Old 30th May 2006, 16:43   #1
Kresner
Junior Member
 
Kresner's Avatar
 
Join Date: Feb 2006
Location: Peru
Posts: 7
Send a message via AIM to Kresner Send a message via Yahoo to Kresner
Spectrum analyzer

Hi all:

I thought that in Winamp 5.22 you would have fixed the SA (Spectrum Analyzer) bug. In older versions it was centered. I would like to know if there's a plugin I must download or do I have to setup something special.

I'll stand for your reply.

Greetings.
Kresner is offline   Reply With Quote
Old 28th June 2006, 16:04   #2
molyere
Junior Member
 
Join Date: Jun 2006
Posts: 2
I have the same problem...I did not upgrade from Winamp 5.12 because the Spectrum Analyzer does not show the right sound frequency or as you say, it isn’t centered. This is a bug or the new Spectrum Analyzer?
I’ll wait for your replay
molyere is offline   Reply With Quote
Old 28th June 2006, 16:35   #3
DrO
 
Join Date: Sep 2003
Posts: 27,873
it's intended design since the older method was in fact not accurate and ignored a large chunk of the fft data

-daz
DrO is offline   Reply With Quote
Old 7th July 2006, 18:56   #4
NotMyBest2Day
Junior Member
 
Join Date: Jul 2006
Posts: 1
I can understand fixing it to make it accurate, but the lower frequencies (the ones on the left) just stay maxed out all the time for me. Whether it's music with bass or so very little you wouldn't notice it, the bars stay up in the red. I've yet to upgrade from 5.13 due to this. I don't know if I'm alone on this, but I'm pretty particular about how the music gets analyzed.

Even if the old SA wasn't accurate, it looks like it was to me. The "accurate" version of it looks like it lumped 400Hz and under all into the left-most bar and the sensitivity got amplified by about 4x.
NotMyBest2Day is offline   Reply With Quote
Old 11th July 2006, 11:46   #5
ulillillia
Senior Member
 
ulillillia's Avatar
 
Join Date: Feb 2002
Location: Minot, ND, USA
Posts: 381
I've been reporting on this for a long time, even before year 2006 started. I use WAV files of otherwise random qualities. On the high end, as shown in the screenshot, I could have 130,000 Hz, 24-bit, mono, but on the low end, I could have 20,000 Hz, 8-bit, mono. With the same song, whether I use the hex edittor to change the sample rate to 150,000 Hz or 30,000 Hz (and yes, I know how to do this), these spectrum analyzer bars never change, even with a speed change as big as indicated. From 150KHz to 30KHz, the tempo drops to 1/5 of what it originally is and the frequency drops to 1/5 (making a 1000 Hz sound 200 Hz). Even with that, the spectrum analyzers never change how they are.

Screenshot #1 - This with a song at 130 KHz, 24-bit, mono has the bar on the far left almost always maxed out.
Screenshot #2 - This with a song at 36 KHz, 16-bit, mono has the bar on the far left almost always maxed out with the second being very close to maxed out. The bars on the far right pretty much never appear. The equalizer doesn't affect it at all.

Oddly enough, if, as a test, I open the song in Windows Sound Recorder, and convert it to use a much lower quality (i.e. 8000 Hz instead of 48,000), the spectrum analyzer bars behave normally.

Since there are supposedly 20 such bars, you could use a 20-step logarithmic scale from 20 to 20,000 Hz making the first one 24 Hz (the center point between 20 and 28 on this logarithmic scale), the second as 33 (between 28 and 40), then 47, 67, and so on to where the last one is 16828. If the frequency band it covers reaches the maximum amplitude (1 in Audacity's side panel), the bar is at the highest possible. If 1/3 of the way there (0.333 in Audacity's side panel), the bar would only get 1/3 of the way to the top. This is the mathematics behind how it should work, for any file type.

void BlueWater() {water.color=blue; while(GameRunning) {if (fox.pos == InBlueWater) {fox.air--; FoxDrown(fox.air);} else {fox.air=1800; fox.flags = WantsToGetWet; } WaitFrames(1); }} // My top favorite thing in 2D Sonic (as C)
ulillillia is offline   Reply With Quote
Old 6th September 2006, 21:17   #6
Kresner
Junior Member
 
Kresner's Avatar
 
Join Date: Feb 2006
Location: Peru
Posts: 7
Send a message via AIM to Kresner Send a message via Yahoo to Kresner
I feel so ignorant, but I did not understand that too much. You tested the SA with a wav file, but what about with a mp3 file? If you could give me a more simple solution how to fix it, I will thank you so much.

