Subject:Hz Hurtin'

Posted by: rusty

Date:12/3/2001 6:23:02 PM

Hello!

I’m somewhat new at this and this ‘Hz’ stuff is kicking my butt! When I want to records something it always wants to know ‘sample rate’ – and knowing zip I take the default. But my real problem comes when I record a dialog or combine a bunch of sound effects and then try to use them together in soming – like the video I’m scoring in Acid – and the sound is all wrong! Or worse – I bring in several things and they sound okay but the rest of the project goes to hell.

For video, is there some ‘Hz rate’ I should use?

Why would you pick one ‘Hz’ over another?

I want to combine or add a wav file I’ve figure that the Hz rates must all be the same or bad things will happen (yes?). So is there a way to ‘convert’ the rates so they are the same?

Finally, does anyone know a way for me to learn more about the dreaded ‘Hz’ so I know what I’m doing?

Thanks for any help you can provide!
Rusty Williamson

Subject:RE: Hz Hurtin'

Reply by: Rednroll

Date:12/3/2001 7:24:27 PM

"at this time...I'ld like to talk in a language that everyone can easily understand...". Lol, always wanted to say that. Ok, anyways.

The Hz is refering to the Sampling Frequency when converting from Analog to Digital. The Nyquist Theorm says to reproduce a certain frequency, the sampling frequency must be twice that. Ok, in simple terms. If your sampling frequency is 48Khz, then you can reproduce sounds of up to 24Khz. The range of human hearing is from 20Hz to 20Khz. So theoretically, the highest Sampling Frequency in (HZ)Hertz that you need to produce the full range of human hearing is 40Khz. The most common sampling frequencies are 44.1Khz and 48Khz. With those Hz you can reproduce up to 22Khz and 24Khz respectively. They add a little overhead to make sure you cover the entire 20Khz. The higher your sampling frequency the larger the file will be when you save to hard disk, because there's more data (ie more samples). Hz=Samples/Second....thus there's 48000 samples taken of the sound in 1 second in a 48 Khz sampling frequency.

Now that you know all that. Acid will allow you to mix, many different files of different sampling frequencies. Within the Acid project, you have to tell it the "project master sampling frequency". This is the final sampling frequency that all the files you are playing will be mixed to, when you make a final mixed wave file. Most people will record files at 44.1Khz or 48Khz. Video professionals tend to record more at 48Khz, audio pro's do more 44.1Khz. A CD has a 44.1Khz sampling rate, thus that's one good reason why audio people record at 44.1 instead of 48, because you can do digital to digital transfers to CD, without having to do a "Sample rate Conversion".

Now, that I've given you all this generalized information of how this stuff works, some A-hole will probably make a post saying how they only record at 96Khz sampling rate, because it is much better than 48Khz and 44.1Khz, because it can reproduce frequencies of up to 48Khz. They would be correct, by the way and then they will argue on how those frequencies are felt and not heard. My reply usually is..."feel this buddy" :-)

regards,
Brian Franz

Subject:RE: Hz Hurtin'

Reply by: jgalt

Date:12/3/2001 7:59:09 PM

Rednroll: Well done!

Subject:RE: Hz Hurtin'

Reply by: rusty

Date:12/3/2001 8:14:19 PM

Well done indeed!

What happens if you try to copy a 48Khz file to a a CD and try to play it? Or will you get an error trying to put it to the CD. Or will it not play or sound bad or... sould okay, just not as good????

But... hmmm... eventually this audio winds up as part of an avi or qt vidio file... I suppose these also allow Hz rates to be specified for the audio (I'll have to do some looking).

If this assumtion is true, and there's a chance I'll be putting the video file on a CD to be played from... should I worry about 48 vs 44.1?

Thanks!
Rusty

Subject:RE: Hz Hurtin'

Reply by: MiguelEttema

Date:12/3/2001 10:21:57 PM

O-kay :) Here is the short version of what you need to know.

Hz is 'Hertz', or frequency. The Hz setting determines how many samples are taken every second. The more samples taken per second, the cleaner the sound, and higher frequencies can be recorded with higher Hz settings. The other setting associated with this is the bits-per-sample: either 8 or 16. 8 bits gives you less steps between amplitude (or, crudely, loud and soft), while 16 bits gives you more, allowing for finer dynamics (louds and softs).

CDs have an (old) standard of being recorded at 44,100 Hz, 16 bits. New CDs are mastered at 96,000 Hz, 24 bits, and then are mixed down to 44,100 and 16 bits: a sort of super-sampling that uses an anti-aliasing (smoothing) function to make the recording sound even smoother.

You are correct in assuming that you need all your samples to be at the same rate. Mixing samples of different rates will result in chaos, as they will play out of tune and at different speeds. Under soundforge, there is a resampling option that takes a file and resamples it to a new frequency and bit depth. However, it is best to record any new material at 44,100 Hz and 16 bits, as this delivers the greatest range of dynamics and frequencies that is currently universally accepted in the recording industry, and by many audio and video programs.

You will find that once you have added sound to your video clip, and you go to save the project, there will be the option to compress the audio using any number of compression agents, be it mpeg-4 (using DViX, an unofficial standard), mpeg-2, or another. There will be settings that will compress the audio to certain standards, depending on how much bandwidth you want the audio to utilise. Just like mp3, which has settings from as low a 24 kbps (kilobytes per second) to as high as 312 kbps, so the audio track of your video will be compressed, and the higher the kilobytes per second, the cleaner the sound.

Hope this helps,

Miguel Ettema

Subject:RE: Hz Hurtin'

Reply by: Rednroll

Date:12/4/2001 1:29:57 PM

"CDs have an (old) standard of being recorded at 44,100 Hz, 16 bits. New CDs are mastered at 96,000 Hz, 24 bits, and then are mixed down to 44,100 and 16 bits: a sort of super-sampling that uses an anti-aliasing (smoothing) function to make the recording sound even smoother."

Just wanted to mention, this is a totally bogus statement, which comes from a misunderstanding of some technologies. Especially "Anti-aliasing", this information is totally wrong and this person that knows how to correctly spell "refridgerator", should stick to spelling words correctly and not give totally misguided bogus information. If you're gonna concentrate so hard on word spelling, next time he should make sure he understands the words he's spelling correctly.

Subject:RE: Hz Hurtin'

Reply by: vanblah

Date:12/4/2001 2:58:52 PM

I would call MiguelEttema's statement an over-simplification rather than completely bogus.

Doug "Defender of the Masses" Walker

Subject:RE: Hz Hurtin'

Reply by: Rednroll

Date:12/4/2001 4:08:36 PM

Ok, I think you may be asking 2 different things here. If you record any audio to a "CD" it will always have to be 44.1Khz. That is the specification of an audio CD. So when you record to a CD using a 48Khz wave file, some kind of sample rate conversion is needed. Most software will do this for you, or you must do a resample of that wave.

A "CDrom" will allow you to record to a CDR using any sample rate. Thus it's just the same as it appears on your hard drive. Thus if you have an .AVI file with a 48Khz sampling rate of audio, the audio does not get converted to 44.1Khz transfering to a CDrom. But this CDrom can not be played back in a "CD" player. If you want to make a CD that will playback in any standard CD player then, this audio must be converted to 44.1Khz. Also a VCD that is able to playback in a DVD player has certain characteristics that must be obeyed when making a video CD. Most programs set up templates that obey these VCD properties for you, which converts the audio to the proper sampling rate and puts the video in the correct MPeg format and also stores everything in an assigned folder, which is then recognized properly by the DVD player. In basic, the DVD is looking in a specific folder for a specific format for everything (ie sample rates, video compressions) and file locations. If those formats aren't obeyed when you make the VCD then the VCD will not play back in the DVD player, because the DVD player is not smart enough to figure it out for you.

Is this what you where asking?

Subject:RE: Hz Hurtin'

Reply by: VU-1

Date:12/4/2001 5:18:11 PM

To answer your question:

Yes, you do have to worry about the sampling rate (or 'Hz', as you call it) of the audio. As was stated earlier, the standard format for CD audio is 44.1kHz/16 bit. Therefore, your production master (CD-R) MUST follow those standards.

If you try to record an audio file that is at a sampling rate/bit depth setting other than that of the standard to a CD-R via a digital or SCSI connection to a CD burner, you will not be allowed to do it. You will get some kind of error message stating your situation. The file will have to be resampled to 44.1/16 before you can proceed. The exception to this is if you are recording the file to a CD as a data file, not as CD audio.


One more thing about sampling rates:

All sound waves are analog by nature. That means that sound travels thru the air (or other medium) as a complex combination of sine waves. As said earlier, the Hertz (Hz) is a measure of cycles per second of a sine wave. Zero amplitude to positive max to zero to negative max and back to zero is one cycle. A Period(T) is the measurement of how long it takes for a particular sine wave to complete one cycle and is calculated as the inverse of Hz (Period(T) = 1/Hz, or T = sec/cycle). If you do the math: a 60Hz tone - very low bass freq. - will complete one cycle in 1/60 cycles per sec, or 0.01667 seconds. Similarly, a 1000Hz (1kHz) tone - mid freq. tone & standard calibration freq. - will complete one cycle in 1/1000 cycles per sec, or 0.001 seconds. As you can see, the 1kHz tone cycles faster than the 60Hz tone. In one second, the 1kHz tone has completed 1000 cycles whereas the 60Hz tone has only completed 60 cycles. If you plot this on a graph of time (x-axis) vs. amplitude (y-axis), you can see that the 60Hz tone has a longer wavelength than the 1000Hz tone.

