Fundamentals of Audio Engineering

[Vijay_kumar74]

In order to try to understand subjective perceptions to objective data, need to understand some fundamentals of audio engineering. E.g. what is mic level output, what is line level output. How does audio signal in DB get converted into analog voltage and current values? E.g. if mic is recording about 60 DB audio, and there is sudden 120 DB loud blast sound, then who limits/clips the audio, if at all, and at what value?
Effectively, between what DB values does audio get limited after mic recordings? How do these analog audio get converted into voltage and current values? How are their voltage and current values determined and get limited? What does PreAmp do to these values (I guess it acts as voltage amp only and not as current amp). Then till voltage/current values out of DAC. What does Amp do and why does current amplification introduce audio noise? What is the limiting DB/voltage/current value out of Amp? Am sure impedances will also play part here.
So .. all in all .. looking for some kind off Sound Engineering 101 course. Any good reference would also be helpful?
Hope I did not ask for too much?

 

[kongwee]

We alway talk about minus dB for recording. Your audio interface will indicates the loudness from mic pres. If only USB compliance and no driver, you have to rely on DAWs. You should not go over 0dB for recording. It is good to maintain around -20dB or lower for mic pres as you build up your tracks. You need to take note the relationship between mic pre volume and DAWs track volume. When you record, the sample recorded in mic pres. Your DAWS can manipulate the mic track with it fader and the sample is not altered. This is non destructive editing. Other way, you can edit the sample directly like lower gain and permanently edited the sample. You may not retrieve the original sample after a few steps.

Also when you record a sample over 0dB, the clips will alway be there in sample even you lower your DAWs fader.

 

[amirm]

I have a tutorial on how I use dB in measurements. It is probably not what fully want to learn but should answer some questions.:

 

[solderdude]

E.g. what is mic level output, what is line level output.

Line level output is somewhere around 1V to 2V
Mic levels are about 60dB below that so around a few mV

How does audio signal in DB get converted into analog voltage and current values?

A microphone converts this. Some may have amplifiers in them, most don't and need to be connected to an input that is suited for a microphone.
That input has an amplifier in it with variable gain settings so you can get the signal to line-level voltages.

E.g. if mic is recording about 60 DB audio, and there is sudden 120 DB loud blast sound, then who limits/clips the audio, if at all, and at what value?
Effectively, between what DB values does audio get limited after mic recordings?

You probably mean 60dB SPL. At which voltage the amplifier behind the microphone clips depends on factors like the used gain setting and electronic 'headroom' that is available.
120dB SPL is very loud and 60dB is quite soft in level. When one wants to record this 60dB range the mic pre-amp gain should be set to record the 120dB SPL peaks and then record 60dB SPL levels. There might be background hiss and the recorded level will be -60dB opposite line level. This means the 'input meter' will hardly move at all.

How do these analog audio get converted into voltage and current values?

In the microphone.

How are their voltage and current values determined and get limited?

Voltage level is determined by the the sensitivity of the microphone. It is limited by the built-in pre-amp or physical properties of the microphone.

What does PreAmp do to these values (I guess it acts as voltage amp only and not as current amp).

A pre-amp amplifies the incoming voltage to a determined output level. This depends on the gain setting.

Then till voltage/current values out of DAC.

You probably mean ADC. Most recording devices have a built-in pre-amp and gain adjust possibility with a range of 60dB or more so between mic and line level and should be adjusted so that the input never clips under the desired level conditions.

What does Amp do and why does current amplification introduce audio noise?

Amplifiers add noise. It is just physics. How much noise, what the spectrum of that noise is depends on many factors. Components, gain, input resistances, design.

What is the limiting DB/voltage/current value out of Amp?

It varies per device. This is why there are spec sheets. It is usually mentioned in there (at least it should)

Am sure impedances will also play part here.

Yes among other aspects.

So .. all in all .. looking for some kind off Sound Engineering 101 course. Any good reference would also be helpful?

