mastered:  They simply turned it up. No limiting, no distortion-cancelled clipping, not even oversampling (to prevent aliasing) -- just turned it up. That, coupled with a sparse spectrum (clean bass line, clean electric guitar, clean voice) meant the clipping distortion cut through like nails on a chalk board, as there was nothing to mask it.

RHCP were pioneers in this type of mastering, but around 2003, most of the music industry followed.

Let's look at Green Day's "Wake Me Up When September Ends",  from 2003.

Here, every single kick drum is a squarewave. This changes the sound of the kick drum from a kick to a crunch, and it first distorts then mutes any other sounds for the duration of the kick drum. Sound waves are additive, and if slammed into the rails, something has to give. In the picture you can clearly see how certain sounds completely disappear at times.

 

2003-green-day-wake-me-up

[2003-green-day-wake-me-up]

In Mr. Foti and Mr. Orban's article about what happens to a clipped CD when played through a typical FM air chain, they mention that due to phase rotation and non phase linear filters, the clipped edges will not stay at the edge. They can in fact end up anywhere in the waveform, even right through the zero line. However, the detail that was lost in clipping is still lost, as evident from the next picture, showing the same moment in this Green Day song as it looks on the Composite Output of the flagship of one of the big processing companies:

 

green-day-typical-fm

[green-day-typical-fm]

The carnage is painfully evident just by looking at it. Imagine what it sounds like on the air. The clipping damage from the CD is clearly visible at the output, nowhere near the edges, and the (previously undamaged) edges now have more damage from the FM processing itself. No wonder TSL is down! And we're still only at 2003.

Fast forward to 2008 and look at Lady Gaga's "Lovegame".

 

2008-lady-gaga-lovegame

[2008-lady-gaga-lovegame]

And finally, unequivocally, we have the winner of the loudness war. Everyone else, put your mastering chains down. The war has already been won -- by Metallica and their Death Magnetic Album.  Witness "The Day That Never Comes" from 2008.  

 

2008-metallica

[2008-metallica]

You may notice how the lines aren't exactly straight. From what I've heard (read on the internet, so it must be true), the clipping was done in mixing, and was already a squarewave when the mastering engineer got it, so all he could do was to EQ it slightly (with a non phase linear EQ), hence the slightly wiggly lines. He didn't have Undo (this is what the Omnia.9's source declipping algorithm/program adaptive-multiband expander is called), as it would be another 3 years before it was invented.

 

That was the back story of Undo and why it is necessary.

On a more personal note, I stopped being able to enjoy contemporary music around 2003, so I got into Classic Rock instead. Steely Dan made some incredible music! I figured the industry would come around at some point and start mastering properly again. However, they didn't, so out of equal parts sheer desperation and hunger for new music, Undo was born.

Undo is two parts. A De-clipper created by Hans van Zutphen (licensed for the Omnia.9), and a program-controlled multi-band expander, created by yours truly.

Together, they work to both remove distortion and undo compression. All on-the-fly, set and forget, with (almost) no effect on material that was properly mastered. Undo cannot repair every mistake, but it goes a very long way.

Here are some actual examples in pictures. Please note that the oscilloscope is zoomed out for the Post Undo shots. The lines are still at 0dB, but 0dB does not actually equal full scale here as the Omnia.9 has 48dB of internal headroom.

Metallica, after Undo:

 

metallica-post-undo

[metallica-post-undo]

Red Hot Chili Peppers, Original CD (repeat):

 

1999-scar-tissue

[1999-scar-tissue]

Red Hot Chili Peppers, after Undo:

 

rhcp-post-undo

[rhcp-post-undo]

Green Day, original CD (repeat):

 

2003-green-day-wake-me-up

[2003-green-day-wake-me-up]

Green Day, after Undo:

 

green-day-post-undo

[green-day-post-undo]

Green Day (composite output of typical FM processor):

 

green-day-typical-fm

[green-day-typical-fm]

Green Day (composite output of the Omnia.9)

 

green-day-o9

[green-day-o9]

The original clipped edges (along with lost detail) are no longer evident in the output audio.

The output, although obviously clipped in the FM final clipper, does not sound distorted, and the detail is still there, both visibly as it is audibly.  Psychoacoustically Distortion Masking Clipping makes it possible.

But, that's another article!

                                                                                         leif tux

                                                                                         Leif Claesson

                                                                                         Creator Omnia.9                                          

Multipath? SSBSC technology may be your solution

Details

Category: Omnia
Created on Monday, 27 February 2012 18:55

ssbsc-dsbsc-notes

Recently, we were privileged to receive some interesting on-air results about our proposed alternative method of using SSBSC (Single Sideband Suppressed Carrier) as the transmission method for FM-Stereo broadcasting. The intention here is to see if SSBSC will reduce the effects of multipath within the coverage area of the broadcast signal. Our proposed method is fully compatible with existing radio receivers.

