HDV Audio: It's data compressed! at DVinfo.net

Ty Ford · February 5th, 2005, 08:48 AM

OK, fine, I'm mostly an audio person who's slowly drifitng over to video. I know that data compressed video in what seem like staggering ratios goes mostly undetected. I have great admiration for the engineers who figure out how to squirt that much data (reduced though it is) onto an HDV tape and make it look as good as it does.

Back when compressed audio became possible, an outcry based on real problems with concantated audio compression algorithms was heard.

So you have an ISDN audio session. That means you data compress the stream with some form of APT-x (or mpeg) data compression. That audio goes into a DAW and the mixed master uses say MP3 (even high bitrate) for distribution. Now your audio is on the radio or TV station hard drive where it might be compressed by their format. Hmm the transmitter and master control are not co-located, so the material takes a hop on a digital STL (studio to transmitter link) to get to the transmitter. The digital STL also has it's own compression. That's concantated algorithms.

So the question is, does anyone have any hard science or anecdotal info for what happens to the audio and video even after 1-2 of these layers?

Back in the 1990s, The degenerative effects of multiple audio algorithms ws demonstrated at the AES show in NY. On the second level, the audio was obviously degraded and things went south fast after that. As a result, I've been very picky about data compressed audio. Knowing that the audio on the Sony HDV is data compressed makes me a LOT less likely to go there.

Has the technology been refined so that the device receiving the compressed stream knows it's compressed already and doesn't apply any further algorithms? Is that posssible in ALL parts of the possible distribution chain? Do we need to be concerned

Your thoughts.

Regards,

Ty Ford

Graham Hickling · February 5th, 2005, 08:36 PM

>>I know that data compressed video in what seem like staggering ratios goes mostly undetected.

Hardly. Everyone is acutely aware of the artifacts that come with DV, and now HDV.

>Has the technology been refined so that the device receiving the compressed stream knows it's compressed already and doesn't apply any further algorithms?

Transfer to a digital file for your NLE computer system via firewire is completely lossless. What you do with the mp2 audio from that point on is entirely up to you.

Personally, with my JVC I usually convert any live sound to 48KHz wav, mix and edit it with wavs from foley, soundtrack etc, and have no particular concerns. And if I was being really picky I wouldn't use the on-camera mic anyway - I'd use an off-camera mic going to DAT.

Chris Hurd · February 5th, 2005, 08:54 PM

Those who follow my occasional rants on this board are already aware that I consider the 384Mbps audio bitrate to be completely acceptable for a consumer-level video format, which is what HDV is.

Peter Jefferson · February 6th, 2005, 03:42 AM

its also the minimum bitrate required for Dolby Digital production/delivery and mixing. Any lower and you wouldnt be able to license through Dolby.

A. J. deLange · February 9th, 2005, 10:24 AM

What happens when compressions are concatenated depends on the compression algorithm. For example if you use Huffman encoding on data which you have previously DCT compressed there is no additional loss because Huffman encoding is lossless (this is exactly what DV does). Thus if an STL is designed to forward a bit stream and uses a lossless encoding/compression scheme what comes out the other end will be the same as what went in and there will be no degradation. It is impossible to say what the systems implications are without knowing the details of the system.

February 5th, 2005, 08:48 AM	#1
Ty Ford Inner Circle Join Date: Nov 2004 Location: Baltimore, MD USA Posts: 2,337	HDV Audio: It's data compressed! OK, fine, I'm mostly an audio person who's slowly drifitng over to video. I know that data compressed video in what seem like staggering ratios goes mostly undetected. I have great admiration for the engineers who figure out how to squirt that much data (reduced though it is) onto an HDV tape and make it look as good as it does. Back when compressed audio became possible, an outcry based on real problems with concantated audio compression algorithms was heard. So you have an ISDN audio session. That means you data compress the stream with some form of APT-x (or mpeg) data compression. That audio goes into a DAW and the mixed master uses say MP3 (even high bitrate) for distribution. Now your audio is on the radio or TV station hard drive where it might be compressed by their format. Hmm the transmitter and master control are not co-located, so the material takes a hop on a digital STL (studio to transmitter link) to get to the transmitter. The digital STL also has it's own compression. That's concantated algorithms. So the question is, does anyone have any hard science or anecdotal info for what happens to the audio and video even after 1-2 of these layers? Back in the 1990s, The degenerative effects of multiple audio algorithms ws demonstrated at the AES show in NY. On the second level, the audio was obviously degraded and things went south fast after that. As a result, I've been very picky about data compressed audio. Knowing that the audio on the Sony HDV is data compressed makes me a LOT less likely to go there. Has the technology been refined so that the device receiving the compressed stream knows it's compressed already and doesn't apply any further algorithms? Is that posssible in ALL parts of the possible distribution chain? Do we need to be concerned Your thoughts. Regards, Ty Ford __________________ Ty Ford's Audio Bootcamp Field Guide Ty Ford's Audio Video Blog

February 5th, 2005, 08:54 PM	#3
Chris Hurd Obstreperous Rex Join Date: Jan 2001 Location: San Marcos, TX Posts: 27,368 Images: 513	Those who follow my occasional rants on this board are already aware that I consider the 384Mbps audio bitrate to be completely acceptable for a consumer-level video format, which is what HDV is. __________________ CH Search DV Info Net \| 20 years of DVi \| ...Tuesday is Soylent Green Day!

February 5th, 2005, 08:36 PM	#2
Graham Hickling Trustee Join Date: May 2004 Location: Knoxville, Tennessee Posts: 1,669	>>I know that data compressed video in what seem like staggering ratios goes mostly undetected. Hardly. Everyone is acutely aware of the artifacts that come with DV, and now HDV. >Has the technology been refined so that the device receiving the compressed stream knows it's compressed already and doesn't apply any further algorithms? Transfer to a digital file for your NLE computer system via firewire is completely lossless. What you do with the mp2 audio from that point on is entirely up to you. Personally, with my JVC I usually convert any live sound to 48KHz wav, mix and edit it with wavs from foley, soundtrack etc, and have no particular concerns. And if I was being really picky I wouldn't use the on-camera mic anyway - I'd use an off-camera mic going to DAT.

February 6th, 2005, 03:42 AM	#4
Peter Jefferson Inner Circle Join Date: Apr 2003 Location: Aus Posts: 3,884	its also the minimum bitrate required for Dolby Digital production/delivery and mixing. Any lower and you wouldnt be able to license through Dolby.

February 9th, 2005, 10:24 AM	#5
A. J. deLange Major Player Join Date: Jun 2004 Location: McLean, VA United States Posts: 749	What happens when compressions are concatenated depends on the compression algorithm. For example if you use Huffman encoding on data which you have previously DCT compressed there is no additional loss because Huffman encoding is lossless (this is exactly what DV does). Thus if an STL is designed to forward a bit stream and uses a lossless encoding/compression scheme what comes out the other end will be the same as what went in and there will be no degradation. It is impossible to say what the systems implications are without knowing the details of the system.