I'm taking older analog audio from cassette or other sources and transferring it to GW so i can play with it, clean it,boost it, etc.
After much playing around, i can usually get something listenable..but nothing near what im looking for.
It turns out that the intitial recordings weren't made with great equipment. basically a mic hanging from a light in the middle of a room for recording music..
SO, im curious ..when i boost the volume, i get distortion w/o really boosting much..as if there is something else on the file other than the desired sounds..like noise or just space or something that turns out to manifest itself within the file.
Is there a way to eliminate everything that is outside the sound signatures?
Ok, im even confusing myself here..
let's try it this way..
let's say you open a new blank file. Lets say to visually represent that file, we have a rectangular cardboard box-tube..The space inside is empty, designating the empty track available space.
Now, you insert or paste or mix an audio file to the new file and you get a wave signature that shows the music as peaks and valleys, representing their position on the db scale.
If the audio does not intrinsically reach the 0db level, is there something else there or does it remain empty--silent-space?
i know in digital there is either a value or there is silence, but not in analog. Am i hamstrung from the get-go because im starting with analog or is there a way to remove everything from the top of the signature boundaries up to 0db..the 'empty space' that seems to become a hum or some other mess and prevents me from really getting at boosting the sound i want boosted...??
does anyone understand what im asking and can help?
thx
cleaning and boosting a recording
The VOLUME -> MAXIMIZE function will increase the volume the exact amount needed to set the peaks at 0dB. The program will scan the file and show the current peak before you click 'OK". If there is already a 0dB peak, it won't do anything.
In general, 0dB is the digital maximum. You can't go over... Maybe like going outside your cardboard tube. If you try to go above 0dB, you get clipping = flat-top waves = distortion. Actually, even something worse can happen with some software.... Something like what happpens when you car's odometer rolls-over.... Big numbers get turned into small numbers... severe distortion!!!
Most CDs are maximized (AKA "normalized"), and I almost always maximize my recordings as a final step after any other processing.
Some processing (such as mixing or equalization) can increase the peaks, so you should maximize last. With GoldWave, you can temporarily go over 0dB without distorting. (It uses 32 bit floating-point.) However, if you try to save a WAV file that goes over 0dB, it will be clipped (distorted). If you maximize before saving, you'll be OK.
There is nothing "filling the space" between your peak and 0dB. However, when you increase the volume, you increase the background noise. GoldWave has some noise reduction filters you can experiment with, as well as an expander function which can be used as a noise gate. A noise gate reduces the volume (or completely silences the audio) at low volumes.
The bad news is that noise reduction works best when the noise isn't very bad... it can make a good recording into an excellent recording. Sometimes it makes it worse! ...I guess this is true for all audio processing...
A computer holds binary numbers. With 16 bits, the biggest number you can hold is 1111 1111 1111 1111. If you convert that to decimal, you get 65,535. But, a 16-bit WAV file uses the leftmost bit as a sign-bit. So, a wave file actually holds values between -32,768 and 32,767. Either of those numbers would be 0dB. (The negative numbers don't represent negative dB... they represent the negative-half of the waveform.)
A CD uses a sample-rate of 44.1 kHz. There are 44,100 of these 16-bit numbers for every second of music. Each number represents the "wave height" at any point in time. When you "connect the dots" you get the wave. In a music file, very few of these samples will actually hit 0dB. ...Imagine measuring the height of ocean waves once per second.
0111 1111 1111 1111 = 32,767 decimal = 0dB = highest value possible.
0101 1010 1010 0000 = 27,200 decimal = about -3dB
0011 1111 1111 1111 = 16,383 decimal = about -6dB
0000 0000 0000 0001 = 1 decimal = about -90dB = quietest 16-bit "sound"
0000 0000 0000 0000 = 0 decimal = minus infinity dB = silence
In general, 0dB is the digital maximum. You can't go over... Maybe like going outside your cardboard tube. If you try to go above 0dB, you get clipping = flat-top waves = distortion. Actually, even something worse can happen with some software.... Something like what happpens when you car's odometer rolls-over.... Big numbers get turned into small numbers... severe distortion!!!
Most CDs are maximized (AKA "normalized"), and I almost always maximize my recordings as a final step after any other processing.
Some processing (such as mixing or equalization) can increase the peaks, so you should maximize last. With GoldWave, you can temporarily go over 0dB without distorting. (It uses 32 bit floating-point.) However, if you try to save a WAV file that goes over 0dB, it will be clipped (distorted). If you maximize before saving, you'll be OK.
There is nothing "filling the space" between your peak and 0dB. However, when you increase the volume, you increase the background noise. GoldWave has some noise reduction filters you can experiment with, as well as an expander function which can be used as a noise gate. A noise gate reduces the volume (or completely silences the audio) at low volumes.
The bad news is that noise reduction works best when the noise isn't very bad... it can make a good recording into an excellent recording. Sometimes it makes it worse! ...I guess this is true for all audio processing...
OK, Let's get technical!If the audio does not intrinsically reach the 0db level, is there something else there or does it remain empty--silent-space? i know in digital there is either a value or there is silence,
A computer holds binary numbers. With 16 bits, the biggest number you can hold is 1111 1111 1111 1111. If you convert that to decimal, you get 65,535. But, a 16-bit WAV file uses the leftmost bit as a sign-bit. So, a wave file actually holds values between -32,768 and 32,767. Either of those numbers would be 0dB. (The negative numbers don't represent negative dB... they represent the negative-half of the waveform.)
A CD uses a sample-rate of 44.1 kHz. There are 44,100 of these 16-bit numbers for every second of music. Each number represents the "wave height" at any point in time. When you "connect the dots" you get the wave. In a music file, very few of these samples will actually hit 0dB. ...Imagine measuring the height of ocean waves once per second.
0111 1111 1111 1111 = 32,767 decimal = 0dB = highest value possible.
0101 1010 1010 0000 = 27,200 decimal = about -3dB
0011 1111 1111 1111 = 16,383 decimal = about -6dB
0000 0000 0000 0001 = 1 decimal = about -90dB = quietest 16-bit "sound"
0000 0000 0000 0000 = 0 decimal = minus infinity dB = silence
-
Perticelli
- Posts: 46
- Joined: Wed Nov 09, 2005 2:41 pm
great reply..thx.
Now, let me redirect.
In digital files, i understand the whole scenario.
However, when dealing in an analog file, it is possible, even probable, that there is noise in the file.
Is there such a thing as 'silent noise'? i know it sounds silly, but it would represent some 'thing' that exists on the file, but you dont hear it. However, the fact that it has some form of 'mass', means it also behaves as such. So when increasing the levels of the soundwave, the wave of the sound you want, this other 'silent noise mass' also gets increased in theory and more or less creates a scenario where the sound spectrum is filled to 0db without the actual wave reaching 0 db..?
To try and illustrate this mess of a question, lets go back to the wave signATure. it looks like a long mountain range...from left to right..with peaks,valleys and levels..if you drew and imaginary line tracing across the top of the signature...now if yopu take that line and continue across the file until you reach the end, you will run into the 'wall' of the GW file..then you trace the line up the wall and then across the celing, whivch represent 0db.If you tehn connect the line back to its origin spot, you have a rectangle that has a straight top (0db point across the file) and a jagged bottom, from the outline of the wave sgnature.
If you lift this shape off the file, you should have the shape and it should constitute an empty box..that is, there is no signal within the box.
This is where im asking if there is unseen noise on an analog file...it seems there is because when i try and boost the wave sig..it reaches 0db quickly..but when i play it back, the actual sound i want boosted isnt much louder..so it must be something else on top of it that is reaching the 0db level..
so how can i eliminate that "box"?
Am i insane ?
Now, let me redirect.
In digital files, i understand the whole scenario.
However, when dealing in an analog file, it is possible, even probable, that there is noise in the file.
Is there such a thing as 'silent noise'? i know it sounds silly, but it would represent some 'thing' that exists on the file, but you dont hear it. However, the fact that it has some form of 'mass', means it also behaves as such. So when increasing the levels of the soundwave, the wave of the sound you want, this other 'silent noise mass' also gets increased in theory and more or less creates a scenario where the sound spectrum is filled to 0db without the actual wave reaching 0 db..?
To try and illustrate this mess of a question, lets go back to the wave signATure. it looks like a long mountain range...from left to right..with peaks,valleys and levels..if you drew and imaginary line tracing across the top of the signature...now if yopu take that line and continue across the file until you reach the end, you will run into the 'wall' of the GW file..then you trace the line up the wall and then across the celing, whivch represent 0db.If you tehn connect the line back to its origin spot, you have a rectangle that has a straight top (0db point across the file) and a jagged bottom, from the outline of the wave sgnature.
