A little sound demo for those who are undecided

[Lewis]

Ive noticed the amount of people recently asking 'what bitrate should i use' etc so i decided to make a little demo. I will add to this as time passes but for now i have ogg at 64kbps and 128kbps

www.freewebs.com/lewisfriend/ogg_q0_64kbps.ogg 429k

www.freewebs.com/lewisfriend/ogg_q4_128kbps.ogg 881k

EDIT mp3 bitrates

www.freewebs.com/lewisfriend/mp3_64kbps.mp3 453k

www.freewebs.com/lewisfriend/mp3_192kbps.mp3 1.3M

[andrewmel]

You may also find this page of interest. It teaches you to hear artifacts when encoding to lossy codecs....

http://ff123.net/training/training.html

But if you do learn this you may regret it as you may start hearing these artifacts more often. Ignorance may be bliss.....

z

[Lewis]

exactly, why would i want to do that???
I posted this for the benefit of the n00bs

[seadzz]

This may make for a good little project and should prove helpful to members here. Thanks for your effort. My only suggestion would be to link short segments limited to about 20 seconds of a given track.

In my experience using the Fobar2000 ABX plugins a short soft passage of music is the only real way to pull off this test. Listening to the entire song causes me to forget exactly where the artifacts are, short segments make finding them easier.

Thanks again,

sdz

[Lewis]

i thought that track was a good example of transparency. btw its off the Rayman PSX disc

[seadzz]

Lewis,

These tracks may be a good example but to tell you the truth i did not download them to try since my ISP is too slow to make it worth the time.

I really think a 15-20 second sound bite with piano, female voice, violin or other soft music is the best for ABX testing. Rock or other loud music tends to mask artifacts.

If you can cut the links to the identical 20 second chunks which would make for quick downloads and fast ABX tests.

I think new users would get a quick earful that way.

Many thanks for your efforts and I do not want to come off as dissing your work.

If this thread proves popular we may make it into a sticky and move it to the FAQ section.

sdz

[G-Rard_B]

To tell you the truth I can't really tell a difference between them.....

[andrewmel]

Originally Posted by Lewis

exactly, why would i want to do that???
I posted this for the benefit of the n00bs

While some of the artifacts are obvious, this doesnt help you tell the difference between 192k and 256k. Hearing some of these takes training.

I thought some might be interested. It wasnt meant to and doesnt detract from the usefullness of your post.

z

[DreamTactix291]

Well, when learning to hear artifacts in problem samples you tend to start at a low bitrate and move your way up. If you train yourself well you can end up hearing these artifacts when they show up at higher bitrates, though as a general rule they tend to be milder unless the encoder just fails with that sample.

[Sango]

SEADZZ I could make a bunch of sound clips if you give my time and a place to host them all.

[seadzz]

Sango,

Thanks for the offer. I would think that a 10 second passage of a give track that has easy to identify artifacts using either piano, female voice or other soft instrument is the best method to ABX test. These clips whould be the exact same in terms of part of the track. No loud R&R is needed for this test since loud music masks artifacts.

I would also make large jumps i.e. ever other bitrate range. This will be the best for gross id and users could always fine tune their own testing once they can learn to hear that 64K and 128k do not sound the same.

As where to host them, threre are any number of free sites on the net that allow for hosting things like this.....google to find or post a question in the forum on "free site to host pix or data" and someone here is bound to offer help.

May thanks for your effort....this is the type of thing that belongs in the FAQ section and will be of real help to all noobs.

sdz

[Sango]

I'll get on it. But I think I will try just to do quality codec tests. Becasue they already have that cool artifact test.

[Sango]

Done
the thread here
the site here