Speeding up GOLDWAVE processing.

GoldWave general discussions and community help
Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Speeding up GOLDWAVE processing.

Post by Coriolanus »

I have been checking out Trial Versions of SOUND FORGE and AUDITION - and two things I have noticed.

1. When I save a file or use one of their preset noise reductions -- they process at almost triple the speed of Goldwave. Is there anything that can be done so speed GOLDWAVE up? I have had proccesses take up to 12 minutes and that is excessive to me.

2. They are very complicated to use and very expensive and I doubt I know enough to spend money on them over Goldwave but their manuals are very informative. SOUND FORGE seems better than Audition at least for repairing bad audio.

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

I meant to respond to this in detail the other day.....but that's the thing about me, i forget a lot of stuff.

I wouldn't say soundforge is better than audition in cleaning up noise. Sound Forge is just more "for dummies" than audition....plus the thing is, Audition supports plugins, i have NUMEROUS plugins to do noise reduction, each one has it's faults and strong points.

As far as processing time...this all goes back to Chris is one man working on an entire program by himself. The processes aren't really very optimized in the same way they are for the bigger programs. The bigger programs are bigger programs for a reason...they have MORE people on it. Seriously, Goldwave's development team is 1, SoundForge and Audition might have more 15 or 20 working on it. When you have 20 people working on a project, then each member spends more time focusing on a specific task...this can either cause problems...or, you wind up with some top notch coding.

The other reason you might be taking forever is you might be doing TOO MUCH processing..you don't always want to go full-force and crank everything up thinking that'll be an end-all, sometimes it doesn't work. The biggest lesson I learned in all the years is figuring out HOW MUCH processing to apply and only apply that amount.

I will say from my expirence that Audition does some things faster and slightly better than Goldwave....but i'm a firm believer in one can have multiple tools for multiple things (and i say that because I know Goldwave is a really good program and I actually owe all my talents TO goldwave...if it wasn't for goldwave, i wouldn't know how to use audition, i also enjoy helping the people here and don't want to seem like a traitor for using something else and possibly upsetting Chris...afterall, he's been really nice to me over the years the few times i've had to deal with him). I mean, for all my simple cutting and pasting audio and analyzing audio...Goldwave is the first thing I go for...I only use Audition when i'm doing serious work (the definition of serious work is bascially when i hit something that might take me a while to do in GW)

With the case of your files (which i'm still trying to get to a point of quality i'm comfortable with putting my name on) - the distortion is just too engrained in the audio to get rid of it...not outside of a frame-by-frame cleaning which would take HOURS...and while i've done stuff before, my going rate of $150/hr for audio restortion work is too steep for most people.

It boils down to this..Goldwave is great for beginners or simple tasks. I've seen it used in a studio (which was when I was using it for some post-production editing), I've noticed a few audio production courses talk about it...it's GREAT for getting the basics down....it's got a LOT more to offer than Audacity, which is a basic open-source audio editor. But, it's not a Pro-Tools killer...nothing can really be a pro-tools killer.

the other thing i'm wondering is do you happen to watch a lot of NCIS or CSI. I get people who have seen that show then happen to have some recording they want clean and get upset when the results aren't "perfect"....they're like "but, they took out worse noise than that on CSI" and i'm like "CSI is a TV show, they got all the noise out for a reason.

I've used the DC7 Audio Forensic lab before....even it has it's limitations. There's a LOT of those in audio.

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

DewDude420 wrote:
I
With the case of your files (which i'm still trying to get to a point of quality i'm comfortable with putting my name on) - the distortion is just too engrained in the audio to get rid of it...not outside of a frame-by-frame cleaning which would take HOURS...and while i've done stuff before, my going rate of $150/hr for audio restortion work is too steep for most people.

It boils down to this..Goldwave is great for beginners or simple tasks. I've seen it used in a studio (which was when I was using it for some post-production editing), I've noticed a few audio production courses talk about it...it's GREAT for getting the basics down....it's got a LOT more to offer than Audacity, which is a basic open-source audio editor. But, it's not a Pro-Tools killer...nothing can really be a pro-tools killer.

the other thing i'm wondering is do you happen to watch a lot of NCIS or CSI. I get people who have seen that show then happen to have some recording they want clean and get upset when the results aren't "perfect"....they're like "but, they took out worse noise than that on CSI" and i'm like "CSI is a TV show, they got all the noise out for a reason.

I've used the DC7 Audio Forensic lab before....even it has it's limitations. There's a LOT of those in audio.
NO I never heard of CSI.

But tell me -- I am willing to commit the time to clean up those two tracks -- could you give me some clue as to how I would do it.? What should I look for in each frame?

Yes SOund Forge does seem simpler than Auditions. I did read several reviews which said that Sound Forge does a better job of restoration than AUdition but then I don't know about the AUdition plugins and I didn't care too much for the interface. I like Goldwave's interface and mostly it does the clean up that I want on recordings. Those two tracks from WAXMAN are the first ones I haven't been able to clean up myself.

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

could you give me some clue as to how I would do it.?
yes and no..it's kind of hard to convey years of expierence. basically what you're doing is looking at the waveform and for any oddities or things that shouldn't be there....and to really get a grasp at that you've gotta look at a lot of waveforms. once you're able to isolate the area you need to fix you can either SOMETIMES run a simple interpolation...but generally the damage is more than a sample wide and in order to prevent all kinds of funky distortion you need to recreate the wave...most of the time magical code is able to do an ok job at it....you can have minimal success free drawing it in goldwave...i mean, overall it's complex.

i'm not trying to dodge you...i'm seriously not really able to tell you EXACTLY what to do becuase I myself don't know EXACTLY what i'd do to the files...i tried all the basic stuff already....and i don't really have the time to seriously dig and go "here's how you can fix the file".....i'd have to fix it before i could do that and i'm not even sure i'm able to fix it.

out of the box sound forge does a better job than audition...but that's out of the box...i've got a whole arsnel of plugins i've added to audition that drastically changes the behavior of various functions....in which case the quality of the processing is now dependant on that program. audition took me quite a few years to learn and i still haven't even mastered it yet.

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

Maybe I am wrong but I have found it easier to deal with problem if the files are in Mono.

I once did a file that had a lot of loud POPS in it -- standard pop/click file wouldn't deal with. But I discovered that those parts where the POPS were had very high points in the wave file and if I deleted those points I would delete the clicks.

I tried the same principal with the current tracks but often ended up with distorted sound -- same when I tried Mouse editing.

What do you mean by interpolation? What is this magical thing you talk about? I don't know how to slow down the speed so I can watch the waveform more closely to see when a problem shows up. How do I do that?

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

magical code...it's the pile of code that when compiled does FFT processing (which is what noise reduction is based off of) - i'm not a coder...i don't know anything about any of the languages so i can't even begin to tell you HOW it works.

you're on to something with the loud pops..i can tell you that. however, deleting them is NOT the best solution becuase you have to remember, you need to maintain the length of the original audio...the distortion is natural...you removed a bunch of samples that contained chunks of words and syllables and such.

interpolation is a simple effect in which the area between two "or more" samples is guessed..although it seems goldwave just basically connects the two samples with a straight line. when you have something like a pop or click, you'd select the spike in the waveform, hit interpolate, and it'll "smooth" out the resulting click without ruining the original construction of the audio. I tried that with yours, there's too much of it to leave ANY original audio after repair.

you can slow down the speed at which the sound plays back...but really, you have to zoom in to 1:1 view and manually look at the area of the file around the distortion...playing it slow enough to watch it won't tell you ANYTHING (your ears will make no sense of it, plus most of the clicks will be inaudible anyway)

it's basically like this...SOMEONE already processed your tracks.....you can't really do ANY MORE processing to it without DRASTICALLY ruining the sonic integrity...and i can't describe what sonic integrity is...it's all about the overall quality. you attempt to remove the noise, you begin eating away elements of the audio you want to keep and killing the sonics. i know you want those two tracks cleaned up...you know something..i've got a bunch of vinyl I really want cleaned up but....the key to being a good restortaion enginner is knowing when to say "i give up" - a beginner is going to go gung-ho and not give up till they solve it, and when they do, they're generally disappointed with the results.

