Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Can you point me to some guides to intermediate-level audio concepts? (Read 2988 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Can you point me to some guides to intermediate-level audio concepts?

I hate to keep this one going but I just have a question reguarding your advice, db1989... I am looking for some really good intermediate level audio "tutorials" or whatever you'd like to call them. I understand the basics of audio; that would mean things such as what was just mentioned... an audio CD is 16-bit depth, at 44.1KHz (aka 44,100Hz/s). This is because human hearing is generally unable to hear anything much past 20,000Hz, so they double that for stereo (20,000Hz for each channel/ear) and then add in some give room, an extra 4,100Hz or 2,050Hz per channel "extra".

So I have the basics of sampling, frequency, and concepts like offsets and how they need to be corrected, which will either cause "x" samples to be truncated from the start/end depending if the offset is negative or positive and may require an overread into the lead-in/out (unless the audio samples are null there... in which case the EAC setting can just fill in the missing samples with null bytes and likely any over-reading will be null, anyway.


My problem or ignorance begins to really come about when the subjects of dB, EQ'ing, and things regarding basically what a DJ would most likely work with and have to know, such as (a simple example) sine waves vs. square wave... logarithmic this or that; what would truly be helpful to me, for one... would be a map or sort of explanation of what frequencies tie in to what sounds, usually. I have trouble (other than screwing around by ear) when I'm trying to EQ (especially one with a ton of bands, such as xnor's "foo_dsp_xgeq" plugin. Say, what range do normal (unaltered) human singing voices typically fall within? Bass (guitar, that is)... Sub-bass frequencies...

And as far as what a lot of filters do... such as high/low-pass filtering, notch filtering... and on up through concepts such as quantization, etc.


My bottom-line question is, are these topics that are covered on this board someplace, or will I need to go searching elsewhere to find this kind of information (in-depth) to learn it well. I find technologies (even "old" ones like digital audio) fascinating; I just run into the problem of not being able to locate the resources that I need. It's a curse. I had a co-worker who was "great at Google'ing" a topic. I would search using all kinds of different terminologies and phrases trying to locate something for 15 minutes and come up with nothing but crap.... and he'd do a single search and hit the damn thing on the head, right there... Before then, I didn't know it was possible to "suck" at using a search engine!? Apparently I am terrible at searching & discovering what I need to find!

Off the top of your head, do you know of any particular areas here on the board or elsewhere if need be, that you consider to be excellent resources for learning these sort of intermediate level concepts? I think part of what makes me bad at searching, is I'm not really quite sure what it is that I actually need to be looking FOR... you get me? I wish there were some sort of "audio technology curriculum" that says you should learn these concepts in group A, then once you understand those fully, move on to your 2000 level concepts in group B, and this is what those are, etc... haha. Maybe I'm asking for too much. I just find myself either reading things I've already learned, or things that go over my head! Maybe I need the pre-requisite "using search engines 101"?

Quote
Concerning Audio CD based files, this is true in  the best case scenario. If you have a poor resampler sound quality will  very likely get worse, unless your sound card also is flawed at certain  sample rates, and you need to resample in software to circumvent poor  hardware resamplers (e.g. SBLive cards).


Just to see if I follow your answer to his question... since your typical Audio CD is only at 16-bit, 44.1KHz -- "upscaling" is essentially what you're doing by changing the bit depth to 24-bit and the frequency range to 192KHz -- to an unnecessary end, almost like decoding an MP3 to WAV, then encoding to FLAC. Your source is as high of a quality as you are going to get. So effectively, transcoding from lossy to lossless obviously is going to do NOTHING to improve the sound quality of the original file -- in fact, if you transcode to another lossy format, all you're doing is simply degrading the quality even further. WASAPI as I understand it is simply an API that bypasses the Windows "mixer" (which mixes in other crap going on like system dings and chimes and this alert noise etc with your audio stream) -- and feeds audio directly to the hardware, allowing no "mixing" to take place to adulterate the sound signal.

I knew a guy years in the past who was transcoding all of his MP3 files to AAC 256k VBR (iTunes Plus, basically) because he heard that it was more efficient (smaller file size for same quality, or greater quality for equal file size) -- yet he didn't understand by transcoding his entire collection he was destroying the fidelity... sinful.... he wouldn't listen to what I had to say, though. He was making his audio "better"... OK.... I have a feeling he, or any of the others could pass a blind ABX test fairly easily after that conversion was done.

Back to topic, even CDs that are supposedly 24-bit are truly only using 20 of those, am I correct? Basically halfway between 16 and 24 bit. "HD-CD" -- I lost my memory there for a second. Or maybe it was the way that they encoded the bits, they either fit 24 bits into 20 bits, or 20 bits into 16 bits... I'll obviously have to rehash that topic, but those discs are so few and far between that I'm not sure how much worth is in it. The first example that comes to my mind of a CD labeled as HDCD is TOOL - Lateralus (2001). I recall listening to a ripped 16-bit copy, and the original CD in WMP or something that was supposedly reproducing the original intended sound quality without dithering, etc... and I could hear no discernable difference.

Bit-depth, that is... going from 16 up to 24 or 24 up to 32-bit essentially widens your dynamic range, and would prevent clipping from happening that might otherwise occur in a 16-bit audio sample? Essentially, it raises the ceiling and lowers the floor... (unless I am confusing bit depth with frequency range!) -- I'll go ahead and shut up until you guys can straighten me out by hopefully at least steering me in the right direction or area that I need to be looking in or what subjects/topics that would be the best natural segue into further study of audio principles?

Thank you so much for your time, I went into a little more detail than I originally intended to... sorry for that.

Can you point me to some guides to intermediate-level audio concepts?

Reply #1 –
Whilst I appreciate the name-drop, please start new threads for new subjects; this was not even tangential!

Your post is a bit . . . large for me to process at the moment, but I would like to address this:
Quote from: Chinch on
I understand the basics of audio; that would mean things such as what was just mentioned... an audio CD is 16-bit depth, at 44.1KHz (aka 44,100Hz/s). This is because human hearing is generally unable to hear anything much past 20,000Hz, so they double that for stereo (20,000Hz for each channel/ear) and then add in some give room, an extra 4,100Hz or 2,050Hz per channel "extra".
This is incorrect. You are right about the ~20 kHz threshold of audibility, but the sampling rate is not double this because there are two channels. It is doubled because representing a tone of n kHz requires a sampling rate of 2n kHz. Each channel receives the latter rate. Thus, CD audio has two channels, each sampled at 44100 Hz and thus able to represent frequencies up to 22050 Hz. This requirement is known as the Nyquist theorem, if you would like to search for more information about it.

Being unaware of any specific threads, books, sites, etc. about these general concepts to recommend, I can only suggest beginning with our Knowledgebase, Wikipedia, and so on. You already know that you want to learn about wave types, dynamic range, etc., so it should be fairly easy to find primers, guides, and whatnot about topics such as these. And I imagine other members will have some good resources to direct you to.

Can you point me to some guides to intermediate-level audio concepts?

Reply #2 –
Thank you for your reply and setting that information straight. Maybe I don't understand things as well as I thought I did... My apologies for the misplaced post, but I appreciate you still addressing some of it, regardless.

Can you point me to some guides to intermediate-level audio concepts?

Reply #3 –
Some resources you might find useful:
harman how to listen
frequency chart
"I hear it when I see it."

Can you point me to some guides to intermediate-level audio concepts?

Reply #4 –
Quote
I'm not really quite sure what it is that I actually need to be looking FOR... you get me?


Indeed, a great obstacle in gaining answers is not knowing what the questions are.

Stick around. Read the discussions. Read old theads. The terminology used will allow you to hit up wikipedia and Google. If someone mentions "sinc curve"; "Nyquist"; "aliasing" — waste no time and go straight to their wikipedia pages. Read up. A lot will be over your head at first, but things will drop into place after a while.

As with all forms of knowledge and skill, there is no magic bullet. There's no plug in the back of your head that will teach you kung fu. You'll be here next year and still only know a fraction of the field. But you'll be less ignorant.

The main page of HA is a summary of all subforums, so you don't have to browse the lot.

Quote
frequency chart


ooh shiny!

Can you point me to some guides to intermediate-level audio concepts?

Reply #5 –
What is it, you are trying to do? Your post mentions so many different topics which are only loosely connected by the general category "audio". Maybe you should set out to achieve a specific goal and acquire the necessary knowledge as a means to an end?

To pick out one topic:
EQing certain types of instruments/voice is what a mixing engineer would do, who has the individual tracks of a song available. He/she might increase some frequencies on the vocal track and reduce the same frequencies on the guitar track. Once the song has been mixed down to stereo you can of course no longer do that.
On the playback side of things, you might EQ to compensate for an uneven frequency response of your speakers or your room. Normally you will use a microphone and test tones. You can of course also EQ for fun (e.g. adding loads of bass) but this is entirely up to personal taste.

Can you point me to some guides to intermediate-level audio concepts?

Reply #6 –
Quote from: Chinch on
I understand the basics of audio; that would mean things such as what was just mentioned... an audio CD is 16-bit depth, at 44.1KHz (aka 44,100Hz/s). This is because human hearing is generally unable to hear anything much past 20,000Hz, so they double that for stereo (20,000Hz for each channel/ear) and then add in some give room, an extra 4,100Hz or 2,050Hz per channel "extra".