Now, transfer this knowledge to your computer screen:

The waveform screen is exactly what I just described - a graphical representation of the digital audio waveform plotted as time (x-axis - horizontal) vs. amplitude (y-axis - vertical). If you will look closely at the waveform, you will see that it consists of MANY little sine waves. Some are short (high freqs.) and some are long (low freqs.). Note that the short sine waves (high freqs.) ride along on top of the longer sine waves (low freqs.) If you will notice the divisions on the timeline at the top, it is normally set to show the time displayed in Hours:Minutes:Seconds.Frames. For standard audio, there are 30 frames per second. If you zoom in further, each frame is divided into ten subframes (in Sound Forge 4.5). Therefore, the smallest division mark of time is 1/30th of a second divided by ten - or 0.00333 seconds. Do the math and we find that our friends, the 60Hz tone, takes just about 5 subframes to complete one cycle whereas the 1kHz tone completes about 3.33 cycles in one subframe. I just looked at the waveform for a hi-hat tick and it looked like it had about 50 peaks within one subframe. That translates into a freq. around 15kHz (0.00333 sec/50 = 0.00007 sec - which is the period (T) of one cycle; since T=1/Hz, then Hz=1/T (cycles/sec=1/0.00007sec=15.015kHz).

What does this mean?

When you sample (digitize) audio at a 44.1kHz sampling rate, you essentially are slicing this compilation of sine waves into 44,100 little pieces in one second. If you could zoom in that far, the waveform would look like a bar graph since each sample (slice) has a set amplitude value (height) between -32767 and +32767. If you sample the audio at 48kHz, the slices become thinner and begin to more closely resemble the actual (analog) sine (sound) wave. When you jump the sampling rate up to 96kHz, the slices become really thin and the curve of the graph (across the peaks of the little bars - sample values) will be dramatically smoother. This is why high resolution digital audio (24/96 & higher) sounds so much better than 16/44.1. The resulting digital waveforms are much closer to matching the original analog waveforms. Keep in mind, however, that although it is very close, it is not an exact replication of the original. There is still - and will always be - the flat tops of all those little bars (samples) that make up the digital waveform since you can't make the slices infinitely thin as they would have to be to exactly match the analog sine wave(s).

Study hard, there will be a test....
Hope this helps you understand a little better.

Jeff Lowes
On-Track Recording

Subject:RE: Hz Hurtin'

Reply by: MiguelEttema

Date:12/4/2001 9:16:34 PM

As I mentioned in the topic, I was trying to keep it simple for the guy who doesn't seem familiar with any knowledge of the limitations of recording digital information from an analogue source.

Was wondering why this comment came up, so I wasn't surprised to find Rednroll once again assuming he knows everything, and others know nothing. Yes, I grossly oversimplified on the 24/96 to 16/14.1 explanation of CD mastering. Do you think the guy who was querying particularly cares? How about Nyquist values? He mentioned them, and said that a Nyquist value is half the frequency of your sampling rate. Can he tell me *why* that is? I can, right now.

If you take a 5 Hz wave, and you sample it at 5 Hz, the point samples taken of the wave will always measure the same position on the wave, therefore your digital sample will result in a flat line. Therefore, no change in the data, no effective wave as such. Double your sample rate to 10 Hz, and you now read 10 points off a sine wave with 5 peaks and 5 troughs. *If* your samples happen to be taken at the peaks and troughs, then you get a triangle-wave simplified representation of the sine wave. But probability states that you won't get your samples from the peaks and troughs of the wave, so your wave will be represented by reduced amplitude, or no amplitude at all if the points are taken at the x-intercepts of the time axis.

Nyquist values represent the highest frequency where you still have the *chance* of representing your wave data respectably, but in practice, the maximum frequency that you will be able to represent with 95% confidence of data integrity being within 90% of your original source is only 40% of the frequency of your sample rate. Which doeesn't matter to the standard 25 year old adult human, who only has a range of hearing up to 18,000 Hz by this age, as high-frequency response is lost by the cochlear system of the ear at a relatively early age due to wear and tear. Only up until the age of 5 years old do people have the full range of hearing of 20 to 20,000 Hz.

Thus, when the fellow who says 96 kHz is the only way to go to really *feel* the music, (well, low frequencies are more for feeling, though high frequencies can set up resonance in teeth and the ossicles of the middle ear), you would be hearing a more accurate representation of the frequencies near 20,000 Hz, and the sound as such will sound cleaner, crisper, and fuller.

But who gives a toss? The guy who asked the original question of this thread? I don't think so. If he wanted to know about that, he would have asked. Why doesn't he try giving people information they can use, instead of trying to foist detailed trivia on everyone. If he kept things simple people would find him a lot more helpful.

Cheers,

Miguel

Subject:RE: Hz Hurtin'

Reply by: VU-1

Date:12/4/2001 10:01:50 PM

Not trying to slam anybody, but Rusty did ask for help in understanding "the dreaded 'Hz' so I know what I'm doing".

Let's ask Rusty...

Is all of the rhetoric you just got bombarded with helpful or just a bunch of bunk? Are you smarter on the subject now, or are you sorry you asked?


JL
OTR

BTW - glad to help!

Subject:RE: Hz Hurtin'

Reply by: Rednroll

Date:12/4/2001 10:11:32 PM

Good explanation, except for the fact I was referring to the bogus inaccuracy of Anti-aliasing and you mentioned nothing about that. I believe I already explaned the nyquist theorem. I'm really waiting to hear that explanation of how things are mastered at 96Khz and then mixed down to 44.1Khz using anti-aliasing techniques to smooth it out. Does this statement also assume that anti-aliasing is not needed for 96Khz? and is only used when going from a higher sampling frequencies to a lower one? If this answer is yes to either one of these questions, then maybe I truly don't understand anti-aliasing and I hope you don't mention this information to my employer the next time I design an anti-aliasing filter of a CD or DVD mechanism. I don't pretend to know everything, but I sure can point out a bullshit explanation when I see it. So will you please explain anti-aliasing for me and everyone else in this forum....because that's what I said the first time and got a further explanation of the Nyquist theorem which I already understand.

Subject:RE: Hz Hurtin'

Reply by: MiguelEttema

Date:12/5/2001 3:53:03 AM

Man, you just don't give up, do you?

I used the Nyquist example as an example of what exactly *wasn't* needed in what I consider to be a straight-forward, easy to understand answer that someone unfamiliar with sampling methodology would need. Hell, I don't know everything either, but that's not my bone with you. Bully (ie, good) for you for what you do know, it's your attacking other users of the forum and making fun of others new to the subject that is getting up my arse. (Point example, the fellow who wanted to 'Download' his music from tape to computer... he didn't need the bollocks you produced as an answer, you didn't help, and if the fellow wasn't familiar with what you were talking about, he may have just tried your bogus suggestions.)

Let's drop it, okay? I'm not going to elucidate on anti-aliasing or anything else just to prove a point to you, and I suspect that you most likely *do* know more about it if you are indeed a software or hardware engineer dealing in the matters of anti-aliasing and music production. Just realise that there are people on this forum who can do well without your acerbic personality.

Subject:RE: Hz Hurtin'

Reply by: Rednroll

Date:12/5/2001 10:15:25 AM

Ok peace, next time don't judge my credibility for incorrectly spelling a word. Spelling is not my forte, but knowledge in audio is. 99% of the time I offer useful information to begginners....point being look at this original post, I was the first to answer the question with useful information. The other 1% of the time I see clueless posts that come from just being lazy to do a little reading. So if a person is that clueless to use the term "download" when trying to get audio into their PC, and I give them a B.S. answer....which was pretty fucking entertaining if you ask me....then that person deserves to spend a lot of time scratching their head in wonder, because if they would had taken the initiative to goto the help menu of sound forge in the first place, they wouldn't be wasting time asking questions like "how do I download a cassette into sound forge?" I spent the time reading and learning and continue to do so. Now you're gonna stick up for some lazy dumb ass who hasn't? You have some knowledge in audio, your reasoning just doesn't seem too keen to me.

By the way, when you referred to "Smoothing", you're actually talking about "Anti-Imaging", NOT "Anti-Aliasing". They're 2 similar items, but totally different applications. Anti-aliasing has nothing to do with smoothing.

Subject:RE: Hz Hurtin'

Reply by: MiguelEttema

Date:12/5/2001 9:40:55 PM

Fair enough... I will agree I'm at fault with the snap judgement of your personality. A lousy week on my part didn't help things, but that's no excuse, so I apologise.

Yeah, they guy may have been lazy and may not have read the help under soundforge, but maybe he did, and still was stumped. Not everyone is blessed with great problem-solving skills, and I know what it is like having to deal with person after person who isn't clued in to what they are doing... can be frustrating, but in my primary line of work (veterinary) I can't afford to get facetious with people. Might be different in the audio industry, I don't know.

Anyway, apologies, and here's hoping we can exchange useful information once in a while.

Subject:RE: Hz Hurtin'

Reply by: rusty

Date:12/5/2001 9:53:07 PM

Dare I reply...

All of this was very helpful (I've been a computer programmer for 22 years so complexity I can handle). And I need to understand this stuff -- eventually. Right now just enough to get (fake my way) through this project I have. And your messages have helped me pull it together and in fact the matterial even seems to have more 'flame' to it for some reason ;-) (though some weirdness still remains -- see more recent post).

I thank all of you for your help -- especially the time you took to go into detail.

Rusty

PS -- BTW, I never could spell worth a crap either.