Have a look here (very, very technical):

Audio calculations in English acoustics calculator convert audio formulas sound dB audio system microphone electro engineering electronics formula sound recording studio useful stuff free audio calculator recording studio acoustic audio engineering s

Calculator calculation conversion of acoustic formulas sound calculations audio system studio electro engineering electronics sound recording useful stuff microphone formula English audio formulas electronics audio engineering physics free thd site map online education sound calcs calculations...

www.sengpielaudio.com

 

[MaxwellsEq]

So .. all in all .. looking for some kind off Sound Engineering 101 course. Any good reference would also be helpful?
Hope I did not ask for too much?

This may be a good place to start: All About Circuits audio introduction

I started tinkering with electronics from about 7 years old. I never stopped and eventually did a degree before working in recording studios. @solderdude did a great job in answering your individual questions. But, if you want to actually understand this stuff (rather than just trust our answers), you have to apply yourself. Some of it may require a grasp of mathematics that is at least college level.

 

[Vijay_kumar74]

Thanks all for clarifying the doubts. All these replies have been great to add to my understanding. It all started from trying to objectively understand subjective terms like smooth bass/muddy bass/punchy bass, sound stage, depth, layering, resolution, holographic sounding and stuff. Why does songs seemingly sounds different from SA1 than Topping A90 when the objective measurements say no difference! Most importantly if cables make any impact to sound perception? I am still not convinced that they don't . To my ears they do seem to make difference.
I started looking/analyzing at the sound waveforms levels using Audacity. Of course this is still very far fetched but in the course it lead to these basic queries. Below is snapshot of different outputs I captured using different kind of DACs/Amps and trying to analyze what (if at all), kind of changes were there at the data/waveform level?

Spoiler: File Listings:

Thanks all for contributing to my journey of learning. Some very basic fundamentals have been cleared, e.g. its typed as dB not DB and that there are also different types of dBs (dB SPL). Am sure it should help other non sound engineers as well.

Will go through the references and keep posting my specific doubts.

Thanks!

 

[Vijay_kumar74]

You probably mean ADC.

First and foremost thanks a lot for your replies! But here I meant DAC only. i.e. data getting captured in digital form (e.g. in flac files by ADC) and then getting converted into analog output from DAC RCA outs. Probably my further doubts may clarify what I want to understand.

 

[solderdude]

A DAC has nothing to do with microphones and various levels.
It just reproduces the recorded file and has line-out level. Some DACs have digital volume control.
There is no clipping unless the recording is clipped or it is a really poorly designed DAC.

It all started from trying to objectively understand subjective terms like smooth bass/muddy bass/punchy bass, sound stage, depth, layering, resolution, holographic sounding and stuff. Why does songs seemingly sounds different from SA1 than Topping A90 when the objective measurements say no difference!

Those subjective things are highly recording and personal as well as the transducers (speakers + room/headphones + all that comes with it)

I can't say why SA1 sounds different to you compared to A90.
The most likely reasons could be:
A: level differences
B: you knowing what DAC is connected (bias)
C: Default filters being substantially different in combination with transducers and IM products or frequency response.

When both DACs are having similar filters and the exact same output levels and are compared blind with statistical relevant evaluations I am quite certain you won't be able to distinguish between them.

Without controls in place... sure ... you could hear any of the things you describe.

Below is snapshot of different outputs I captured using different kind of DACs/Amps and trying to analyze what (if at all), kind of changes were there at the data/waveform level?

It is possible a DAC with a filter that does not filter the images properly and is then recorded using an ADC that does not have a proper anti-alias filter may 'worsen' things that may not have been audible when listening directly to the same DAC.

You should look at the awesome comparator software (Deltawave) from @okane.

 

[MaxwellsEq]

Electrical conduction has been understood at a scientific level since the 1800s. Radio has been scientifically understood since the late 1800s. The links between electricity and magnetism have been scientifically understood since the 1860s. Quantum mechanics has been scientifically understood since the 1920s.

Baseband audio frequencies of interest are within the limits of 10Hz to 50kHz with most important stuff being 20 to 20k. At these frequencies, Ohm's law dominates. It's possible to understand a cable in terms of its resistance, inductance and capacitance. With properly designed electronics and well built cables, we can predict how a cable will perform and we can measure it between 20 and 20k. When we do that, we are unable to measure significant differences between sensibly designed and well built cables.