Greetings.
Kresner is offline   Reply With Quote
Old 20th November 2006, 00:08   #7
OdessaCubbage
Guest
 
Posts: n/a
I agree, with the last few versions (currently still a problem with 5.3), the classic SA display is skewed way to the left. The entire audible frequency range is crammed into the leftmost 20 or so bars (in thin mode), leaving the right 1/2 to 1/3 of the bars unused, never registering a value. I am using decent quality MP3 and WMA files (160-192 kbps or better). Perhaps we need an option to set the highest and lowest frequencies to be displayed, then everyone can set it as they prefer, whether audible or "correct"?

And it always takes me several frustrated minutes to remember how to change the SA options. Not sure why right-click->Configure Plugin doesn't get you to the configuration options. I think it used to, if I'm not mistaken. Very unintuitive now.
  Reply With Quote
Old 20th November 2006, 00:50   #8
DJ Egg
Techorator
Winamp & SHOUTcast Team
 
Join Date: Jun 2000
Posts: 35,821
That would be the way your files were encoded then.
What software/ripper/encoder/settings did you use?

All my Lame_Enc encoded mp3's show the full spectrum, though only the top quality ones, i.e. --alt-preset standard/extreme and higher, show much in the right-most 3 frequency bars - which again is expected behaviour, and accurate.

Though I do agree maybe that the default signal levels could be boosted a bit for the classic vis . . . it does seem a bit higher for modern skin vis. I find that turning up the preamp level a bit in the EQ helps, but of course, turning up too much tends to cause unwanted distortion, heh.

And by the way, the 'right-click > configure plugin' option has always taken you to the active vis plugin config, and not the built-in spectrum/oscilloscope vis options. Those are found via: Prefs > Skins > Classic Skins > Classic Visualization, and it's been like that ever since v5.0
DJ Egg is offline   Reply With Quote
Old 2nd December 2006, 18:07   #9
ColdCold
Junior Member
 
Join Date: Dec 2006
Posts: 6
yup, this is definitely a bug of some kind..

Hi all, I agree, this is definitely a BUG. I too have noticed it for a long time (maybe a year??) and today I searched the Net and found this thread specifically because I knew there was something wrong with this SA display.

Just to qualify my opinion, I'll mention I have a degree in electrical engineering, with a special emphasis on audio technologies and music recording. I have several years of experience with audio signal processing (dsp). On top of that I'm a musician (mainly guitar, some bass & drums). My opinion is, that the SA in WinAmp used to work just dandy and it doesn't now (at least not with the MP3s in my large library that all used to look fine in the WinAmp SA.)

I have not confirmed ulillillia's observation that at lower sampling rates the problem disappears. (Testing every permutation of encoding format and encoding parameters is someone else's job. errrr, ok, basically I'm just lazy.) For the record, all of my MP3s were encoded with LAME v1.30, at 48kHz, stereo, highest quality (q=0), variable bit rate (32-320kbps), VBR quality = 0, VBR method = default. I wonder, are those of you who experienced this problem also using variable bit rate encoded files?

Mind you, the music sounds as good as ever, of course, we're just talking about the SA display. The upper half of the spectrum looks like it's under a lead weight and the energy is getting squished down into the lower half. heheh

DrO, I'm curious, what are you referring to? Where did you get this information? What was inaccurate with the old method and what chunk of fft data was missing?

In any case, this bug needs investigation... how do we get the attention of the WinAmp developers?

- the Coldest
ColdCold is offline   Reply With Quote
Old 2nd December 2006, 18:15   #10
DrO
 
Join Date: Sep 2003
Posts: 27,873
my comment was based on what one of the devs said when things were changed

-daz
DrO is offline   Reply With Quote
Old 5th December 2006, 08:41   #11
Gaymar
Junior Member
 
Join Date: Dec 2006
Posts: 1
Here's my take on the issue.

Human hearing ranges from about 20 hz to 20khz. Our hearing is biased towards the mid range (~4kh being loudest). Winamp's old spectrum analysis was biased towards this midrange frequency range (and in my opinion rightly so.) The new analyzer however focuses on the entire range; while this may sound good, I don't think it is.

Take sound itself into consideration; there's the fundamental frequency of the sound, and it's following harmonics. The harmonics become more and more weak as the frequency increases - by the time we get above maybe 12,000 or so, the harmonics are so weak that the analyzer isn't showing them anyway. That's why in the new analyzer we see a big chunk of powerful harmonics hanging waaay to the left, and so little anywhere else. You can't even tell what's going on, whereas the old analyzer was focused on the more "important" content - the midrange. So while the new one may be showing ALL of the information, it's arguable whether that information is relevant or not.

I think there should at least be an option to change between the two. I liked the old one a LOT better. It was particularly useful for comparing songs I recorded and mixed to commercial recordings. I can't do that worth a damn with the new analyzer, because it's just not specific to the important content.

Just my two nerdy cents.
Gaymar is offline   Reply With Quote
Old 15th January 2007, 02:20   #12
ColdCold
Junior Member
 
Join Date: Dec 2006
Posts: 6
see for yourself

go to http://www.dogstar.dantimax.dk/testwavs/ and download the "3stepoct" test file. this file has 31 one-third octave steps from 20Hz to 20kHz. play it in winamp and watch the spectrum analyzer.

these are the 31 frequencies in the file (in Hz): 20, 25.2, 31.8, 40, 50.4, 63.5, 80, 100.8, 127, 160, 202, 254, 320, 403, 508, 640, 806, 1016, 1280, 1613, 2032, 2560, 3226, 4064, 5120, 6451, 8128, 10240, 12902, 16256, 20480.

if it were working correctly you should see just one or two bars at full amplitude start at the far left and end up at the far right by the end of the track, agreed?

instead, you can't see the left-most bar even flinch until the ninth note of the series (127Hz) and it doesn't reach full peak until around the thirteenth note, which is already more than one third of the way through the spectrum (at 320Hz)! the second bar does not even turn on until the 18th note (1016Hz). the left-most bar is still maxed-out during the 19th note of the series (1280Hz)! by the 24th note (4064Hz), the bars of the SA are only one fourth of the way across the display! then the bars take longer steps to the right as it goes, where only the last four notes make it to the right half the display. (what... is it on a linear scale instead of logarithmic?)

how much more proof do you guys need to admit that the spectrum analyzer is screwy ?!?!!!

please look into it!

-the Coldest
ColdCold is offline   Reply With Quote
Old 17th January 2007, 14:41   #13
Kresner
Junior Member
 
Kresner's Avatar
 
Join Date: Feb 2006
Location: Peru
Posts: 7
Send a message via AIM to Kresner Send a message via Yahoo to Kresner
Hi all again:

Now that I have read the posts, I should ask, what encoder/converter/ripper/etc software should I use?
I have this free one: Free Mp3 Wma Converter.

About lame_encoder, where should I find it? Is it easy to use? I must refresh the fact I'm not good in this.

Well, that's all I have to say for now. I'll stand for more support

Greetings.
Kresner is offline   Reply With Quote
Old 27th January 2007, 23:34   #14
ColdCold
Junior Member
 
Join Date: Dec 2006
Posts: 6
nobody cares if the spectrum analyzer displays garbage? what's the world coming to?

did anyone try the tests I suggested?
does anybody else agree that there is a problem here?
how do we get the attention of someone who can fix it?

(I keep posting here to keep the issue alive.)

TIA to whoever can help
ColdCold is offline   Reply With Quote
Old 29th January 2007, 09:32   #15
kzuse
Senior Member
 
kzuse's Avatar
 
Join Date: Oct 2005
Location: (D)
Posts: 480
Send a message via ICQ to kzuse Send a message via Skype™ to kzuse
I have tested it and fully agree to ColdCold. The current Spectrum Analyzer (WA 5.32) is crappy. Please fix it.
kzuse is offline   Reply With Quote
Old 23rd April 2007, 15:46   #16
xiradio
Senior Member
 
xiradio's Avatar
 
Join Date: Aug 2001
Location: Ohio
Posts: 170
I've been upgrading to new versions periodically when they become available, and end up going back to 5.13 because of this problem. Seems strange that no one has addressed it by now.
xiradio is offline   Reply With Quote
Old 30th April 2007, 10:20   #17
Tin Weasel
Junior Member
 
Join Date: Apr 2007
Posts: 1
I'm not sure this is the same issue, but I've read elsewhere that plugins have a choice of providing pre-calculated sa info to the display, or allowing WinAmp to do it for them.

