[colonelkernel8]

toastmaster, put my logo in and then start a new SongView.org thread.

 

[DJGeorgeT]

Quote:

Originally Posted by Nerull /img/forum/go_quote.gif Hey guys, I'm using audacity right now, and in actuality, some of the waveforms do not actually correlate with sound quality. One notable offender in sound quality, Take Me Out by Franz Ferdinand, (and it sounds like utter crap on my headphones), measures well on Audacity.

If you can post the waveforms you are seeing, that will help in the future.

There are many factors that determine the quality of recording. It could have been recorded poorly, but mastered...well...ok and yet sound utterly like crap.

Quote:

In some cases, it does sound like crap according to graph (Opeth and noodles, sadly, for example.) And it does sound good, according to graph (The Orchard by Rose Melberg.) But, one of my best quality CDs (revealing the best of my system in sound stage, nuance, and placement of instruments) is Architecture in Helsinki, which has waveforms that sometimes don't look too good. I'm pretty sure there isn't any compression either, because the instruments and sounds almost exactly as they should (rather amazing sounding really) and if I look for changes in my system in quality, I'll use that CD. Subatomicglue, also sounds really good (techno music) but measures really badly (perhaps that's just because it's techno though...). Some of the older Judas Priest stuff also sounds just horrible or is decent, but lacking a little in dynamics, but looks good on Audacity. What I'm trying to say is, sometimes it doesn't exactly translate into a bad CD.

We would have to see it and hear it to comment. Waveforms do not have to be pretty, but certainly not square. Mostly sine waves. [/quote]

Quote:

Other things like poor recording technique, heavy compression, and thrashing of dynamics sometimes take place and don't translate into the waveform on screen and you don't notice until it reaches your ear. For some things though, I do agree, the crappiest can be seen in full glory by looking through Audacity. Some others though can slip by, so I suggest playing the song before actually deciding if it does indeed sound like crap. I would post pictures, but it's way too cold in my computer room and my hands are freezing...so good day for now! ~Tom

In order for songview to succeed, we would need to post both waveforms and a sample clip (not the whole song).

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @@@@@@@@@@@@@@

One cannot assume the following:

a) Sound must be pretty good because its waveform plot does not show clipping. The fact is...it could have been recorded poorly and heavily compressed or mildly compressed with a lot of compression distortion.

http://www.sendspace.com/file/n8nj87

It should sound good, but it doesn't because of poor recording and overcompression.

b) Sound must sound like crap if waveform plot shows that it is clipping. The sound might have been through some mild flattening/limiting if not mild clipping and it could still sound very good. As long as clipping is not severe, or heavy compressed/limited, you could still enjoy these recordings.

@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @@@@@@@@@@@@@@@@@@

However, if you see no daylight on the waveform plot, then I would not buy it becuase it probably combines heavy compression/limiting with clipping or clipping with heavy compression/limiting

previously posted:

Good post!

 

[colonelkernel8]

DJGeorgeT, your analysis is exactly what we need at the new SongView project wiki. Go on and help us out. http://www.songview.org

 

[DJGeorgeT]

Quote:

Originally Posted by colonelkernel8 /img/forum/go_quote.gif DJGeorgeT, your analysis is exactly what we need at the new SongView project wiki. Go on and help us out. http://www.songview.org

Of course. I am on it.

Let me propose a logo I have in mind. Maybe sometime tonight, I'll post it.

 

[colonelkernel8]

I already made a logo, but if you have something better...

 

[BigLu89]

"Jean Luc Ponty" has some of the best sound quality of any band I've ever heard. I might be biased because I love their music, but, the music is so clear, every detail is audible, giant variations in volume, everyone should hear what I'm talking about.

 

[colonelkernel8]

Check out this thread. http://www.head-fi.org/forums/showthread.php?t=219198

 

[SeagramSeven]

All of this is pointless if it is not done properly. The overall waveform view only tells part of the story. It must be accompanied by a spectral view.