seriously, i'm trying to save you a LOT of headache and a lot of work. you can work on this for the next 2 months and never find a result you're happy with...i'm telling you from expierence...ok? I was once just like you..I had no idea what the noise reduction did or how it worked or anything and it took me a good 8 years before i finally figured out in the logic in my head what it's doing how it's doing it and all that. the key to remember it's called noise *REDUCTION* not noise *REMOVAL*...nothing is going to remove 100% of the noise every time. don't let these two tracks be your white whale. you will SERIOUSLY drive yourself bonkers over it...not only that, but when doing audio...focusing so intently on ONE project is the worst way to learn anything..you're so focused on the outcome you want and not learning to be happy with what you can get. you can't learn any of this stuff out of fustration/desperation.

my advice to you if you wanna learn noise reduction is to put those files away...stop messing with them...grab some stuff you really DON'T care about and work with it...then when that gets boring switch to another task.

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

DewDude420 wrote:magical code...it's the pile of code that when compiled does FFT processing (which is what noise reduction is based off of) - i'm not a coder...i don't know anything about any of the languages so i can't even begin to tell you HOW it works.

you're on to something with the loud pops..i can tell you that. however, deleting them is NOT the best solution becuase you have to remember, you need to maintain the length of the original audio...the distortion is natural...you removed a bunch of samples that contained chunks of words and syllables and such.

interpolation is a simple effect in which the area between two "or more" samples is guessed..although it seems goldwave just basically connects the two samples with a straight line. when you have something like a pop or click, you'd select the spike in the waveform, hit interpolate, and it'll "smooth" out the resulting click without ruining the original construction of the audio. I tried that with yours, there's too much of it to leave ANY original audio after repair.

you can slow down the speed at which the sound plays back...but really, you have to zoom in to 1:1 view and manually look at the area of the file around the distortion...playing it slow enough to watch it won't tell you ANYTHING (your ears will make no sense of it, plus most of the clicks will be inaudible anyway)

it's basically like this...SOMEONE already processed your tracks.....you can't really do ANY MORE processing to it without DRASTICALLY ruining the sonic integrity...and i can't describe what sonic integrity is...it's all about the overall quality. you attempt to remove the noise, you begin eating away elements of the audio you want to keep and killing the sonics. i know you want those two tracks cleaned up...you know something..i've got a bunch of vinyl I really want cleaned up but....the key to being a good restortaion enginner is knowing when to say "i give up" - a beginner is going to go gung-ho and not give up till they solve it, and when they do, they're generally disappointed with the results.

seriously, i'm trying to save you a LOT of headache and a lot of work. you can work on this for the next 2 months and never find a result you're happy with...i'm telling you from expierence...ok? I was once just like you..I had no idea what the noise reduction did or how it worked or anything and it took me a good 8 years before i finally figured out in the logic in my head what it's doing how it's doing it and all that. the key to remember it's called noise *REDUCTION* not noise *REMOVAL*...nothing is going to remove 100% of the noise every time. don't let these two tracks be your white whale. you will SERIOUSLY drive yourself bonkers over it...not only that, but when doing audio...focusing so intently on ONE project is the worst way to learn anything..you're so focused on the outcome you want and not learning to be happy with what you can get. you can't learn any of this stuff out of fustration/desperation.

my advice to you if you wanna learn noise reduction is to put those files away...stop messing with them...grab some stuff you really DON'T care about and work with it...then when that gets boring switch to another task.
Just tried Interpolate with both Goldwave and Sound Forge - they both seem to work the same way, but I found Sound Forge more awkward -- I think for this file the trick may be to narrow the particular part of the form you are operating on.

But I like speed of saving and preset processing in Sound Forge --But I don't like the interface much -- but the noise reductions presets are very, very good. But it lacks a hum removal.

I know I am beating my head against a wall but I will keep trying.

I did have luck deleting very high peaks with one file -- but I am only talking about 2-3 peaks in a 1 Hours plus file - so it never affected the rest of the audio. The pop usually came in between or after words. But this file is more complicated.

How can you tell processing was already done on it? I was able to clean up tracks 19-21 which weren't that bad -- just a couple of places. But I will beat my head against a wall with this wone.

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

muffled sound...it varies on how it's muffled. generally the only way you can really work around it is with an equalizer...more specifically a very very tight parametric equalizer to adjust the bands....other than that, there's not much you can really do.

stereo/mono shouldn't really make a bit of difference..if you're THAT worried about the amount of time...then you can go to mono which will save you some time.

basic interpolation seems to work that way...but i've seen (and have) some interpolation plugins that will do some "reconstruction"

to be honest....one shouldn't ever really rely on "presets" for noise reduction...this is known as the cookie-cutter method of noise reduction. ideally you have to create your own "preset" depending on the amount of noise you have and the specific type of noise. i personally *never* use a noise preset, i always go by samples of noise i need to reduce...the problem is that noise is by it's very nature...very very random. so you can't really match up a type of noise from even within the same source...when the noise you want to remove is entirely random....a generic preset is only going to get close and an actual noiseprint is only going to get closer.

i can tell processing was done on it by looking at the frequency response between words. the first thing i noticed is it was ENTIRELY silent. this is not normal for an audio book recording..there's generally SOME kind of ambient noise in there...and even if it IS rather silent between the words...i could hear the leftover rements of processing the noise.

you're probably going to sit there and go "how can you tell"...it's like this..i've been playing around with this stuff a while...i've made stuff sound good, i've made stuff sound bad...i made sure i knew the negative effects of the effects i was applying...and when you spend a lot of doing this stuff you begin to identify things.

if you really don't believe me...maybe Doug or don can chime in (who have been silent and I'm kind of worried about them) and toss thier two cents in...

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

This may be a little off-topic but I would like to know about digital processing.

Basically these things on the PC maybe wave forms but those forms are really sets of 01 010101 -- justbits and bytes. What can't we process these files digitally - at the BIT/BYTE level.

If you know hex and octal it would be so much easier to see patterns etc and clean them up. I was an assembler programmer for IBM and I can read HEX like English so anything with bits and bytes is easy for me.

Even though retired, I still HEX edit files if I need to do a recovery of some sort or rebuild stuff. Why can't we do the same thing with AUDIO files -- I mean at their essence they are bits and byts no matter how they get represented visually.

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

i've never seen ANYTHING that processes digital audio in something that wasn't digital processing

...unless it was already converted to analog *eyes his equalizer and remembers that it's processing former digital audio in the analog domain*

digital audio IS 0's and 1's, but, here's how it works.

analog audio as you know is bascially just an electrical signal...an AC one to be specific. it's taken, amplified, run through a voicecoil where the amplified version moves a speaker cone...and boom...audio. it's so simple i wish i had a time machine to bring this technology back to the early 20th century and PROFIT (sorry, a little joke for the under 30's internet crowd)

anyway, in order to get this into the digital domain, a LOT has to happen..i'll break this down as simply as I can because, i don't know the specifics of how an analog digital converter works.

digital audio is made up of samples. each sample is a binary repsentation of a voltage level that USED to be an analog "waveform". this is where the term "sample rate" comes from..it's how many times a second the waveform has it's voltage level sampled.

now, each sample has a resolution..this is where the bit rating of digital audio comes in. most audio these days is done in the 16-bit domain..meaning it's using a 16-bit resolution to record each sample.

basically when say, an audio CD is played back. all this binary information is fed in to a decoder and it in turn converts the samples into a voltage...when do this 44,100 times a second (or 44.1 khz), you get rather acceptable CD quality audio.