 

[Vijay_kumar74]

If I see the original song waveforms using Audacity, I see lot of clipping in waveform after 55 secs.

Spoiler: Original Flac File

Were these clipped at t5he time of recording itself or limitation of Audacity or just representation to show them clipped, when actually they are not? Is there a way to see full amplitude (dB) data (i.e. beyond 1dB and -1dB)? Will I be able to see it in flac file if I try to read data from the file itself?

 

[antcollinet]

Thanks all for clarifying the doubts. All these replies have been great to add to my understanding. It all started from trying to objectively understand subjective terms like smooth bass/muddy bass/punchy bass, sound stage, depth, layering, resolution, holographic sounding and stuff. Why does songs seemingly sounds different from SA1 than Topping A90 when the objective measurements say no difference! Most importantly if cables make any impact to sound perception? I am still not convinced that they don't . To my ears they do seem to make difference.
I started looking/analyzing at the sound waveforms levels using Audacity. Of course this is still very far fetched but in the course it lead to these basic queries. Below is snapshot of different outputs I captured using different kind of DACs/Amps and trying to analyze what (if at all), kind of changes were there at the data/waveform level?

Spoiler: File Listings:

View attachment 253414

Thanks all for contributing to my journey of learning. Some very basic fundamentals have been cleared, e.g. its typed as dB not DB and that there are also different types of dBs (dB SPL). Am sure it should help other non sound engineers as well.

Will go through the references and keep posting my specific doubts.

Thanks!

If you are hearing things that are not in the audio signal (differences in cables, or differences in transparent measuring electronics) - or even if you are not - you also need to understand the role that cognitive biases can play in what you (we all) hear.

That is equally important (and equally part of the science of audio) as is the electronic recording and reproduction topics you have asked about.

 

[kongwee]

If I see the original song waveforms using Audacity, I see lot of clipping in waveform after 55 secs.

Spoiler: Original Flac File

View attachment 253419

Were these clipped at t5he time of recording itself or limitation of Audacity or just representation to show them clipped, when actually they are not? Is there a way to see full amplitude (dB) data (i.e. beyond 1dB and -1dB)? Will I be able to see it in flac file if I try to read data from the file itself?

What you see in 1 and -1 is actually range. You can treat as percentage. 1 = 100% in positive cycle and -1=-100% in negative cycle. 1 or -1 you are seeing zero dB. You able to see the track dB reaching 0dB. To change to dB scale, you need to search how can to done in Audacity. 1 or -1= 0dB, 0= -∞ dB.

 

[DVDdoug]

and that there are also different types of dBs (dB SPL). Am sure it should help other non sound engineers as well.

The dB levels are directly correlated, but usually not calibrated. Let's say you have a digital level of -3dB and an acoustic SPL level of 80dB. If you reduce the digital level down by 3dB to -6dB, the SPL level will drop by 3dB to 77dB.

Movie theaters are about the only place where digital levels are calibrated to an SPL level. Or, I should say that's the only place with a standard calibration. Live shows and recording studios can be calibrated but everybody can be calibrated differently. There is also a calibration standard for home theater (THX?) but most people listen to lower levels at home and adjust the volume by-ear.

The digital reference of 0dBFS (zero decibels full scale) is essentially the "digital maximum" so digital levels are usually negative. With integer formats it's the maximum you can "count to" with a given number of bits. A 16-bit file has "bigger numbers" than an 8-bit file, but when you play the file everything is automatically scaled by the software & drivers so a 0dB file is the same volume.

The 0dB dB SPL reference is approximately the quietest sound that can be heard so SPL levels are usually positive.

Our ears are more sensitive to mid-frequencies than high & low frequencies and SPL measurements are usually A-weighted (giving more weight to the mid-frequencies) so the measurements better match perceived loudness.

With digital data we are mostly concerned with the peaks because if we go over 0dBFS we'll (usually) get clipping. ...There are formats that go over 0dB. Audacity uses floating-point internally so it can go over 0dB, but ADCs, DACs, regular (integer) WAV files, and CDs are all hard-limited to 0dB.

Peaks don't correlate well with perceived loudness. Most commercial recordings are normalized/maximized for dB peaks but some are still louder than others. The RMS (a kind of average) correlates better than peaks and there is something that correlates even better called LUFS that takes into account the frequency content and the short-term average.

Also if you are in Audacity (etc.) and you reduce the level by 3dB, the peak, RMS, and LUFF are all reduced by 3dB.

There are different formulas for calculating dB, depending on if we are measuring amplitude/voltage or power/wattage. With voltage or digital amplitude a factor of two is a 6dB change. With power (wattage) a factor of two is a 3dB change. But no matter what you measure, a 3dB change results in a 3dB change in loudness.

If I see the original song waveforms using Audacity, I see lot of clipping in waveform after 55 secs.

FYI - When Audacity is configured to "show clipping" it shows red for potential clipping. You can get false positives and false negatives. For example, if the recording isn't clipped and you amplify for boost the bass or something you can push the peaks over 0dB and it will show red. The waves aren't really clipped (yet) and if you reduce the volume before exporting you can prevent clipping. On the other hand, if the file is really clipped and it shows red, if you lower the volume the wave shape isn't fixed and it's still clipped but Audacity will no longer show red.

 

[Vijay_kumar74]

Thanks all! Different dB levels are making more sense now.

Checked on my Audacity and found the earlier -1 to 1 on Y-axis was linear scale. It had option of changing it to dB scale also. Attached below is dB scale snapshot. Now it ranges between 0dB to 0dB with -60dB being center or minimum value.

Spoiler: 0 dBFS scale

I also compressed it by 10dB so that new peak amplitude is at -10 dB.

Spoiler: -10 dB compressed

Even in compressed waveform, peak amplitudes after 55 secs are clipped. Seems peak amplitudes have been clipped in the file itself (probably at the time of recording itself or as part of production processing of the recording).

 

[caught gesture]

Whatever medium you have for playback, nothing about the process leading up to its creation is going to be a variable in judging the differences between playback equipment. If I understand it correctly, you are hearing differences and wanting to know why. Scientifically, first you have to deal with the differences in the primary playback equipment. Your brain. Scientifically, cognitive bias has to be immediately overcome in order to validate any further conclusions. We want to keep it scientific, n’est-ce pas?

 

[Vijay_kumar74]

Whatever medium you have for playback, nothing about the process leading up to its creation is going to be a variable in judging the differences between playback equipment. If I understand it correctly, you are hearing differences and wanting to know why. Scientifically, first you have to deal with the differences in the primary playback equipment. Your brain. Scientifically, cognitive bias has to be immediately overcome in order to validate any further conclusions. We want to keep it scientific, n’est-ce pas?

No.. no .. no! To understand why I hear differences will come much later. For now am trying to understand fundamentals of sound, its waveforms, its end to end flow and what each equipment in the system tries to do on it (including any changes/distortions added). In the process, understanding sound pressure outputs and how it gets limited, is added learnings (E.g. if more powerful amp has any added advantage like producing better bass or better dynamics.. if at all?).

So basically what kind of changes could be present in waveform (if at all) that results in seemingly better sound perception using different (or more powerful?) equipments.

 

[MaxwellsEq]

what kind of changes could be present in waveform (if at all) that results in seemingly better sound perception using different (or more powerful?) equipments

The only place "more powerful" has an impact in your playback system is in driving loudspeaker or headphones. Power amplifiers usually have an input stage, a gain stage and a "grunt" stage which is responsible for making the motors move in the drivers.

If you have very efficient speakers (such as horn-loaded), you don't need a lot of grunt to go loud. It's unlikely that any normal power amplifier will "change the waveform" at all. If you have inefficient speakers, you will need more grunt. You may not have enough grunt, in which case the most likely impact will be that the peaks will be clipped off waveforms.

 

[solderdude]

No.. no .. no! To understand why I hear differences will come much later. For now am trying to understand fundamentals of sound, its waveforms, its end to end flow and what each equipment in the system tries to do on it (including any changes/distortions added). In the process, understanding sound pressure outputs and how it gets limited, is added learnings

read:

Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms​

by Floyd Toole

 

[Vijay_kumar74]

The only place "more powerful" has an impact in your playback system is in driving loudspeaker or headphones. Power amplifiers usually have an input stage, a gain stage and a "grunt" stage which is responsible for making the motors move in the drivers.

If you have very efficient speakers (such as horn-loaded), you don't need a lot of grunt to go loud. It's unlikely that any normal power amplifier will "change the waveform" at all. If you have inefficient speakers, you will need more grunt. You may not have enough grunt, in which case the most likely impact will be that the peaks will be clipped off waveforms.

I was thinking of focusing on speakers later. For now I planned to focus on headphones and headphone amps only but since it has been brought so few related questions:

1. How can power requirements of power amp or headphone amp be determined for a given speaker/headphone based on its provided impedance and sensitivity? Subjectively it is said that more powerful the amp (can provide more watts into that impedance), the better it is. It helps in providing better dynamics and bass response! Objectively it may not be true but how much of power should be sufficient?
One practical problem related to it is that I have set of Revel F208 floor standing speakers. I drive it from Yamaha Aventage Rx1050 AV receiver. To me they sound good even in ATMOS mode (with 7.1 setup) but some suppliers say that I am not doing justice to my speakers. They need powerful power amp (to be able to provide about 200 watts into 8 ohms per channel) to drive them properly. Could this be True? Do they require more "grunt"? How to determine my speakers power/grunt requirements?

2. How does headphone/speaker sensitivity relate to its impedance? Are these even related or not?

3. What are pros/cons of more sensitive and/or higher impedance drivers? Are these in any way related to sound quality?

4. How to calculate output SPL dB based on input dB (0dBFS), headphone/speaker impedance and sensitivity?

 

[MaxwellsEq]

1. How can power requirements of power amp or headphone amp be determined for a given speaker/headphone based on its provided impedance and sensitivity? Subjectively it is said that more powerful the amp (can provide more watts into that impedance), the better it is. It helps in providing better dynamics and bass response! Objectively it may not be true but how much of power should be sufficient?
One practical problem related to it is that I have set of Revel F208 floor standing speakers. I drive it from Yamaha Aventage Rx1050 AV receiver. To me they sound good even in ATMOS mode (with 7.1 setup) but some suppliers say that I am not doing justice to my speakers. They need powerful power amp (to be able to provide about 200 watts into 8 ohms per channel) to drive them properly. Could this be True? Do they require more "grunt"? How to determine my speakers power/grunt requirements?

This is a good question and you've probably worked out there are no simple answers to some of your questions in your post.

When a speaker manufacturer states "Power requirements of 100 to 300W", it simply doesn't mean anything. For normal domestic speakers at sensible levels in a normal room you are probably idling at 5 Watts or so! A 5000 W amplifier won't damage your speakers if you listen at normal levels in a normal room; but you probably have sufficient power to damage them if you keep turning the volume up

"Grunt" is either lots of Volts (for high impedance drivers) or lots of Amps (for low impedance drivers). In practice, given the variation of impedance values in a speaker, you need a bit of both. The first requires that the power supply rails are high enough and the output devices can safely swing high volts. The second requires a stiff power supply and output devices that can handle high currents - and the higher temperatures created by the current flow.

2. How does headphone/speaker sensitivity relate to its impedance? Are these even related or not?

There is no mathematical relationship between speaker/headphone sensitivity and impedance. You can't look at the specs of headphone at say 70 Ohms and say "oh yes, that will have a sensitivity of x"

3. What are pros/cons of more sensitive and/or higher impedance drivers? Are these in any way related to sound quality?

High sensitivity drivers go loud with little power input. As a result, noise in earlier stages is often lower.

High impedance drivers put less current load on the amplifier

There is no mathematical relationship between impedance or sensitivity and sound quality. Compression drivers and horn loading are good for high sensitivity, but some people consider they are "coloured".

4. How to calculate output SPL dB based on input dB (0dBFS), headphone/speaker impedance and sensitivity?

You can't calculate this without knowing the gain of the amp driving the speakers/headphones. If you have a power stage with low gain but lots of grunt, speaker/headphone impedance would be irrelevant, but you would need high sensitivity speakers/headphones to listen to it.