Gordon Monro home
Link to News page
Link to Gordon Monro's blog
Link to Downloads page
Link to Bio page
Link to GommoG page
Link to Music page
Link to Images page
Link to list of pieces
Link to Talks page
Link to list of CDs
Link to Archive page
Link to contact info
Link to search box

Audio and video formats

I have to deal with a number of different audio and video formats. Here I describe each one briefly, according to what it is used for. This list is not meant to be comprehensive, just to describe formats that I actually use, or am trying to use. A little while ago I moved in part to the Macintosh platform and acquired a new set of software tools (the Final Cut Pro suite).

I am not an expert on formats; if anything here is wrong, or I am missing things, please tell me.

 

I am not an audiophile

I am not an audiophile. For distribution I am happy to go with CD-quality audio, 16 bits per channel at 44.1 kHz. At present I am not looking to go beyond this (well, to 48 kHz for DVDs).

For processing sound I tend to use more than 16 bits in the intermediate stages (often 32-bit floating point).

Even if I am happy with the CD standard, common sound formats for download from the Internet and for DVD surround sound are below CD quality. So there is something to consider.

 

Audio for surround sound

By "surround sound" I mean multi-channel sound with discrete channels, typically with five speakers and a subwoofer (5.1). There seem to be some other uses around that I haven't sorted out.
 

With pictures

For me, surround sound with pictures means a video DVD.

Dolby Digital - The common surround sound format on DVDs. Highly compressed. Also known as AC-3. Not to be confused with Dolby Pro-Logic, which is an earlier (analogue) system. AC-3 isn't only for surround sound; it is also capable of compressing stereo soundtracks so that they occupy less space.

DTS - (Digital Theater System). This is less compressed than the commoner Dolby Digital (AC-3). Most DVD players now handle it, but older ones and cheaper ones may not. The encoder is not cheap, but it is not totally out of reach.

I have more or less decided to stick with AC-3, as it seems to be good enough. By all accounts the audible differences between AC-3 and DTS are subtle and easily overshadowed by differences in playback equipment, listening conditions, etc.
 

Without pictures

DVD Audio - Pure audio format, written onto a standard DVD. For my uses, this is an uncompressed format.

AC-3, DTS - As for surround sound with pictures, but with blank video.

DVD Audio requires a DVD player which will handle this format. There are a couple under $300 Australian. However, a DVD-Audio player has six analogue outputs, and the surround amplifier therefore needs six analogue inputs, which many surround amplifiers do not have. The reason for the six analogue outputs on the player is apparently that the digital sound connection (S/PDIF) cannot handle the data rate from a DVD-Audio disc. (There is also a copy-protection issue in the background.)

I have the (cheapish) Minnetonka Bronze DVD-A encoder. This is bare-bones, with no facility for still pictures (which the DVD-A standard allows), and no support for the highest-quality formats. However it does support six channels at 24 bits and 48 kHz, which is plenty for me now. Encoders for the full DVD-A spec cost several thousand dollars.

The SACD (Super Audio CD) format from Sony and Philips seems to be completely out of reach for home studios, and it is not clear that it is better than DVD-A.

 

Stereo sound

CD - still going strong, and an uncompressed format.

DVD - a standard video DVD will support uncompressed stereo at 48 kHz, so better than CD quality. However, to save space on the disc and transfer bandwidth off the disc, stereo sound can be compressed using the AC-3 format.

 

Audio downloads

These need a highly compressed format, because ordinary stereo CD sound takes about 1 megabyte per 6 seconds.
 

Stereo

MP3 - Very widely used.

AAC - This format is used by QuickTime and other software from Apple. On PCs, Winamp and Foobar (at least) handle it. Unlike MP3, AAC has support for Digital Rights Management (i.e. you paid for it, but you don't own it).

Ogg - A free format in every sense, but unfortunately seems to have stagnated. Under Windows, Winamp and Foobar support Ogg. See www.vorbis.com, which also explains the Mac options.

Since I now have Apple software, I now use AAC.
 

Surround

In the past I looked at putting up surround sound files on the Internet, but I am not pursuing this now. I have the impression that there is not much interest in this. If I'm wrong, please let me know!

 

Video formats

DVD: PAL versus NTSC

I would like to work with PAL only (not NTSC). PAL is the standard format in Australia, and I understand that it is better than NTSC. It also seems a little easier to work with. I am hoping that if my work is played at any North American festivals, the organisers will have access to a PAL DVD player. Unfortunately a DVD cannot mix both formats, so a compilation DVD has to be either all PAL or all NTSC.

I have twice tried to convert PAL to NTSC. The first time I paid a bureau to do it; the second time I used Final Cut Pro. The results were not wonderful either time.

For DVD the only reasonable video format is MPEG-2 (though there are options within that format, including PAL/NTSC).

At this stage I don't want to think about high definition: HDV/HDTV/HD-DVD/Blu-ray.


Downloads

For downloading video from the Internet there is a big problem. Any practical video format is already heavily compressed, but is still too large for comfortable download. I have put up some video extracts using the H.264 codec (which is part of the MPEG-4 standard) at 320x240 pixels (less than a quarter of the original number of pixels). High quality cannot be expected.

