Samples for a new multiformat public test

[IgorC]

It will be useful if people will submit a samples for the upcoming public listening test.

Only one condition. It should not target one particular codec but rather be an interesting in a context of testing lossy encoders.

[IgorC]

MPEG sequences

MPEG have used these sequences in their verification test of latest audio format USAC(aka xHE-AAC).
This file is a concatenation of different samples: music, speech and mixed (music+speech, background + speech etc).

[darkbyte]

I would like to see the U96 stereo pre-echo problem sample in the test: www.hydrogenaudio.org/forums/index.php?showtopic=101993
Although it's reported as a problem sample for Opus i've noticed that it's a difficult one for the other ones aswell.

That "MPEG sequences" longsample is very good by the way

[win32vb]

    I would like to see some ambient nature sound samples included in this listening test.  I have recorded many good ones, but practically all of my recordings were done directly to Opus and are thus illegible for the test.

    Searching the Internet for good samples was going nowhere until I found a goldmine at www.freesound.org/.  Currently, I am looking through their library of nature sounds and making a list of unique, quality (and possibly difficult to encode) sound samples.

    My question is, how many individual nature sound samples do you want to include in your test?  For instance, we could have a "birds" sample (melodious singing and chirping characteristics), and a "insects/frogs" sample (mostly trilling and monotonous tones).  Each sample could contain short snippets of several different recordings.

    What do you think?

[kamedo2]

Jan 1, 2014 20:07:02 GMT IgorC said:

MPEG sequences

MPEG have used these sequences in their verification test of latest audio format USAC(aka xHE-AAC).
This file is a concatenation of different samples: music, speech and mixed (music+speech, background + speech etc).

Where did you find it? It's a great testing material.

[IgorC]

darkbyte
Your sample is interesting and it's in a big list.

win32vb
It's possible to include 1 of an ambient samples. Max 2 if they're interesting enough.

kamedo2
These sequences were available during that test.

[win32vb]

    IgorC, I have two ambient nature sound files for you to evaluate.  Both of them are 50 seconds long, and each one crossfades between four different scenes.  One sound file focuses on musical nature sounds (like birdsong), and the other on percussive nature sounds (like insect and frog calls, as well as a bubbling brook).  The files are in 24-bit, 48 kHz FLAC format.  I carefully selected only the highest quality recordings, and had to throw away several "lossless" recordings after looking at the spectrogram because they were obviously from a compressed source.

Ambient Nature (musical).flac
Ambient Nature (percussive).flac

    If these aren't interesting enough to warrant two samples/recordings, let me know your inclination as to which scenes to keep and I'll gladly make you a single file.  Also, if you would rather 44.1 kHz sample rate or 32-bit PCM instead of 24-bit FLAC, let me know.

[Gainless]

2 more samples from me. The first one is very aggressive-dusbteppish, the second has some interesting mouth organ sounds in it.

Deal With It (Sample)
All I Really Want (Sample).flac (915.25 KB)

[IgorC]

Thank You gainless and win32vb. Sorry if I don't answer to all. We take into account all your samples and suggestions.

[eahm]

How big/long do the samples need to be? I just saved two awesome pieces from a Swedish band, let me know if they are "good enough":



dropcanvas.com/btq8k

[eahm]

Jan 9, 2014 19:00:27 GMT eahm said:

How big/long do the samples need to be? I just saved two awesome pieces from a Swedish band, let me know if they are "good enough":



dropcanvas.com/btq8k

Ops, my bad, selection done.