in essence you are kind of manipulating the binary information directly...becuase you're manipulating samples in the bit stream...changing thier value, adding new ones, subtracting old ones...but seriously, you're dealing with a LOT of them...i mean, if each sample uses 16 bits of information and you're doing this 44,100 times a second....you can do the math, it's a LOT of samples.

the other thing is unless you REALLLLLY knew your audio, looking at a raw sample stream won't help you AT ALLL. it's literally impossible for a human to edit audio at the bit level (that i know of)....this is why we have software. the software converts all this into something we can manipulate easier....yes, you can zoom in and edit sample values directly and such....but you're STILL not working on raw bitstream....you're manipulating a sample in a WYSIWYG interface.

the reality is no one edits in 16-bit anymore..it has to many limitations. goldwave and most audio programs i've used now edit in 24-bit floating point...which is a LOT more accurate.

if you thought CD Audio had a lot of bits...well, DVD-Audio uses upwards of 192,000 samples per second with 24-bits per sample (integer, not floating point...floating point is really much only used in the mixing stage...i don't believe it's even possible to create a floating point capeable DAC..it all has to be converted to integer at some point....generally when you save, which generally causes negliable quality loss.

essentially ALL audio processing is just the manipulation of samples...it's just it's impossible to tell on a sample by sample basis what is noise and what is not...imagine a connect the dot with a LOT of stuff in the picture that doesn't belong...you can't really tell WHAT it is till you've drawn at least SOME of it out and look at it. most processing plugins work with chunks of audio at a time...so it's probably impossible to look at a series of samples and go "hey, that's noise"....because noise is just random movement of the waveform.

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

DewDude420 wrote:it's not exactly that simple.

digital audio IS 0's and 1's, but, here's how it works.

analog audio as you know is bascially just an electrical signal...an AC one to be specific. it's taken, amplified, run through a voicecoil where the amplified version moves a speaker cone...and boom...audio. it's so simple i wish i had a time machine to bring this technology back to the early 20th century and PROFIT (sorry, a little joke for the under 30's internet crowd)

anyway, in order to get this into the digital domain, a LOT has to happen..i'll break this down as simply as I can because, i don't know the specifics of how an analog digital converter works.

digital audio is made up of samples. each sample is a binary repsentation of a voltage level that USED to be an analog "waveform". this is where the term "sample rate" comes from..it's how many times a second the waveform has it's voltage level sampled.

now, each sample has a resolution..this is where the bit rating of digital audio comes in. most audio these days is done in the 16-bit domain..meaning it's using a 16-bit resolution to record each sample.

basically when say, an audio CD is played back. all this binary information is fed in to a decoder and it in turn converts the samples into a voltage...when do this 44,100 times a second (or 44.1 khz), you get rather acceptable CD quality audio.

in essence you are kind of manipulating the binary information directly...becuase you're manipulating samples in the bit stream...changing thier value, adding new ones, subtracting old ones...but seriously, you're dealing with a LOT of them...i mean, if each sample uses 16 bits of information and you're doing this 44,100 times a second....you can do the math, it's a LOT of samples.

the other thing is unless you REALLLLLY knew your audio, looking at a raw sample stream won't help you AT ALLL. it's literally impossible for a human to edit audio at the bit level (that i know of)....this is why we have software. the software converts all this into something we can manipulate easier....yes, you can zoom in and edit sample values directly and such....but you're STILL not working on raw bitstream....you're manipulating a sample in a WYSIWYG interface.

the reality is no one edits in 16-bit anymore..it has to many limitations. goldwave and most audio programs i've used now edit in 24-bit floating point...which is a LOT more accurate.

if you thought CD Audio had a lot of bits...well, DVD-Audio uses upwards of 192,000 samples per second with 24-bits per sample (integer, not floating point...floating point is really much only used in the mixing stage...i don't believe it's even possible to create a floating point capeable DAC..it all has to be converted to integer at some point....generally when you save, which generally causes negliable quality loss.

essentially ALL audio processing is just the manipulation of samples...it's just it's impossible to tell on a sample by sample basis what is noise and what is not...imagine a connect the dot with a LOT of stuff in the picture that doesn't belong...you can't really tell WHAT it is till you've drawn at least SOME of it out and look at it. most processing plugins work with chunks of audio at a time...so it's probably impossible to look at a series of samples and go "hey, that's noise"....because noise is just random movement of the waveform.
I have just been looking at some plain NOISE - basically low level hum from a tape recording in between the speech -- AND I see them same digital pattern repeated (FFx FEx). And IF I ZERO out the pattern (00x) with a hex editor -- I get a line with NO NOISE. I suspect that 00x is NO SOUND at all.

SO I am sure it can be done - the trick is spotting the digital bytes that comprise the noise and what makes up the speech. I expect it will take of lot editing and testing to see my results -Probably using just one small words until I can isolate the noise.

I know you think I am nuts but I really can see things with bits and bytes that make no sense to me with waveforms - I can see patterns. But as you said do I want to take the time for maybe 6 minutes of audio.

DougDbug
Posts: 2062
Joined: Wed Feb 16, 2005 3:33 pm
Location: Silicon Valley

Post by DougDbug »

This may be a little off-topic but I would like to know about digital processing.
There is a good FREE online DSP book (Digital Signal Processing) at DSPguide.com. I haven't read the whole thing, and I don't understand everything I've read, but I do believe it's a good book!

And there is a simplified introduction to digital audio on the Audacity website.
If you know hex and octal it would be so much easier to see patterns etc and clean them up. I was an assembler programmer for IBM and I can read HEX like English so anything with bits and bytes is easy for me.

Even though retired, I still HEX edit files if I need to do a recovery of some sort or rebuild stuff. Why can't we do the same thing with AUDIO files -- I mean at their essence they are bits and bytes no matter how they get represented visually.
Uncompressed WAV files are simple... just a sequence of samples. Each sample simply represents the "height" of the wave at a given instant in time.


If you want to make it "easy", open an 8-bit mono file with your hex editor. You can find links to the WAV file spec at Wotsit.org.

Again DewDude is right... It's far too much raw data to analyze or manipulate manually (i.e. 44,100 samples per second). As you probably know, a 16-bit, 44,00khz stereo file has 176, 400 bytes for every second of sound!

Plus, a couple of minor things make "visual analysis" of the raw hex data difficult. In a stereo file, the left and right samples are alternated, and the bytes are in big-endian (reversed) order. And, 16-bit WAV files use signed integers. You could probably write a little program to display the data in a more friendly format.
And IF I ZERO out the pattern (00x) with a hex editor -- I get a line with NO NOISE. I suspect that 00x is NO SOUND at all.
Right! A sequence of zeros is absolute digital silence. A 0dB (full scale) 16-bit WAV file has a positive peak of 32,767 (decimal) and a negative peak of -32,768... I'm not going to attempt hex, because I'd probably screw-up the negative number (one's compliment?). :P

P.S.
I just remembered, 8-bit WAV files use unsigned integers, but that doesn't make it any easier.... It's not so easy to "look at" an 8-bit file either... The waveform still has positive and negative half-cycles, but the data is not stored that way... Digital silence isn't zero in an 8-bit file... It's "half way"... I don't recall exactly... probably 7F or 80 (hex).
Last edited by DougDbug on Tue Dec 02, 2008 11:50 pm, edited 2 times in total.

Coriolanus
Posts: 181
Joined: Tue Apr 29, 2008 3:50 pm

Post by Coriolanus »

DougDbug wrote:
This may be a little off-topic but I would like to know about digital processing.
There is a good FREE online DSP book (Digital Signal Processing) at DSPguide.com. I haven't read the whole thing, and I don't understand everything I've read, but I do believe it's a good book!
If you know hex and octal it would be so much easier to see patterns etc and clean them up. I was an assembler programmer for IBM and I can read HEX like English so anything with bits and bytes is easy for me.