If you lift this shape off the file, you should have the shape and it should constitute an empty box..that is, there is no signal within the box.
This is where im asking if there is unseen noise on an analog file...it seems there is because when i try and boost the wave sig..it reaches 0db quickly..but when i play it back, the actual sound i want boosted isnt much louder..so it must be something else on top of it that is reaching the 0db level..
so how can i eliminate that "box"?
Am i insane ?
-
piano nick
- Posts: 423
- Joined: Wed Jun 16, 2004 8:33 pm
Perticelli:
I don't think there is a "box" as you describe it, and there is absolutely nothing above the tops of the waves. Compare it to the ocean - the water is in and below the waves - above that is air.
When you maximize your wave files, I would suggest that you set your vertical axis to dB if it is not already there - this is standard recording terminology.
When you do a Max to "zero dB", one or both channels will be at 0.00 dB. You cannot go any louder.
If your wave file has a large range from the quiet parts to the loud parts, the limit is governed by the maximum loudness point on the wave file, and this can't go above 0.00 dB. Even if there is one single spike that has a volume above everything else, this spike will limit the perceived loudness of the file. In the case of a file with huge level variation (classical music very often has enormous dynamic range), compression can be used to reduce the dynamic range. But caution is in order, because compression also removes dynamics. Dynamics give music life.
The amount of compression used varies with the type of music - disco music is usually played at dB levels that is harmful to the inner ear. If it has quiet spots, the dancers would not hear it. Consequently this type of music typically has the crap compressed out of it. Looking at it in Goldwave, it is all red and green - very little blue background. A classical music wave file will have spikes right to the top and also portions that you can hardly see above the "no sound" point - there is a lot of blue background.
A good learning tool is to rip a number of cd's to your HDD, and compare them visually and of course listening to them.
PN
I don't think there is a "box" as you describe it, and there is absolutely nothing above the tops of the waves. Compare it to the ocean - the water is in and below the waves - above that is air.
When you maximize your wave files, I would suggest that you set your vertical axis to dB if it is not already there - this is standard recording terminology.
When you do a Max to "zero dB", one or both channels will be at 0.00 dB. You cannot go any louder.
If your wave file has a large range from the quiet parts to the loud parts, the limit is governed by the maximum loudness point on the wave file, and this can't go above 0.00 dB. Even if there is one single spike that has a volume above everything else, this spike will limit the perceived loudness of the file. In the case of a file with huge level variation (classical music very often has enormous dynamic range), compression can be used to reduce the dynamic range. But caution is in order, because compression also removes dynamics. Dynamics give music life.
The amount of compression used varies with the type of music - disco music is usually played at dB levels that is harmful to the inner ear. If it has quiet spots, the dancers would not hear it. Consequently this type of music typically has the crap compressed out of it. Looking at it in Goldwave, it is all red and green - very little blue background. A classical music wave file will have spikes right to the top and also portions that you can hardly see above the "no sound" point - there is a lot of blue background.
A good learning tool is to rip a number of cd's to your HDD, and compare them visually and of course listening to them.
PN
-
Perticelli
- Posts: 46
- Joined: Wed Nov 09, 2005 2:41 pm
okay, you answered my exact question. Just as there is air at the top of the ocean, and that air has mass, i thought whatever "mass" was above the signal might artrificially reach the 0db mark before any sound would, thereby limiting the range.
But since there's nothing there, it must be spikes of noise from the poor recording.
I enjoy compression on the guitar sometimes..but not so much on what im doing here.
So now i need to learn how to identify and clean up the noise on the track so when i boost the volume, im boosting as much source as i can, as opposed to source and junk.
Any idea's on that, other than painstaking analysis of the signature and manually cleaning it?
In audacity there is an effect called noise filter that allows to scan a file and get a noise sample..then its supposed to remove that noise if you agree with the sample. I tried that, thinking this was exactly what i need..and it probably is..but the result was to also remove most of the source as well!
I tried audacity fopr this feature only..,i didnt like the rest of the functionality compared to GW..but anyway....
But since there's nothing there, it must be spikes of noise from the poor recording.
I enjoy compression on the guitar sometimes..but not so much on what im doing here.
So now i need to learn how to identify and clean up the noise on the track so when i boost the volume, im boosting as much source as i can, as opposed to source and junk.
Any idea's on that, other than painstaking analysis of the signature and manually cleaning it?
In audacity there is an effect called noise filter that allows to scan a file and get a noise sample..then its supposed to remove that noise if you agree with the sample. I tried that, thinking this was exactly what i need..and it probably is..but the result was to also remove most of the source as well!
I tried audacity fopr this feature only..,i didnt like the rest of the functionality compared to GW..but anyway....
-
GoldWave Inc.
- Site Admin
- Posts: 4375
- Joined: Wed Mar 10, 2004 6:43 pm
- Location: St. John's, NL
- Contact:
-
piano nick
- Posts: 423
- Joined: Wed Jun 16, 2004 8:33 pm
Perticelli:
I probably used a poor analogy, but above the audio/wave file, there is NOTHING, absolutely nothing.
Any un-wanted sound is in the wave file. There is no sound outside the wave file. Think about it for a moment - there is no sound outside the wave file - the wave file IS the sound.
Let's examine your very first statement, "I'm taking older analogue audio from cassette or other sources". The sound on cassettes was never very good to start with - not as good a a new vinyl LP. Play the cassette tapes a few dozen times, and the sound worsened.
Now you play this less than stellar sound source on a tape machine that may or may not be that good, and top of that, if you are using the onboard sound chip of your computer, you are degrading it further. What will be the result? Nothing very good.
There is an old expression - "you can't make a silk purse from a sow's ear". Don't forget that all the tape hiss, and other background noise is in your recording. Refer to my post about separating all the instruments from a recording - tape hiss has some frequencies that are the same as music - when you take these out, you also remove some of the music.
Bottom line - if you don't have a good source, you are not going to get an excellent result - maybe passable, but not the best.
PN
I probably used a poor analogy, but above the audio/wave file, there is NOTHING, absolutely nothing.
Any un-wanted sound is in the wave file. There is no sound outside the wave file. Think about it for a moment - there is no sound outside the wave file - the wave file IS the sound.
Let's examine your very first statement, "I'm taking older analogue audio from cassette or other sources". The sound on cassettes was never very good to start with - not as good a a new vinyl LP. Play the cassette tapes a few dozen times, and the sound worsened.
Now you play this less than stellar sound source on a tape machine that may or may not be that good, and top of that, if you are using the onboard sound chip of your computer, you are degrading it further. What will be the result? Nothing very good.
There is an old expression - "you can't make a silk purse from a sow's ear". Don't forget that all the tape hiss, and other background noise is in your recording. Refer to my post about separating all the instruments from a recording - tape hiss has some frequencies that are the same as music - when you take these out, you also remove some of the music.
Bottom line - if you don't have a good source, you are not going to get an excellent result - maybe passable, but not the best.
PN
I've noticed that sometimes a file will look like it has plenty of room to "grow" before it reaches the maximum and it confused me for a while but I think I've figured it out.
What GoldWave draws on the screen is an approximation of the wave at the zoom level at which you are currently viewing it. I noticed that when I went to maximize a file that looked to have headroom GoldWave actually found 0db peaks that I couldn't (or didn't) see visually until I zoomed in much further at the point specified. I believe that's an accurate explanation anyway.
One thing you could try is to proccess your file using the higpass filter with the cutoff set to a low frequency. For a fairly poor quailty recording you most likely won't notice a difference in quailty if you roll off everything below 60hz or so. Open the maximize effect window first to find out what level the peak is at. Click on cancel. Then do the highpass filter. Click on maximize again and see if it lowered the peak level. If not, then low frequency hum or noise is not the problem.
What GoldWave draws on the screen is an approximation of the wave at the zoom level at which you are currently viewing it. I noticed that when I went to maximize a file that looked to have headroom GoldWave actually found 0db peaks that I couldn't (or didn't) see visually until I zoomed in much further at the point specified. I believe that's an accurate explanation anyway.
One thing you could try is to proccess your file using the higpass filter with the cutoff set to a low frequency. For a fairly poor quailty recording you most likely won't notice a difference in quailty if you roll off everything below 60hz or so. Open the maximize effect window first to find out what level the peak is at. Click on cancel. Then do the highpass filter. Click on maximize again and see if it lowered the peak level. If not, then low frequency hum or noise is not the problem.
-
piano nick
- Posts: 423
- Joined: Wed Jun 16, 2004 8:33 pm