samples for AAC public test
samples for AAC public test…
Herbie_Hancock.flac ( 1.16MB )
An excellent sample with rare (for music recordings) noticeable spectrum up to 24 kHz!
Excellent not only for AAC, but for any format encoding-decoding tests.
Will you please post its “original”, not FLAC-compressed, file?
Thanks in advance.
Will you please post its “original”, not FLAC-compressed, file?
Thanks in advance.
Don't you know that FLAC is lossless?
Will you please post its “original”, not FLAC-compressed, file?
Thanks in advance.
Don't you know that FLAC is lossless?
Is there now anyone who doesn’t know this?.. A good joke.
The aim of the question was to know the “initial” sampling frequency and spectral view of the original sample. For my inverse (uncompressing) conversions of the FLAC sample posted into Wav uncompressed with different sampling frequencies and bit depths gave unequal spectral results:
The FLAC sample posted itself, and “Wav”s obtained by conversion of the FLAC sample posted into “Wav 16/44,1” and “Wav 16/48” gave a spectrum bandwidth a little bit wider than 24 kHz (up to 25 kHz at some moments); and “Wav”s obtained by conversion of the FLAC sample posted into “Wav 24/48”, “Wav 32/48”, “Wav 16/96”, “Wav 24/96” “Wav 32/96” gave a spectrum bandwidth abruptly cut at 22 kHz. (Conversions made with “dBpoweramp Music Converter”. Spectra viewed on “CoolEdit Pro 2” “Frequency Analysis” window, and on “SillanumSoft Visual Analyser”.)
Thus, the experiments described show that in some cases the “inverse” (uncompressing) conversion of FLAC samples into Wav uncompressed with different sampling frequencies and bit depths do not always give spectrally equal results.
Herbie_Hancock.flac has samplerate=44100 Hz and bitdepth=16. Don't do any conversions, just decode.
The FLAC sample posted itself, and “Wav”s obtained by conversion of the FLAC sample posted into “Wav 16/44,1” and “Wav 16/48” gave a spectrum bandwidth a little bit wider than 24 kHz (up to 25 kHz at some moments)
So, WAV file sampled at 44.1 kHz contains frequencies up to 25 kHz? "A good joke".
(Maybe this discussion should be moved from this thread somewhere...)
As this is the samples collection thread I repost the links to the samples I'd love to see in the test:
harp40_1 (http://dl.getdropbox.com/u/2681777/ProblemSamples/harp401.flac)
trumpet_MyPrince (http://dl.getdropbox.com/u/2681777/ProblemSamples/trumpetMyPrince.flac)
lead-voice (http://dl.getdropbox.com/u/2681777/ProblemSamples/leadvoice.flac)
trumpet (http://dl.getdropbox.com/u/2681777/ProblemSamples/trumpet.flac)
herding_calls (http://dl.getdropbox.com/u/2681777/ProblemSamples/herdingcalls.flac)
keys_1644ds.flac (http://dl.getdropbox.com/u/2681777/ProblemSamples/keys1644ds.flac)
Là Ou Je Suis Née (http://dl.getdropbox.com/u/2681777/ProblemSamples/LaOuJeSuisNee.flac)
AAC and Mp3 have problems with the cymabls and synth on this sample at low bitrates.
I also have reposted some links in this thread. Since it is more suitable thread:
- Linchpin (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=4937)
- The Robots (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=5655)
- Musique Non Stop (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=5656)
- Show Me Your Spine (3:00 - 3:30 Edit) (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=5657)
- Human Disease (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=4794)
- Hexonxonx (http://www.hydrogenaudio.org/forums/index.php?act=attach&type=post&id=5593)
Berlin Drug from 4.6s-9s is pretty killer.
Berlin Drug from 4.6s-9s is pretty killer.
Ahh my sample .. Recently i found that it destroys vorbis up to Q6 - no need to abx as its obvious. Q7 is close to transparent.
Very annoying at 128 kbps.
Sample the should be difficult enough for AAC 128 kbps
P.S.: The highest bitrates i've seen so far with LAME -V5
Another sample for AAC listening test
Sample from musical Cats "Memories"
I actually planned to post this later, but since it seems to be default behavior on some Windows 7 installations, I think it's good if I post it twice.
To everyone doing ABX tests running at 44.1 kHz sample rate, especially those planning to participate in this one:
Play back this file on your computer!
It's a 10-second sweep from zero to 100 Hz. In the first two seconds, you should not hear anything.
If you hear a constant (but modulated) beep starting at the beginning of your file, read this
http://www.acourate.com/OperatingSystemsHa...SampleRates.pdf (http://www.acourate.com/OperatingSystemsHandlingOfSampleRates.pdf),
then correct your operating system's configuration accordingly. Windows 7: "Control Panel (German: Systemsteuerung)" -> "Sound" -> In "Playback (German: Wiedergabe)" tab, double-click the sound card you are using. In the last tab, select "16 Bit, 44100 Hz" or "24 Bit, 44100 Hz". Then verify that the above file sounds correct.
Chris
[attachment=5789:SoundCardTest.flac]
In case we need it, the Mandylion item originally posted here (http://www.hydrogenaudio.org/forums/index.php?showtopic=73344).
Chris
[attachment=5796:06_Mandylion_shrt.flac]
A couple of more samples. One of the samples is uploaded by request and the other sample was from a yet another Kraftwerk track that was used on a AAC ABX test at very high bitrates (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=74381&view=findpost&p=694909).
Also another suggesting would be The Missing (http://www.hydrogenaudio.org/forums/index.php?act=Attach&type=post&id=5587) by Ministry, which has smearing issues at low bitrates.
Thanks a lot, /mnt! I noticed that most of the Kraftwerk samples you uploaded in the past have a similar sound: isolated beats and high-frequency clicks or zaps (similar to the phasor sound on Star Trek). Most of the time you claim to hear pre-echo artifacts. Could you give a little more detail to us insensitive pre-echo listeners on where exactly in the spectrum (higher or lower frequecies) the artifacts occur and how they sound. Are these clicks or zaps where you can hear the artifacts?
Thanks,
Chris
Thanks a lot, /mnt! I noticed that most of the Kraftwerk samples you uploaded in the past have a similar sound: isolated beats and high-frequency clicks or zaps (similar to the phasor sound on Star Trek). Most of the time you claim to hear pre-echo artifacts. Could you give a little more detail to us insensitive pre-echo listeners on where exactly in the spectrum (higher or lower frequecies) the artifacts occur and how they sound. Are these clicks or zaps where you can hear the artifacts?
Thanks,
Chris
The artifacts on the Kraftwerk samples always appear on the high freq zaps or clips. Which usaully produces a more harsher and wet sounding zap noise. Also the artifacts may make the zaps sound smeared or it can present a "psst" noise at random places like on iTunes AAC and LAME.
Sadly pre-echo also effects Rock and Metal music aswell, since it can produce drum smearing. Which to me sometimes sounds like smashed glass when the drummer was hitting the cymbals, on a handfull of LAME V2 encodes.
I find that most of the tracks from The Man Machine and Electric Cafe / Techno Pop to contain plenty of tracks that have very sharp transients. Which brings lossy codecs such as Mp3 and AAC to its knees.
AAC frame size test data in Excel format. 7-Zip package.
[attachment=5801:fatboy_30s_xls.7z]
EDIT
Related to this forum post: http://www.hydrogenaudio.org/forums/index....st&p=695185 (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=77272&view=findpost&p=695185)
Here's a thing which worries me a bit. Might be off-topic, though, but here it goes.
Attached are the first 30 seconds of the old mastering of Kraftwerk's The Robots, amplified by a factor of 2 for better loudness matching to the other items. Listening to this in comparison to /mnt's upload of the remastered version makes me realize how much noise reduction has been applied in the latter. Unfortunately, to my ears, the noise reduction didn't always do good to the signal. Please judge for yourself. Here are also two spectrograms for comparison. Note how some of the horizontal and vertical lines (tones and clicks, respectively) have disappeared in the remaster, and some pre-echos have been added (esp. end of the section)!
I think that, for optimal testing of transient response of the encoders under test, we should revert to the old master. The added (or better, non-removed) noise might even be another useful encoder stress factor (since it's clearly audible, it "pulls the bits away from the actual signal," so to speak). /mnt, since you're the official Kraftwerk uploader what do you think?
Chris
Original Master: | 2009 Remaster:
[attachment=5803:robots_old.png] [attachment=5804:robots_remaster.png]
Download of old master:
[attachment=5802:Robots_old.flac]
I think that, for optimal testing of transient response of the encoders under test, we should revert to the old master. The added (or better, non-removed) noise might even be another useful encoder stress factor (since it's clearly audible, it "pulls the bits away from the actual signal," so to speak). /mnt, since you're the official Kraftwerk uploader what do you think?
I find the 2009 remaster to sound smeared on the sharp transient parts, which is mostly due to the noise reduction. Also most of the codecs i have tried seemed to be more prone with the 2009 remaster, which sometimes makes me wounder if i am transcoding from a lossy source.
The old master seems to encode more better dispite the extra background noise.
Vocal sample.
"Ectasy" is a killer sample for all AAC encoders at 128 kbps.
http://www.hydrogenaudio.org/forums/index....st&p=590404 (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=64410&view=findpost&p=590404)
AAC encoder bitrates (Excel sheet)
[attachment=5848:aacenc_b...0414_xls.zip]
The related forum post: http://www.hydrogenaudio.org/forums/index....st&p=700610 (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=77272&view=findpost&p=700610)
The "ABC-HR_bin" package from Sebastian's MP3@128 test.
[attachment=5859:ABC_HR_bin.zip]
The related forum post: http://www.hydrogenaudio.org/forums/index....st&p=701999 (http://www.hydrogenaudio.org/forums/index.php?showtopic=77272&view=findpost&p=701999)