Greater Media did some on-air testing in Boston to compare the legacy DSBSC system vs. our proposed SSBSC alternative for FM-Stereo. Their chief engineer, Paul Shullins, is one of those top-of-the-line get-it guys, who'll give good thought to testing and proving a technology. What Paul did was to drive a well known multipath challenging route within the coverage area of 106.7FM, the Greater Media station there. He did something which further validates the testing, he ran video from the dashboard of his car, and captured the audio. This validates the location of the captured off-air audio.

Here are two videos (back to back), about 30 seconds in length:

http://www.youtube.com/watch?v=KlxeI2uqCOw

They were taken from the exact same points along the route. The first one, dated November 15, 2011, is of the standard DSBSC stereo system, and the other one is taken the next day on November 16, 2011, using the SSBSC method. Watch the videos and compare the results.

What you'll observe here is not an isolated case, but the first, we know of, where someone took the time to present an easy way to compare the difference between the two stereo transmission systems.

By the way, both the Omnia.11 and the Omnia.9 have switchable SSBSC capability as a standard feature.

For a detailed explanation on the subject of SSBSC technology, click here:

http://omniaaudio.com/downloads/white-papers/MPX-SSB-White-Paper.pdf

IP links: Ready for STL?

Details

Category: Telos
Created on Wednesday, 08 February 2012 19:20

This article guest-written by Telos Customer Support guru Ted Alexander.

Your STL is your lifeline. Everything you are travels over that STL. Ideally, the STL should be a “nailed up” connection, whether via microwave, or leased lines.  It should be always there, always on.

But, because of the rising costs of leased lines and other “nailed up” lines such as a T-1, or switched lines such as ISDN BRI, some stations are looking for an alternate path for their STL. If the station is willing to accept the slight risk of audio (and sometimes control) interruption in STL usage, a connection from ISP to Internet, and back to ISP can be considered.

So, is IP via the public Internet a worthy solution for STL? The answer right now is “maybe.”

Z/IP ONE IP CodecLet’s examine this scenario. Using the Telos Z/IP ONE, you can establish a direct connection using the TCSP protocol for connections between 2 units with static IP addresses and some guaranteed bandwidth (QoS) if you can get it. The connection between the ISP and the studio, and the ISP and the transmitter should be provisioned for the best possible reliability.

To set up an STL that uses the TSCP protocol, you will need two Z/IP ONEs, one at the studio and one at the transmitter, to utilize the advantages of Telos ACT error correction and concealment of anomalies in an Internet data connection. You also can connect by using the Z/IP Server (a free service Telos provides to Z/IP owners, allowing them to register their devices and get around NATS and firewalls) to establish the connection.

So, how do you connect 2 Z/IP ONEs together directly, without going through the Z/IP Server? You use a direct “TSCP” call which maintains all of the ACT advantages. This can also be useful for backup or temporary STL service, but if you want to set up a permanent STL, you should seriously consider asking your ISP to provide QoS service for this, which guarantees bandwidth. You will need your ISP to provide outside IP addresses for each of the two units in any case.

Configure each of the Z/IP ONEs’ WAN port network parameters with a normal internal IP address and the proper gateway address (usually the address of the Router). To call the other unit, press the CONN button and set the Device Name to the IP address of the other unit. Delete any “Group Name” that is present so that it indicates “(For TSCP Calls)” and set the “Call Type” to TSCP. You should now be able to connect to the far-end unit. Be sure to set the Codec parameters as appropriate for the available bandwidth. For STL use, “regular” AAC is suggested at 256kbps or higher (bandwidth permitting). It would also probably be a good idea to set General Settings / “Autoredial Broken Connections” to “Forever”. 

You will need to make available ports 24 and 308 (used for updating), port 11926 (used for listening), and ports 5060 and 5061 (used for SIP negotiation). Also, the default port for TCP is 5060, and 9150 for UDP.

Using the setup outlined here, it’s possible to achieve an STL link via IP over public Internet connections. However, remember that, without QoS, the IP link provided by your Internet supplier is reliable only so long as their network remains intact and uncongested!

While we don’t believe that the reliability of the public Internet is solid enough to support full-time STL connections yet, the steady increase of bandwidth everywhere indicates that the time will come when this is possible, if not inevitable. Right now, however, an STL connection via IP should be considered as extremely viable for use as a backup path to your transmitter.

tedbugTed Alexander
Telos Technical Support

 

Do you need audio processing in a file-based encoder?

Details

Category: Omnia
Created on Sunday, 08 January 2012 22:46

 

That question comes up on a regular basis                          .   

The first answer that comes to mind is “because it’s cool!” but, on a more serious note, there are some very practical reasons for it.

If all of our listeners were in the studio, and each one used a professional grade headphone set, we could arguably get by with less processing. The reality is that many listeners will play the audio through cheap PC speakers. They may be listening in a noisy environment, maybe in the car on the way to work, or on the bus or subway during the morning commute. Audio with a high dynamic range is difficult to enjoy under these conditions, as the user needs to keep adjusting the volume so they can hear the softer passages and not get their ears blasted by the loud passages.  For example, the multiband AGC of the Omnia F/XE audio processor takes care of this issue automatically, and ensures consistent levels. The listener no longer gets annoyed by large volume swings, the audio is more clear and easier to enjoy.

                                                                

A similar situation occurs on the production side. Podcasts may be produced from multiple audio sources. Each audio source may have been recorded in different locations and perhaps at different levels. Dynamics processing ensures consistent levels across and a consistent sound signature, not only in the one file but across all material you produce. And because the processing happens automatically as you encode the file, you spend less time in the audio editor.

The multiband dynamics processing can also improve the perceived audio quality as it can apply different adjustments to each band.  If you need to add more bass and loudness or are going for a more subtle effect, you can easily add some sparkle to the audio (as a matter of fact, “Sparkle” is one of the included factory presets on the Omnia F/XE).

Another reason for audio processing is that it can help the encoder with the conversion to MP3 or AAC. By processing the audio specifically for the target bitrate, you have the potential to reduce coding artifacts and get a better sounding file then when encoding the audio without preprocessing.

The best part is that, once you create your preset, you can automatically apply it to every file you encode. And you don’t need to start from scratch; we include a number of presets to get you started. You can use them as provided, or better yet, tweak them until you get the sound you like best for your application.

I mentioned above that you can automatically apply one of the processing presets to every file you encode. This is certainly the case, but our recommended choice -- the Omnia F/XE --  is more flexible than that. You can use it on a file by file basis or in batch mode with the File Processor application or you can automate the process with Folder Bot. In File Processor you create one or more “job” targets, where each “job” specifies the processing preset, encoder settings, and any post encode actions (e.g. to FTP the encoded file to a server). Then you drag and drop files on the targets to process and encode them. Folder Bot, on the other hand, sits in the background and watches one or more folders (that you specify) for new files. When a new file is detected, Folder Bot springs into action and automatically processes and encodes the file according to your configured instructions, in a completely hands-off manner.

In addition to keeping detailed logs, F/XE can also notify you by email, so you don’t have to sit by the computer and watch the progress meters tick away (although I do find this more exciting than, say, watching paint dry).

 

                                                                                                                       Ioan Rus        

                                                                                           Omnia F/XE  Research and Development

       

 

                                                                                                                            

Four "Must-Dos" For Planning your IP-Audio network

Details

Category: Axia
Created on Thursday, 05 January 2012 07:28

Although most engineers have some experience with computer networks, not all of us have had to build one from scratch. If you're planning for an Axia IP-Audio network (or maybe you've just purchased some Axia gear), here are four things you should do to help ensure that the audio network you're planning performs the way you want it to.

1) Clearly define your goals. How many studios do you want your network to support? Five? Ten? Just one? Knowing this at the outset will help you determine which network switch you should specify. The switch is the heart of any Ethernet-based network; buy too much switch and you've spent money needlessly - buy too little switch and you'll be wishing for more ports in short order. A list of Axia-approved network switches is found at www.AxiaAudio.com/switches/ , and Axia sales and support technicians will gladly help you determine which one is right for your needs. For smaller installations, you may not even need a core switch -- Axia integrated mixing engines have a switch built-in that can daisy-chain up to three studios.

2) Put it on paper (or PDF). If you're the kind who can visualize complex assemblies in your head, great! Even so, you've got to communicate what you see to others, so sketch it out. Sit down with your tool of choice, whether it be graph paper or AutoCAD, and lay out your studios. Place every audio component that you can think of, no matter how small, and connect them all. This will give you a clear picture of how many audio nodes, adapter cables, CAT-5 runs, etc. you will need -- and also give your team the opportunity to help you fill in items you may have missed. The finished product will be the roadmap you use to assemble the pieces when all those cardboard boxes begin arriving in your lobby.

3) Don't take shortcuts. IP-Audio networking, at its simplest, is the movement of packet-based data. In that respect, it doesn't differ much from your business LAN. Because of this, the temptation might arise to use network switches already on hand, or to buy cheaper switches with similar specs, or to try to let your studio audio "ride along" on your existing business network. Don't give in to this temptation! Although both networks use Ethernet for data transport, an IP-Audio network is a very different beast, requiring guaranteed bandwidth and Quality-of-Service (QoS) prioritizing to ensure that your real-time broadcast audio gets where it needs to go without interference. Networks and switches built only to handle email and Internet traffic simply don't have the smarts to keep your program audio flowing seamlessly.

4) Call for friendly advice. If you get stuck, don't be proud -- call us with your question. Chances are, it's something we've run across before, and we're always happy to help you past whatever roadblock's got you stopped. After all, our success depends on your success!

Clark Novak
Axia Audio

 

More Articles...

  1. Five things your competition is doing to sound better than you
  2. 4G Success with Telos Z/IP and ProSTREAM
  3. Retiring a king, and crowning a new one
  4. Cabling for Audio Networks: How to Save Time, Money and Trouble by Forgetting What You Know
  5. Audio Processing: Your Ears vs. Specs and Features

Page 1 of 3

  1. Start
  2. Prev
  3. 1
  4. 2
  5. 3
  6. Next
  7. End
feed-image Feed Entries
Back to Top

Telos-Systems.com | OmniaAudio.com | AxiaAudio.com | LinearAcoustic.com

All rights reserved. © 2012 The Telos Alliance.