In audacity, click edit---->preferences, and point to the "Spectrograms" tab.

Set max frequency to 22050. (Default 8000). If your desktop resolution is 1024x768 or less, use FFT size 128. For 1280x960 or more, use 256-FFT. And on the remote chance you are running some massive resolution, above 1920x1440, use 512-FFT. Click ok once those two options are set.

Open a waveform, click the drop-down arrow on the left side, and select Spectrum.

And the result:

This spectral view is what you must use to identify the vast majority of "problems" in the source material.

 

[adanac061]

Quote:

Originally Posted by SeagramSeven /img/forum/go_quote.gif All of this is pointless if it is not done properly. The overall waveform view only tells part of the story. It must be accompanied by a spectral view. In audacity, click edit---->preferences, and point to the "Spectrograms" tab. Set max frequency to 22050. (Default 8000). If your desktop resolution is 1024x768 or less, use FFT size 128. For 1280x960 or more, use 256-FFT. And on the remote chance you are running some massive resolution, above 1920x1440, use 512-FFT. Click ok once those two options are set. Open a waveform, click the drop-down arrow on the left side, and select Spectrum. And the result: This spectral view is what you must use to identify the vast majority of "problems" in the source material.

Thats a good point.. assuming that songview is about anything else other than dynamic compression .... adding the spectral view into the mix might be outside the scope of the project???

 

[Corrupt^]

Hey I just noticed this thread and er...

I don't really understand alot of all that technical mumbo jumbo but If I'd upload a part of a song to my webspace, would you guy's be able to test it for these flaws?

I downloaded a couple of songs from E.S. Posthumus some time ago and I liked it but the lack of detail in the mp3's annoyed me so I figured I go buy the CD. Now even with the CD I hear distortion at some points, and it's not my gear and it is, quite frankly, pissing me off.

 

[SeagramSeven]

People do not need to know what everything means, all they need to know is how to activate the spectral view (configured to 22050 hz), and take a (preferably high-resolution) screenshot. Of course, anything is better than nothing, but considering it only takes 2 mouse clicks to activate spectrogram view (assuming you use audacity), I see no reason why it is not feasible.

(And sure Corrupt^, go ahead and upload

)

 

[Corrupt^]

Haven't got time to upload it right now but I took a quick peak in Sony Soundforge and this is what came up:



And this is straight from the CD!

The Song is E.S. Posthumus - Pompeii

 

[DJGeorgeT]

Spectral analysis is useful if you have the input and output waveforms to compare. For instance, if I wanted to determine the THD of an amp, I can inject a 1 kHz sine wave into the amp and measure the output. Spectral analysis of the input will show all energy concentrated at 1 kHz. However, the output waveform will show mostly all energy at 1 kHz, but some energy at 2 kHz, 3 kHz, etc. It is a lot more complicated if you have music instead of a single 1 kHz sine wave because I do not have input/output to compare.

I believe it is best to just look at the dB vs. time plot to make a quick determination if a song is severely clipped, compressed/limited or otherwise good.

 

[DJGeorgeT]

Quote:

Originally Posted by Corrupt^ /img/forum/go_quote.gif Haven't got time to upload it right now but I took a quick peak in Sony Soundforge and this is what came up: And this is straight from the CD! The Song is E.S. Posthumus - Pompeii

Severe clipping. However, it could be worse and I have seen worse. When I used to master really hot, I would compress the snot out of the track. It would look clipped, but it wasn't. I am not sure if you are hearing high pitch distortion. Having said that, both are undesirable. I am no longer for compressing the snot out of a track, but then I hear from recording/mastering people that as long as the track is not clipping, then it is OK.

 

[slwiser]

Please provide a definition or explanation of what is being shown with the spectral view.

It looks like to me it is showing the frequency weight of each sample of a passage which has little to do with the loudness issue as I understand it. The scale on the left is frequency and the density of the image seems to indicate that where the tone balance is coming from not the intensity/loudness of the sound.