Even though retired, I still HEX edit files if I need to do a recovery of some sort or rebuild stuff. Why can't we do the same thing with AUDIO files -- I mean at their essence they are bits and bytes no matter how they get represented visually.
Uncompressed WAV files are simple... just a sequence of samples. (If you want to make it "easy", open an 8-bit mono file.) You can find links to the WAV file spec at Wotsit.org.

Again DewDude is right... It's far too much raw data to analyze or manipulate manually (i.e. 44,100 samples per second). As you probably know, a 16-bit, 44,00khz stereo file has 176, 400 bytes for every second of sound!

Plus, a couple of minor things make "visual analysis" of the raw hex data difficult. In a stereo file, the left and right samples are alternated, and the bytes are in big-endian (reversed) order. And, 16-bit WAV files use signed integers. You could probably write a little program to display the data in a more friendly format.
And IF I ZERO out the pattern (00x) with a hex editor -- I get a line with NO NOISE. I suspect that 00x is NO SOUND at all.
Right! A sequence of zeros is absolute digital silence. A 0dB (full scale) 16-bit WAV file has a positive peak of 32,767 (decimal) and a negative peak of -32,768... I'm not going to attempt hex, because I'd probably screw-up the negative number (one's compliment?). :P

P.S.
I just remembered, 8-bit WAV files use unsigned integers, but that doesn't make it any easier.... It's not so easy to "look at" an 8-bit file either... The waveform still has positive and negative half-cycles, but the data is not stored that way... Digital silence isn't zero in an 8-bit file... It's "half way"... I don't recall exactly... probably 7F or 80 (hex).
Maybe it is me but I convert all the files to mono to process - it is faster and I find it easier than dealing with stereo. So all the files are right not in MONO format to look at. Same thing in Audition - brought them in as MONO files -

I think I see the noise but don't know what to do with it as it is wrapped into the audio. If I had a color picker I could really go to town with the file - but the tools provided are clumsy compared with Photoshop editing - junky lasso, box etc. useless in Photoshop for image editing too. Only use the color picker and sometime I interpolate the color pixels by hand to get it right. But their color spectrum needs a color picker -- I suspect not something audio people think about.

Can't believe I am doing this for 6 minutes of Audio.

DougDbug
Posts: 2062
Joined: Wed Feb 16, 2005 3:33 pm
Location: Silicon Valley

Post by DougDbug »

If I had a color picker I could really go to town with the file -
Hmmmm... I've read about a program called iZotope RX. It has something called "Spectral Repair ", which might be what you're looking for. But, the "cheap" version is $350 USD. :( There is a trial version, but you can't save your results.

DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

Izotope RX wouldn't be able to fix those files anyway.

*edit: the files in question (so everyone knows) are here:
http://jay.is-lost.org/22.mp3
http://jay.is-lost.org/23.mp3

Doug: she's got noise engrained over audio...i mean the noise is MUCH higher aplitude than the audio..and it's not really noise it's more of a static.

Ok, look, I'm hitting my breaking point. This is getting silly. I've told you OVER AND OVER you're not going to be able to compltely fix those files...and now you're talking about color picking and whatnot.

YOU CAN'T TREAT AUDIO LIKE GRAPHICS OR COMPUTER CODE!

That's all I've got to say. You can't fix the files. Period. Done. DONE. You're at a point that you're comming up with real silly ideas. You obviously don't have the head for understanding audio at all. You think it's all cut and dry like computer code..it's not..it MIGHT be digital but digital audio is probably just about the most "living" creature on the PC you'll ever encounter.

You can't believe you're doing this for 6 minutes of audio? EXACTLY! Move on to something else. Give it up. Live with the noise.

I'm finished. This thread is going no where but getting on my nerves and making me very irate. I don't have to be here helping people..I'm being nice. I've been nice in the past. I'm done with nice. Give it up. You're not going to fix them. Maybe Doug and don or Emmet or one of the other guys might continue to try and help you...but i don't see why. it's obvious you've only heard what i've been telling you and not listening to what i've been saying.[/b]

Post Reply