As was already pointed out, the 44.1k sampling frequency is not for 2 channels of 20k response. The reason for the higher sampling rate is because one needs to sample at double the desired frequency to be able to reproduce that frequency (the Nyquist theorem). The "extra" 4100Hz is not really anything extra; the reason that is in there is more because of physics necessity than anything.

Because it is physically impossible to design a "brick wall" filter that will let everything at 20K through to the sampler while blocking absolutely everything above 20k - there has to be some "drop off slope" to the filter - the sampler must sample high enough to cover this filter slope. If it doesn't, a type of high frequency noise called "aliasing" can be introduced into the signal. By allowing the sampler to sample higher than the Nyquist sample frequency of 40k, (in theory) such aliasing noise can be avoided. In real world practice, such aliasing can still occasionally occur due to a poorly designed converter, but the increased sample rate does take care of 99% of it.

Quote
My problem or ignorance begins to really come about when the subjects of dB, EQ'ing, and things regarding basically what a DJ would most likely work with and have to know, such as (a simple example) sine waves vs. square wave... logarithmic this or that; what would truly be helpful to me, for one... would be a map or sort of explanation of what frequencies tie in to what sounds, usually. I have trouble (other than screwing around by ear) when I'm trying to EQ (especially one with a ton of bands, such as xnor's "foo_dsp_xgeq" plugin. Say, what range do normal (unaltered) human singing voices typically fall within? Bass (guitar, that is)... Sub-bass frequencies...

And as far as what a lot of filters do... such as high/low-pass filtering, notch filtering... and on up through concepts such as quantization, etc.
A DJ needs to know next to nothing about waveforms or logarithms or any of that stuff. Whether a DJ, engineer or musical enthusiast, #1 is getting your ears trained for critical and analytical listening of music and sounds. Without the ear, you may as well take up knitting instead of audio.

I created that frequency chart (yes, that is my website) as an aid to help train the ear, not as a guide for what to actually do. your ears must be your guide. A great way to get started on training you ears in that regard is to sit down with a few of your best (not necessarily favorite) recordings of a cross-section of musical genres and a 2/3rd  octave graphic equalizer (i.e a graphic EQ of 15 bands or thereabouts). Start listening to one of the recordings with all bands set to flat (or 0db boost/cut). Then, starting at the lowest frequency band, slide that fader slowly up and down while carefully listening tp how that specific band affects both the overall sound and the individual instruments in the mix. Return that fader to 0, move to the next band and repeat. Look at the frequency labels for each band as you do.

Run this exericse for about an hour a day for just a couple of weeks. Then bring a friend/partner in to "test" you. Have them play some new songs that you haven't practiced the exercise with, and put them in control of the EQ with you facing away from the EQ so you can't see what they are doing. Then have them randomly pick one band at a time, either boosting it or cutting it, and have you try to identify with your ears only just what band they are adjusting, and in which direction. You may not get them all 100% correct the first - or even the second or third - time around, but you should rapidly find yourself getting at least close enough for horseshoes.By the end of the month, you should be starting to feel pretty good about being able to recognize the main frequency band sounds within a 10-15% margin or so, whilch will get you 90% of the way to figuring out what to do or not to do EQ-wise.

Two good general guidelines to remember when EQing (*general* guidelines, not always absolute law): 1) Use EQ boost to make things sound different, use EQ cut to make things sound better, and 2) Boost wide and shallow, cut narrow and deep.

The underlying principles behind these guidelines are A) that often the best way to make something sound better is to suggically remove the frequencies that sound bad and leave the "good" stuff to be good on its own. And B) when you do want to boost the good stuff, it often sounds better to boost a wider range of frequencies by just a couple of dB than to boost a narrow range by a large amount, creating a more gentle EQ curve just to "nudge" the sound in the right direction.

I invite you to also check out the "parametric sweep" technique for removing bad frequencies from your audio. This is one of the most useful EQ techniques in both the studio and the lab. I have a quick write-up on it at http://www.independentrecording.net/irn/co...index.php?id=69.

And for general reference material, one of the best starting points on the web is the "Reference" section at Rane Corporation. There are two main areas: Rane Notes ( http://www.rane.com/library.html ) and the "Pro audio reference" ( http://www.rane.com/digi-dic.html ). Both are chock full of useful information.

HTH,

G.
 

Can you point me to some guides to intermediate-level audio concepts?

Reply #7 –
Quote
My problem or ignorance begins to really come about when the subjects of dB
I forgot to address this one. Head on over to http://www.independentrecording.net/irn/resources/index.php and click on "Metering and Gain Structure" for an applet with a comprehensive tutorial on dB and how the various dB scales relate.

G.