Adding mic or upgrade recorder?

[Hxd Ped]

Small detail: if the mics can not pick up anything above 20 kHz, there will be no harmonics above 20 kHz that can be recorded. So why oversampling?

.

OK, I don't know a lot about film, but this is the example I was given in school.

Use this comparison. Take a picture with 35mm film, then take the same picture with 70mm film. The 70mm looks better because there is more detailed information on the film. The same colors are there, the film didn't add any, but the information that's there is more detailed on 70mm vs 35mm.


If there is something wrong with that comparison, please point it out. Like I said, I'm not a film expert.


The same information is there no matter what sample rate you used to record. Higher sample rates give more detailed information about what's recorded, they don't record higher sound frequencys.

 

[ullanta]

10s "Dynamic Range" vs "Resolution" in sample sized is an old, old discussion here...

see, e.g., http://www.dvxuser.com/V6/showthread.php?t=53336 and http://www.dvxuser.com/V6/showthread.php?t=117636 ; this will give you a lot of detail and analysis on my viewpoint (which seemingly agrees with yours) and Petrus's different viewpoint.

Hxd - If you ask me, your comparison to image resolution is correct. However, if you extend the metaphor, there may be some details that show up in the 70mm images that are just to fine to be seen in the 35mm images. These correspond to higher frequencies. So, you can see how your final statement is mistaken - higher sample rates indeed are able to represent higher frequencies than lower sample rates.

 

[Petrus]

Use this comparison. Take a picture with 35mm film, then take the same picture with 70mm film. The 70mm looks better because there is more detailed information on the film. The same colors are there, the film didn't add any, but the information that's there is more detailed on 70mm vs 35mm.


If there is something wrong with that comparison, please point it out. Like I said, I'm not a film expert.


The same information is there no matter what sample rate you used to record. Higher sample rates give more detailed information about what's recorded, they don't record higher sound frequencys.

Your film example is valid, but only if there actually is more detail to record. In the mic audio example there is no finer detailed information than 20 kHz signals, so recording up to 44 kHz or so is of no use, there is nothing there.

The ONLY thing higher rample rates do is record higher frequencies. Detail or resolution if you wish comes in only two flavors: amplitude and frequency. I repeat:

Sample depth ONLY determines the dynamic range (difference between maximum level and the smallest amplitude detail = smallest recordable amplitude change). This is expressed in dB and each bit correspond to about 6 dB.
Sample rate ONLY determines the highest possible recorded frequency, nothing more.

These are the two cornerstones of digital sampling and they are not affected by opinions, hearsay, misinformation, misunderstanding or gut feelings.

 

[Hxd Ped]

Sample rate ONLY determines the highest possible recorded frequency, nothing more.

I'm sorry, you are wrong.


There are many good articles here and on the internet. Please read them because I don't know any other way to explain it then the way I already have.

Don't mean to sound like a jerk, but...

 

[Petrus]

What else does the sample rate detarmine besides the highest possible recorded frequency? Are there some new scientific findings?

I have to admit I am sorry for many things, for many reasons.

 

[10s]

There are mainly three main components of audio:

Frequency: The number of times the sound wave occurs per second, we call this frequency because it's how frequent it occurs. Pitch directly correlates to frequency. An A on the piano is made up of 440 cycles (waves) of sound pressure waves persecond hitting our ears and/or microphone diaphram.

Amplitude: Another is how intense the wave is, this is called amplitude, we often call it loudness or volume. When sound is quiet and then loud we call this dynamics. The difference between zero audio signal to ear/microphone/ recorder/maximum ability to receive.... is the dynamic range.

Harmonics: The third component is harmonics. They are secondary waves that occur and give unique distinction/identity to the fundamental frequency/tone of the wave. Example, a violin and a trumpet can play the same A 440 cycle tone but sound differently because of the unique harmonics they each produce differently than the other. If harmnonics did not exist then all sounds would be similair to the sine wave tone you hear during bars & tone.

Sample Rate: The higher the sample rate, the greater the ability to capture higher frequencies. Though we may not hear above 20kHz, the tones within our hearing range are affected by these sympathetic high frequency waves that interact and color them. There is valid controversy about sampling above 44.1 kHz.

Where there isn't controversy is in higher bit depth. More depth gives greater dynamic ability. 16 bit is good, but 20 & 24 bit is even better.

 

[Gohanto]

Sampling frequency doesn't change the range of frequencies recorded. It just takes MORE samples every second.

And here's a simple diagram showing what bit-depth is...

 

[10s]

The number/frequency of samples per second does affect the range of recordable frequencies. More samples per second = ability to record higher frequency sound waves. Using the Nyquist theorem, a 192kHz sample rate is able to record 96kHz sound waves. A 96kHz sample rate is able to record 48kHz sound waves, a 48kHz sample rate (std video) is able to record 24kHz sound waves, and a 44.1kHz sample rate is able to record 22,000 sound pressure waves per second, and we can only hear up to 20,000 of thoses waves.

 

[ullanta]

Gohanto -

Your images aren't working for me! Can't see 'em. But, indeed, sampling frequency determines the range of frequencies that can be represented/recorded. Sampling frequency ALSO affects the amount of detail/accuracy of soundwaves across the spectrum. But The relation between sampling frequency and highest recordable/representable frequency is clear and non-controversial. The only controversy involves whether it's WORTHWHILE to record at high sample rates, given other limitations in the recording chain.

 

[ullanta]

"We can only hear up to 20kHz" is not necessarily a true/complete statement. Some research shows that we are able to react/process signals at higher frequencies, even though we don't consciously "perceive" them. They may indeed be used for things like localization and other factors below our level of consciousness. This leaves open the question of, given a front-end that can handle higher frequencies than 20 kHz, whether it's worthwhile to record 'em. But, given the equipment (esp. mics) currently available, we won't have an answer for a while.

That said, a mic with a frequency response of "20Hz-20kHz" does indeed respond to higher frequencies, just in general at a reduced output level. Frequency response doesn't drop off suddenly. So there is indeed SOME higher-frequency content in many mic signals. Some people claim this makes a difference, and that they can hear/analyze said difference when comparing recordings at different sample rates. I don't have any personal opinion on the matter so far...

 

[10s]

Ullanta, well said!


... PS, I like your robots.

 

[ullanta]

They like you, too!

 

[Petrus]

Sampling frequency ALSO affects the amount of detail/accuracy of soundwaves across the spectrum. But The relation between sampling frequency and highest recordable/representable frequency is clear and non-controversial. The only controversy involves whether it's WORTHWHILE to record at high sample rates, given other limitations in the recording chain.

The first sentence is false. Raising the sample rate, say, from 48 to 96 kHz has absolutelly no effect on frequences below 22 kHz or so. Study Nyquist theorem a bit more. That detail/accuracy you are talking about are the exact higher frequences higher sample rate is able to capture.

If we can capture (mics), reproduce (speakers) or hear them (us) is another matter.

 

[10s]

No, he's correct. Think of the samples as vertical lines that when put together represent the amplitude (volume) of the wave. With more samples (vertical lines) the more closely they are able to follow the dynamic amplitude contour of the wave. In effect they create a more detailed and accurate representation of the pressure wave.

Higher sample rates allow recording of higher frequencies & more detailed representation of all audio waves regardless of where they sit in the range of frequencies.

 

[ullanta]

The first sentence is false. Raising the sample rate, say, from 48 to 96 kHz has absolutelly no effect on frequences below 22 kHz or so. Study Nyquist theorem a bit more.

Petrus, I'm trying hard not to get into this kind of discussion with you again, which is why I pointed to the previous discussions. To say it in the least argumentative way, your understanding differs significantly from that of most audio engineers (and by "engineers" I mean folks with engineering degrees, not "people who record audio"). I wish you'd tone down your rhetoric and insistence on calling everyone else "wrong."

Any interpolation technique will tend to work better with more data. The Nyquist theorem specifies the range of frequencies that can be represented by a given sample rate. However, it does not say anything about the accuracy with which a given sampling frequency can approximates any particular waveform. More sampled points = more accurate representation of the waveform, regardless of the frequency.

Except in the rarest cases, audio waveforms are highly complex and irregular and accuracy of recording benefits from high sample rates. Again, however, what is "Sufficient" accuracy is an open question.

love,
barry

 

[Petrus]

Opinions and misunderstandings do not change mathematical and scientific facts.

"Staricase" illustration has caused a lot of misinterpetations.

I am not trying to convert you, Ullanta, anymore away from the dark side, but now only trying to save the younger generation to the side of science.

We are recording only sine waves, which can be perfectly descibed by very few samples (in ideal conditions only 2 samples per cycle). For that reason higher sample rates have no effect on the recording quality of low frequences.

The code words are Nyquist and Fourier. Do your own studying.

Peace to all...

 

[gco]

The first sentence is false. Raising the sample rate, say, from 48 to 96 kHz has absolutelly no effect on frequences below 22 kHz or so. Study Nyquist theorem a bit more. .....

Impossible.
The Nyquist Theorem establishes the MINIMUM rate at which to sample an analog waveform at discrete points in time in order to obtain an intelligible digital approximation. The more often you sample, the more accurate the approximation of the original. Raising the sampling rate absolutely does have an effect on the accuracy of approximating any frequency be it within hearing range or not. There is no way that it cannot. To say otherwise flys in the face of basic engineering principles. Whether or not one can hear a difference in resolution between 48 and 96 is another subject as pointed out previously.

Opinions and misunderstandings do not change mathematical and scientific facts....
I am not trying to convert you, Ullanta, anymore away from the dark side, but now only trying to save the younger generation to the side of science.
We are recording only sine waves, which can be perfectly descibed by very few samples (in ideal conditions only 2 samples per cycle). For that reason higher sample rates have no effect on the recording quality of low frequences.
The code words are Nyquist and Fourier. Do your own studying.
Peace to all...

That is totally illogical.
Sure, digitizing a very very low frequency with a very low sampling rate is consistent with the Nyquist Theorem, i.e. MINIMUM sampling rate = 2x highest frequency you are sampling. A 1Hz sine wave could be sampled acceptably by a 2Hz sampling frequency. But what is your point? Nobody is recording sine waves here. We are talking about audio which involves highly complex waveforms rich in harmonic content that is dynamic. The more often you sample, the greater accuracy with which you approximate the original analog waveforms. The effect may be more noticeable in the higher frequencies, i.e. transients but it is just as true for low frequencies as it is for high frequencies. Please don't mislead anyone here.

 

[Petrus]

Nobody is recording sine waves here. We are talking about audio which involves highly complex waveforms rich in harmonic content that is dynamic. The more often you sample, the greater accuracy with which you approximate the original analog waveforms. The effect may be more noticeable in the higher frequencies, i.e. transients but it is just as true for low frequencies as it is for high frequencies. Please don't mislead anyone here.

People here either forget or do not know that even the most complex audio waveform consist only of pure sine waves of different amplitude, frequency and phase (Fourier series). If we say 48 kHz system can record up to 24 kHz perfectly, if can describe the 24 kHz frequency, which is a sinusoidal wave, as the Nyquist theorem says. If there are some "datail" in the audio that 48 kHz sampling can not record, it MUST be some even higher frequency.

So, in this sense we are recording only sine waves, a complex mixture of them, that is for sure.

 

[ullanta]

"A complex mixture of them" is perhaps the key for you, Petrus! To determine what that complex mixture is, without foreknowledge, is a tricky approximation process in which higher sample rate always does a better job.

 

[Michael T]

Well I hate to upset all of you guys but that being said there are people like me that can hear extremely higher than most of you on this list. People use to think I was wacked when I'd ask them if they could hear that high pitch or sound. They would just look at me funny and I'd keep on listening. Well come to find out I can hear things that most people can't hear. I was tested by the government and I hear much higher than the so called normal person. The guy doing the test came out and looked at me a little funny so I asked him what, why you looking at me that way. He just wanted to give me the test again so I said ok. He gave me the test three times and then a different test and finally gave up. He printed out some paper work for me and sent off his finding to whom ever. Finally he explaind that he had never met anyone that could hear this high before and thought that I was cheating somehow. Well I wasn't and he finally realized this is the way it is. I don't hear dog whistles but do hear harmonics and high frequencys 30 to 40% higher than most people according to him.

The problem is two fold because I don't hear base as well as other people it has always been a problem for me being distorted where others say its just fine. They get mad when I turn down the bass so I can hear the music better. I fell in love with my first stereo system that I just had to buy because of the bass response. It was a Macintosh stereo and if any of you have heard one the bass is super clean and I loved it. Great speakers that reproduce dynamic sounds very well. Modern equipment is much better now at reproducing sounds including bass so its much better than the old equipment was. My best friend Ed doesn't hear above 14k and can't hear a floresent light ballast or any high sound at all. In a TV signal that has snow mixed in with the sound and sounds terrible he can't hear the snow and can pick out the people talking like it isn't there. He is a engineer and has been tested lots of times because of his job clearence. Its just how he is. He can't hear any harmonics at all but can hear orders that others can't make out over a bad signal in headphones so it has helped him many times.

Now for the bad, I can and hear lots of noise that you can't hear so I hate dirty sound. I hear what sounds like high hiss and noises when others just think I'm nuts. I have gotten use to it but am very aware that I tune it out a lot. I love creating sound using good headphones from sounds produced in very quite sound rooms. I love music and today's is so much better. Harmonics make a much sweeter and richer sound thats fuller to my ears and on good recordings I can hear them. Hell nobody can really tell what you hear but I'm interested in clean clean clean sound. Higher bit rates capture more samples and do reproduce harmonics better but also produce sounds you can't hear. In a clean studio that use a lot of filters to make clean electricity and quieter lights its better, out side it can be more noisy in the upper register. I say be thankful you can't hear this high. I went to music school (college) with a kid that had perfect perfect pitch which is much better than perfect pitch. Willy Maden our instructor who played on the road for 30 years with the best big bands had never met a person with this talent. In the begining of each class he woud have 3 students come up and all four would hit 10 notes and Jim could start at the bottom note and tell every note that was being played. Now thats impossible I know but I saw it done for years while we were in school together. It never ceased to amaze Willy, he couldn't believe someone could do this. But Jim who played guitar couldn't play with others for two years until he trained himself to play with people that were always out of tune. Jim tuned his guitar for two years in every class and that got old. We thought he couldn't play, but finally he stopped tuning and played great. He just could not listen to the symphony because it gave him an instant migraine. He would have to leave rehersal sometimes because to many people were out of tune. So be careful what you wish for.

I don't usually share this because if people can't hear it, it just doesn't exist.