I have not used Xvid or Divx; I understand they are based on the H.263 codec, which is being superseded by H.264.

 

Note on compression

Audio compression

Unfortunately, in the audio world "compression" has two different meanings. On this page, by compression (as in "compressed format") I mean a process which makes the size of audio files smaller. For CD-quality originals this usually means reducing the quality of the sound, sometimes severely.

The other meaning of "compression" in the audio world is a process where large changes in volume are flattened out, during the recording process or in the studio; this is particularly done for radio broadcasts. This other meaning has nothing to do with file sizes.

Some sound formats, such as CD audio, are not compressed.

Format Compression factor
DTS Roughly 2.5 to 1 (high quality option).
Dolby Digital (AC-3) Roughly 10 to 1.
CD audio Not compressed.
DVD-Audio Need not be compressed if original at 48 or 44.1 kHz.
Internet audio (MP3, etc.) Around 10:1 for CD-quality originals.

As a general rule, more compression means lower quality, though there seems to be some argument about whether DTS is better than AC-3. If lossless compression is possible then the original sound can be recovered exactly. For CD-quality originals it appears that the best lossless compression can do is about 2 to 1, which is nowhere near enough for downloading in reasonable times.

The DVD-Audio format allows an encoding called Meridian Lossless Packing (MLP). At present to buy an encoder is very expensive. It is only needed if one is working in surround sound with sampling rates above 48 kHz, or if one wants to get more than 80 minutes of surround sound on one DVD-Audio disc.
 

Video compression

In the video world, as far as I know "compression" refers to reducing file size. All practical video formats are compressed: if each frame of video were transmitted uncompressed it would require at least 25 megabytes per second. Video on a DVD normally occupies one megabyte per second or less. The video format on DVDs is the MPEG-2 format, but this allows various options for compression.

Whatever options are chosen, the video on a DVD is compressed from the original frames by a factor of 20 to 40.

When video is further compressed for the Internet, the compression factor is huge. The video extracts I have put up for download are compressed by at least a factor of 12 from the DVD, so the compression from the original uncompressed frames is something like 300 to 1. The much smaller picture in the extracts is one consequence.

 

Note on codecs and money

Generally two pieces of software are required to deal with a given format, one to encode a recording to this format, and the other to decode the format in order to play it back. The two pieces of software are together known as a codec (for "coder/decoder", or maybe "compressor/decompressor").

Unfortunately, most formats are covered by patents, which means that typically an encoder to create the format costs money, sometimes a lot of money. The decoder is often free for non-commercial use, or comes built into the DVD player or whatever (though the cost of the hardware is likely to include a licence fee).

MP3 encoders come in quite a few commercial products. The use of MP3 compression is subject to patents and licensing fees, including fees for the commercial distribution or streaming of MP3-encoded material. (Small-scale distribution is exempt.) Some of the patents underlying MP3 are apparently due to expire soon.

AAC files can be created by Apple's iTunes (free download for Mac and PC). Dolby seem to be the prime developer. The format is subject to patents and licensing fees, though there is no restriction on distributing AAC-encoded material.

The really free format that does the same job as MP3 or AAC is Ogg Vorbis.

DTS encoders cost several thousand US dollars when Minnetonka Audio had the monopoly on them. Now one is around US $250 (go to www.surcode.com and click on "Products"). It appears that not all DVD authoring programs handle DTS.

Dolby Digital (AC-3) encoders are not particularly cheap either, though with this format all DVD authoring programs accept the results. Commercial video editing packages may include AC-3 encoders.

Meridian Lossless Packing encoders are very expensive. Minnetonka's costs US $2495 as of January 2007 (www.surcode.com again). This by itself will not allow you to make DVD-Audio discs using this technology; you will need a compatible DVD-Audio authoring program. DiscWelder Chrome will cost you an additional US $2995!

Commercial video codecs are included with commercial video editing packages such as ULead VideoStudio, Adobe Premiere and Final Cut Pro.

 

Open-source encoders

There are a number of projects developing open-source encoders for some of the formats mentioned above. I haven't used these, and the only one I have had any contact with is LAME, an MP3 encoder. This was developed from the standard independently of Fraunhofer/Thomson Multimedia (the main patent holders). After permitting LAME's use for a long time, Thomson Multimedia got aggressive about it, though LAME is still flourishing.

Other names of projects I have heard of are: ffmpeg, faac, x264, dffshow.

My understanding (which is minimal) is that most or all of these projects are developed direct from the relevant standards; that is they do not use proprietary software. They may still run into trouble with the patents. Depending on the country, it may be legal to download the source code for one of these projects but not to distribute a compiled version.

The legal situation seems to be a complete mess, with a large amount of fear, uncertainty and doubt fostered by the patent holders. It appears that even the patent holders do not necessarily know what the situation is: Thomson Multimedia say "We believe that some of the patents of Thomson are also used in AAC... We have not yet determined our licensing policy for this format." They are saying this in 2007, and the AAC standard was completed in 1997!

 

Reference

Some of the information for this page comes from: Tomlinson Holman, 5.1 Surround Sound: Up and Running. Focal Press, 2000. ISBN: 0-240-89383-3.


horizontal rule

© Gordon Monro 2007.       Last modified: November 12, 2007.
Home page:
Email: