Syntrillium Backup Forums :: View topic

steve01 · Posts: 4

Hi everyone:
A friend just purchased CE Pro and has layed down some great acoustic tracks. It sounds really sweet! He has asked me to help him with some of the editing and transform features and so far so good.

One thing I don't get is the included spectrum analyzer. The one that is on CE Pro is probably just above my head. I was looking for something like a real-time bar graph that would show the amplitude of each frequency as the file is playing. Something like the analyzer in WinAmp. This would really help in ajusting the acoustic guitar's EQ settings while the song plays. I can imagine this feature displayed just above the included 30 band EQ. Now that would be nice!
Any thoughts?

syneryder · Posts: 56

In case you're not aware of it already, the Frequency Analysis graph in Cool Edit can update itself in realtime, just set the FFT value to 1024 and play your track. Voila, instant 1024 band graphic equaliser display!

I realise that's probably far too detailed for what you want to do though. Surely there must be a DirectX EQ plugin that will do what you're wanting? If I come across something I'll be sure to post again....

steve01 · Posts: 4

Thanks for the reply.

I'm sure the CEP analyzer is one of the best. But for me, I'm a little lost. I've never used an analyzer that is so advanced. To my unprofessional eye, (read-I'm not a broadcast or recording professional so I need the layman's crutch)it looks like just a bunch of squiggled (sp?) lines. I was just used to using a simple vertical bar graph with each band (20-20k) laid out below each bar. Is there a Direct-X Plug-in that might do the same that?......Syntrillium?

dkistner · Posts: 50

I'd love to see Syntrillium do a tutorial on how to read the Spectral and Frequency Analysis, Histograms, etc.

I think one point of confusion for me is I can't even figure out how to read the basics...the vertical graph on the wave form, for example! (I know, dumb, dumb, dumb.) I get very confused about what something like "-6dB" means. Minus from what? So I can't take full advantage of statistical analysis and the like for trying to come up with the best transforms for what I am doing.

I know Syntrillium must assume a limited range of knowledge in its users, but some of us just don't have it but want to confidently learn. Analagously, somebody who knows how to write words may assume words are made up of letters because that's so obvious to them they don't even think about it. But somebody who has never learned that little piece of information would be hard pressed to construct words. You can't do higher math or calculus without first knowing how percentages and fractions work. Etc.

Any "for dummies" links would be most appreciated!

oldman · Posts: 86

For a good article on decibels try http://www.prorec.com/prorec/articles.nsf/articles/EA68A9018C905AFB8625675400514576

dkistner · Posts: 50

Oldman, thank you for this link. I just saved it out to my hard drive to print it out. But one sentence caught my eye: "What's more, this lack of understanding isn't limited to beginners. There are folks I know who've worked in this field for a long time who aren't clear on the whole thing ..."

Now I feel a little less like a total BOOB! :)

Diane

dkistner · Posts: 50

I've had an epiphany. I read the article at Oldman's posting, and that helped (although I ain't even gonna try to do that math!), at least as far as the 0dB and -96dB stuff went.

The epiphany I had is this: The sample rate on the amplitude ruler (that my CE2K defaulted to) has no relationship to decibels--at least not that matters to me! I discovered I could change the display to either % or dB format, and that really works to clear up my confusion! I mean, I was trying to figure how in heck the sample rate was related to the size of the waveform and what that meant in terms of the transforms I was trying to use. I was flat out blasted CONFUSED!

There's hope for me now....

Diane

oldman · Posts: 86

Enjoy, but those are samples not sample rate.

Edited by - oldman on 02/11/2001 3:07:54 PM

dkistner · Posts: 50

Oldman, forgive my ignorance, but what is the difference?

oldman · Posts: 86

I am sure Diane that there are people on this forum that are a lot more qualifed than I to answer your question but I'll take a crack at it.

When you digitaly record a waveform your sound card does a number of pretty cool things. It divides the waveform up into many constant intervals and it assigns values to the magnitude of the waveform in each interval. In this manner one second of the waveform is represented by a finte number of values. The sample rate is the number of divisions or intervals taken per second of the waveform. If you record at 44100 Hz there will be 44100 samples of the waveform taken for ever second recorded. Magnify the data in the waveform display the maximum amount and you'll be able to see each data point.

The value of the waveform in the interval is determined by the analog/digital (A/D) convertor in the sound card. The A/D assigns a discrete number which represents the amplitude of the waveform within that interval. An 8-bit A/D has a capability of assigning one of 256 values while a 16 bit A/D has 65,536. These are the values that are identified on the Y axis, or "amplitude ruler" to the right of the Waveform display. They can be displayed as samples, percent of full scale, decibels, etc.

For an easy to follow discusion on the subject find the tutorial section of Syntrillium i.e. http://school.syntrillium.com/tutor/ and download "A Short Course In Digital Audio" audcours.exe

Edited by - oldman on 02/16/2001 10:34:46 PM

Edited by - oldman on 02/18/2001 3:16:47 PM

invader9000 · Posts: 299

And for everyone that is curiou about how on earth 44.100Hz came along with 16-bit resolution to make the CD audio, have in mind that the inventors of the CD, figured out that those values, should be the best "compromise" for Hi-Fi, non-compressed audio, which was su[posed to fit it's data into a 12cm disk. And Philips created the CD!
So, now the DVD-Audio standard comes in, to promise much more than Hi-Fi audio, but High End, as it has advanced features from CD. Fierst of all, the disk can store more data, meaning that music can be digitally put ot DVD with more accuracy, much better than 44.100 16-bit, but 96.000 24-bit! This, will mean much more UNCOMPRESSED data stored into the disk, because each sec of audio is "chopped" in 96.000 pieces, not 44.100 and analyzed in 24 bits, not 16 any more. Total conclusion! More precisely digitally recorded audio, and the most important; Maybe no difference any more from LP sound!
Cheers todigitall recording!!!

ScottF · Location: USA Posts: 3

There's a site out there @ www.7shades.com with a DirectX-based RTA (spectrum analyzer) for $99 and change. It's an app separate from CoolEdit. I've downloaded the demo but not had a chance to play with it. From what I've read, though, it can be setup to give you the look of a 1/3-octave EQ unit. Try before you buy -- it may be what you're after.

sf

mahana · Posts: 13

try, http://www.monumental.com/rshorne/gram.html this is the home site for a fine spectrum analyser. it costs about $25.00 in its current version , but earlier versions were free. the new version will display full screen.

SJM · Posts: 78

This topic has been visited a number of times in this forum (and I helped). Looks like there's a real desire for a simple plugin in that will do the following:
1) Sample the average frequencies generated by a wave and superimpose those points on a 30 band equalizer.
2) Allow the user to adjust his wave to meet those points.
Maybe I'm oversimplifying the process, but I'd welcome a plugin that did that and nothing more.
Thanks, StephenJ

jester700 · Posts: 546

RobertM · Posts: 299

Jester,

A couple of points cannot be disputed: 98db is certainly enough dynamic range for the human ear, and CD format can reproduce frequencies from 20hz to 20khz, which is beyond that which people can usually hear. So everyone sould be happy with the results, right? But remember that there is much more to sound reproduction than that.

In CD audio, the digital sample points are limited to 32,000 discrete values above 0 and 32,000 levels below 0, and these sample points are provided 44,100 times per second. These discrete points are then plugged into a mathematical algorythm in an attempt to reproduce the original analogue signal. The resulting signal is, therefore, an approximation to the original. The issue is how faithfully is the original signal recreated? How accurately are the most subtle of details reconstructed? Consider the sampling rate. A 20khz tone in CD audio will be re-created from only about 2 points per cycle. You can see this by using the "generate>tones.." feature in CE. Create a 20khz tone (a small fraction of a second is enough) and zoom WAY in. You'll see what I mean.

So, while the CD audio specs are quite valid, they can also be a little mis-leading. I've heard gorgeous sounding analogue equipement, horrid CD players, and vice versa. I'd like to think that the 96khz/24bit format would put the above concerns to rest, but you'll still have to extract the digital information, split it up into 2 parts (left and right), synch to a clock, run through a D/A algorythms, phase match the resulting 2 analogue signals and amplify them before leaving the back of a CD player. Still lots of opportunity to muck it up there.

Also, I'm not so sure about the "people talk about 25k & above all the time" thing. Even if the human voice box could produce such frequencies, I would call it noise, not speech, because no one could hear it.

jester700 · Posts: 546

The whole "missing music between the samples" thing is always proffered. It's baloney. Individual sample are meaningless - it's all about averaging over time. And proper digital systems with proper digital filtering capture ALL the same music that analog systems of the same bandwidth do (and analog systems ARE bandwidth limited). With proper dither & filtering, a digital system will reproduce EXACTLY the waveform that enters, within that bandwidth & dynamic range limit. The SAME waveform that a good analog system would reproduce.

There are reasons some digital gear sounds cruddy (especially ultra cheap or older gear), but it has nothing to do with limitations of the 16/44.1k format as a final distribution format; it has everything to do with poor implementation further up the line. If there is ONE example of 16/44.1 that is exemplary, it's enough to prove the format.

I mention the 25k & up thing because that's also often proffered (usually by the same people as the "missing music between samples" thing). I don't buy that one, either, particularly since it's often brought by old farts that can't even hear a 15k TV whine...

RobertM · Posts: 299

Sorry Jester, I goofed up on your last comment. I thought you were saying that there is >25k content in speech, but you were really referring to recording at greater than audible frequencies to allow "just that little bit of extra definition"? I'm not a big believer in that notion either. Another theory is that higher than audible frequencies will re-combine, upon stereo playback, into audible interference patterns, thus yielding better sound reproduction, etc. I'm not sure why those interference patterns would yield better sound.

As to the discrete samples issue, is it REALLY baloney? I agree absolutely with you about the problems of "poor implementation further up the line", and I even mentioned several examples of potential "down the line" problems in my post. These types of problems will exist regardless of the sample rate and bit depth. But is the mathematical representation of the analogue signal really IDENTICAL to the original waveform? I've always thought that it should not be too difficult to take an analogue signal (with no >20k content) and record it at some outrageous sampling rate and bit depth (high enough so that there really can be no question as to how well is represents the original waveform), then downsample to 44.1-16bit, upsample again and invert paste on top of the original. The downsample/upsample stage should take out ALL of the supposed higher definition, and if you are left with a flat line, with all noise comfortably below the -90bd region, or even just a file full of zeros, then I suppose there really is no difference between the analogue and 44.1 signal.

Say...(not totally unrelated)...I had a conversation with a salesman at a well regarded audio shop the other weekend, and he maintained that the most important part of a CD player is the transport; that the real difference between the top-of-the-line Linn and the one-step-down Linn is nothing more than the transport (and some thousands of dollars). He was quite firm in his stand that data taken from a CD is LOADED with errors, LOADED (he really emphasized this), so a better transport means fewer errors. I think he's an example of someone who's loaded: with baloney! He claimed no knowledge of PC data, but seemed comfortable with the idea that all PC data and executable files are equally error prone (is this where I'm supposed to interject a comment about our favourite OS?). At least he was right about one thing: he apparently knows nothing about PCs. But one thing which I was curious about: I have always presumed that the CD transport will load the digital info into a buffer, perform any required error correction (presumably correcting ALL errors along the way) THEN pass the info to the D/A sections. The D/A then does all of the sorting, sequencing, filtering, etc. This thinking makes me comfortable with the notion that any reasonably competent transport will do, and that the real money should go into the D/A box. But is there more to the transport than that? Could my 'friend' at the audio store be correct to a degree? What do you think.

Graeme · Member Location: Spain Posts: 4663

RobertM · Posts: 299

Thanks, Graeme, I agree with those points.

But what do you think about the errors that go uncorrected. Are THOSE common? My 'friend' was suggesting that MANY errors make it though the error correction stage, and have a corrupting effect on the D/A process. That was the basis of his thesis, that a better transport would result in fewer UN-corrected errors being presented to the D/A. I, on the other hand, think that ALL errors are corrected, otherwise, virtually ALL of our computer files would contain lots of corrupt data. Truly fundamental errors (scratches, dirt on the surface, mechanical shock to the playback system, etc.) WILL make it through, of course, but we were not arguing about that magnitude of error. I can't imagine that even mid-fi CD transports are worse at error correction than the run-of-the-mill 50x CD-R drives that we are stuffing in our PCs of late.

Further: would not ANY data error be VERY noticeable? I don't think that there's much subtlety in digital errors, the number will be wrong in a range anywhere from 0 to 64k, so could these un-corrected errors really result a hard-to-distinguish loss of detail, akin to the sound differences between two good quality amplifiers?

jester700 · Posts: 546

Robert - yes, you got my meaning Re: >25k correct. And yes, that's a common reason offered. What people forget is that analog is ALSO bandwidth limited. With real world mics, real world vinyl after a few plays (or tape at 15ips & lower), real world phono styli, & real world speakers, there isn't anything above 20k anyway.

With proper filtering & dithering, yes - the waveform is reproduced EXACTLY. The secret here is "within dynamic range & bandwidth limits". Dither allows there to be signal "in the noise" - like analog. Yes, the noise floor is raised a couple dB in the end - so what; the benefits are big. The music between the samples is simply higher harmonics, which are filtered by ANY bandwidth limiter, be it a digital antialiasing filter or an analog tape/head combo. The "timing" issue is a non-issue. Again - it's about averaging, and you can't forget the time domain.

Jitter - yes, a problem. Most modern circuits deal with it pretty well, though. MY problem with cheap digital is usually the cheap analog sections tacked on to the tail end so you can hear the numbers...

The sales guy is an idiot. PC data has NO errors - it can't. When it does, we have what is commonly referred to as a "crash".

I have a cheap $50 CDRom that can rip audio BIT PERFECTLY at 28x. I'm sure my Onkyo player has no problems with errors on normal discs. Now, imperfect discs may cause both to burp; my $50 PC one can re-read the data 100 times if it wants, the Onkyo can't. Here's where there's some difference. But that's not common, and it CERTAINLY ain't on every disc. Graeme's right - error correction takes care of most of this. Pro CD players sometimes have a BLER counter, and will show you when interpolation hits; it ain't often.

jester700 · Posts: 546

Data errors may not be as noticable as you'd think on audio. The first stage is correction, and CD audio has a couple methods to do this. The second, if correction isn't possible, is interpolation - the player "averages" the preceding & following data, kinda like you'd do on a graph if you wer missing a data point. If done enough, this is noticeable, but not with one sample.

If enough interpolation is occuring, the player decides "it just ain't happening" and mutes the audio (or skips, or...). Mostly you hear this on CDs from people who pull them off the floor of the car (sans case) and answer your look of incredulity with "hey, they're indestructable, right?" (musta been told that by the audio salesman from the other post ;-) )

RobertM · Posts: 299

Thanks for the info, Jester. Much appreciated.