As I understand it, WinAmp shows the frequencies on a linear scale, rather than the more common (in my experience) logarithmic scale. I think this would make the bars appear more skewed to the lower end (left side).

Apparently some plugins used to do it themselves, and used a logarithmic scale, but many now leave it up to WinAmp, and the resulting display (for many people) is odd.

Would this explain the results you're seeing? Or is this a different issue entirely?

- Tin
Tin Weasel is offline   Reply With Quote
Old 2nd June 2007, 15:26   #18
BuzzyKrongard
Junior Member
 
Join Date: Jun 2007
Location: In The Bowels of the Justice Dept.
Posts: 1
You Definitely are NOT Alone

Quote:
Originally posted by NotMyBest2Day
I can understand fixing it to make it accurate, but the lower frequencies (the ones on the left) just stay maxed out all the time for me. Whether it's music with bass or so very little you wouldn't notice it, the bars stay up in the red. I've yet to upgrade from 5.13 due to this. I don't know if I'm alone on this, but I'm pretty particular about how the music gets analyzed.

Even if the old SA wasn't accurate, it looks like it was to me. The "accurate" version of it looks like it lumped 400Hz and under all into the left-most bar and the sensitivity got amplified by about 4x.
I've been using Winamp for almost 10 years for its SA. It's truly one of the most accurate I've ever seen in ANY commercial application. I was going crazy for 3 days thinking it was my soundcard or something until I googled: winamp "spectrum analyzer" inaccurate... and found this page, thank God.

It puzzles me how something like this can get by any half decent/attentive quality control. I went back to 5.12 and everything is fine.

Winamp's Classic SA lets me know (or gives me a vivid idea, after years of cross-referencing) how the high-end of my music will respond on a high-end system and allows me to know just when enough is enough on the lower ends (if I'm looking for a super "hot", push-the-envelope bass line).

Thanks for the help guys...it probably would have been a week before I realized it was a bug and reverted back to an older vrsn.

Much Appreciated!
BuzzyKrongard is offline   Reply With Quote
Old 10th August 2008, 13:28   #19
j7n
Member
 
Join Date: Jul 2005
Posts: 75
Re: You Definitely are NOT Alone

So, is the spectrum analyzer considered fixed now? Then how come it performs worse than the old built-in MP3 one? The beautiful vis was the only thing that kept me from switching to the MAD plugin in Winamp 2.

The current generic FFT has too low frequency resolution to satisfy even the default small display. But the problem gets really painful if you decide to launch a fullscreen vis, like the Nullsoft Tiny Fullscreen (voiceprint). I tried to take a screenshot to demonstrate the issue but could not do it.

I asked you (DrO) once on Hydrogen Audio and I shall ask again: What did the in_mp3 do that made the animation look so good? Why couldn't the same process be applied to generic PCM audio? If the vis was specific to the mp3 format, why did you remove it from in_mp3?
j7n is offline   Reply With Quote
Old 10th August 2008, 20:29   #20
DrO
 
Join Date: Sep 2003
Posts: 27,873
i've no idea and i've never replied to the message at HA since i no longer have access to the account there.

http://forums.winamp.com/showthread....43#post2387043 is the only comment i can currently find from the dev team in print about the analyser implementation so if you've issues then you know who to moan at about it since i don't particularly care about the player and it's issues anymore.
DrO is offline   Reply With Quote
Old 18th September 2008, 01:30   #21
TRIUMPH
Junior Member
 
Join Date: Sep 2008
Posts: 1
Hey everyone, needed some expert opinions on spectrum analyzers, thought winamp would be a good place to start. I'm trying to figure out what the standard classic bar equalizer shows when 2 channels (right / left) are analyzed.

Do they show the channels side by side or an average of the 2? The reason I'm asking is because I've created an online mp3 player in flash with a visual EQ and want it to be accurate.

You can see my player here:
http://www.flashsavant.com/testingsite/mp32/mp3.html

I made the first song the "3stepoctive" file ColdCold speaks of, the rest of the songs are of the Metal variety. Sure would like an opinion, and ultimately an answer to the above question about both channels being an average or both channels being separate.

Thanks
TRIUMPH is offline   Reply With Quote
Reply
Go Back   Winamp & Shoutcast Forums > Winamp > Winamp Bug Reports

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump