Where do all the bits go?

[@Bassotronics]

From Paul’s loudest remark to his softest end of video “bye” is about 32 bit float.

[@wagsman9999]

I am enjoying this channel. I was a bit of an audio nut in the 80s, but when the kids started coming the money went elsewhere. Now I would like to get back into it. This channel will be a great start. I imagine a lot has changed, especially in the digital area (although CDs were available in the 80s, of course). And it seems tubes have made a bit of a (strange) comeback. (p.s. I retained my big pile of vinyl)

[@ThinkingBetter]

Dynamic range of PCM = 6 dB per bit. With 24 bits you get 144dB theoretical max dynamic range. That’s already ridiculously high and due to analog electronic noise in the amp circuits, you won’t get value of all those bits. Even 16 bits CD quality with 96dB is already excellent. The reason we use bits of 16, 24 or 32 is because these bits are sent as bytes and one byte is 8 bits thus audio PCM samples are either 2, 3 or 4 bytes. Of course 32 bits make sense in modern chips that are natively operating with 32 bits instructions of math even it’s overkill from an audio fidelity perspective.

[@geoff37s38]

Agree. Paul constantly conflates sample rate/bit depth used in the recording process with samples/bits used in the playback system. A higher dynamic range on playback that exceeds 96dB is highly undesirable as low level sound will be inaudible under ambient room noise or loud peaks would be screamingly loud.

[@trippmoore]

If they use 4 bytes (32 bits) as the data packet size they should still state the sample rate as the more important stat. It doesn’t tell us any useful info if they say they are encoding the audio in a 32 bit stream when it was sampled at only 20 bits to begin with. I really don’t care about the 12 bits of zeros tacked on to the front of the 20 bits of my sample just to make the encoding fit perfectly in a 4 byte dword (or word if the architecture defines words as 32 bits).

[@ThinkingBetter]

@@trippmoore There are cases where you actually benefit from a 32 bit PCM architecture (but for the content itself, it’s completely overkill). For example, if you attenuate the sound down to a low level in your PC or Mac, you will shift the bits down and those lower bits will start to represent real content. You can attenuate a 32 bit PCM down 96dB (which is a bit insane) and you still have 96dB of actual resolution in the stream. But, of course a DAC would never operate with so many bits as usable bits. A 24 bit PCM stream really at best can make sense only of 20 of the bits while 4 bits usually are below thermal noise in electronics. And yes, there really isn’t any meaningful value in making music with a dynamic range of more than 96dB. The red book CD standard from 1980 is still an excellent standard for dynamic range also today.

[@donaldchamble6897]

I just want to thank you for your time and energy to share with us . You are providing us with valuable information that allow one to develop a finely focused systems that are properly setup for audio pleasure.

[@davidfromamerica1871]

The bits just floated away never to be seen again 🥺

[@ford1546]

A big mistake many people make is to look more at specifications than build quality. Most good amplifiers today have really good specification, more than good. There are other things that are more important! As build quality! It is the end result out of the speaker that is most important. the speaker is the weakest link!

[@EnpuerKT]

The Speaker is the weakest link, just like the cast piston in a S550 coyote.

[@Mark-lq3sb]

@@EnpuerKT Throw the name Shelby in there and you'll get Manley forged. Just like 20 years ago when Ford built my 03 Cobra.

[@johnnytoobad7785]

Bits go into the bit-bucket. That's where they've been going for years.🤗

[@paulenciska3136]

Great and clear explanation Paul! Thanks!

[@stimpy1226]

A great explanation Paul.

[@user-od9iz9cv1w]

Nice simple explanation of bit and sample rate.

[@InsideOfMyOwnMind]

I look at it in terms of resolution across two axis. For given frequency spectrum and a given dynamic range you can improve resolution by increasing the sampling rate or similarly by increasing the bit depth. In the digital world it also comes with bonuses like as Paul mentioned, reduction of phase anomalies. Increased bit depth can be harvested in two ways. Through dynamic range or quantization error reduction most notably in the softer passages. Think of back when digital video was in it's infancy where dark scenes were pixelated. Same thing. When they fixed it the picture didn't get any brighter.

[@davidtomsett]

The analogue music signal is sampled at 96khz per second, so in a way to describe that, there are 96,000 sliced samples per second of the music put back together and played back by your dac when it outputs the analogue signal to your amp. Behind this simplistic analogy there are a whole host of electronic events going on in your dac, such as jitter, dithering, roll offs etc at play which all affect the quality of the sound you hear.

[@ford1546]

You have music with a quality of 41.1 16bit and you put it inside something that is 3 times as big. you don't lose anything as long as you don't go below 41.1 16bit. at 41.1 A DAC. has a high pass filter that removes sound above 20k. This filter is steep and takes hard. You can get phase shift and other things. By upsampling, this will happen far beyond what humans can hear. Which means you don't hear these problems. But many DACs. upsample automatically.

[@UP_URANUS]

Be careful not to knock those speakers over Paul. 😂 They seem to move quite a bit when you touch them

[@LuxAudio389]

Excellent explanation 👨‍🏫

[@larrywe3320]

24/48 is all you need *IF* the recording is done properly

[@housepianist]

100% correct! 👍🏻

[@stimpy1226]

I’d just like a bit more for safety sake. It won’t hurt and it might not help but that may be just a hair too low for sample rate. Totally agree with bit depth.

[@endrizo]

96khz is already overkill. you wont hear anything more... dont get crazy by numbers and specs. just relax and enjoy.

[@gizmobowen]

Shout out to Omak! Home of the famous Omak Stampede!

[@imkow]

i think to human ears nothing is lost from downsampling 384khz to 96khz.. 384khz oh heaven that's overkill,too high-res...even 96khz is more than enough for a master format.. 48khz,-- the old Sony Pro Audio .... studio standard of the 90s. is for me merrily good

[@scottwolf8633]

Albums with 65 to 70 db of dynamic range? Are these values available from modern records? We were very lucky, back in the Day, to be able to purchase records with 55 db of dynamic range, from the 30 second pressings and many, pressed from recycled vinyl. Doubtful that the cutting head or electronics could cram such a vast range into the microgrooves of the lacquer 50 years ago. Please feel free to correct me as I will learn more.

[@thomaskandersen7250]

Stacking your equipment like those behind, isn't that a huge no go? Or are we getting lazy here 😜 ✌️❤️🇩🇰

[@JonAnderhub]

Hey Paul, is that a shot of a FR 30 and your PS Audio equipment in your new mixing room? How is that setup working out for you and your engineers? Just a few clarifications on your video. PCM is actually available in 64-bit floating point. Of course, the dynamic range of 32-bit floating point PCM is insane and there is absolutely no clipping (within the context of a digital file) but more importantly, though is the resolution. It is of course already known that even 16-bit 44.1 kHz PCM has 32,000 times the resolution of DSD 64 but with 32-bit floating bit depth, the resolution becomes even greater than 128 or even 256 DSD. As your engineer, Ted explained the dynamic range is incredible and the signal-to-noise ratio far exceeds DSD 64. Imagine not only being able to hear the room a sexy saxophone is played in but hear the sax player take his breaths or the swirl of the brushes on the snare and then realizing you just heard the tap of the handle on the ring of the drummer's hand. Now THAT is dynamic range! Nobody cares about single molecules hitting their eardrum? They do when those molecules say you are here with the musicians in the room. Oh, by the way, I'm sure you caught your error at 4:40 where you mistake bits for samples, and then you mistakenly say that CDs can only get in about 20 kHz of frequency response when in fact CDs have 22 kHz of frequency response before any Nyquist filtering is applied at A\D conversion which is 2 kHz beyond what's considered normal human hearing and exceeds the frequency response of SACD because of the need to apply filtering at playback on DSD 64 to filter out quantization noise that reaches down into the audible range, which as you point out, causes phase shift in the audible range. You of course made an error also when you stated that PCs and Mac aren't good sample rate converters. The process of converting samples is mathematical and the processors in computers are built for math. Converting samples in digital audio files is far simpler and takes much less processing power than say changing resolution in video yet the average computer is quite capable of doing both. Anyway Paul, you never gave us a look at that mixing room when you did your studio tour videos. Sure would like to have a peek at it!

[@mauriciojv7901]

Don´t think the dude understood any of that.

[@imkow]

Mr.Paul. a question may I. What sample rate the audio industry uses when mastering a redbook CD in the later 80s? is it 88.2khz,or 48khz? or is it just the same 44.1khz First pack of CDs came in 1983.. you've got to use some higher sample rate(higher than 44.1khz) to make those CDs...the rate can't be 2.8mhz of the DSDs .. my bet is on 48khz...

[@imkow]

I kinda figure out an answer myself.. quotes from wiki: "In 1989, Sonic Solutions released the first professional (48 kHz at 24 bit) disk-based non-linear audio editing system" so allow me assume that from 1983 to 1989, the audio industry use the old means of making analog audio e.g. multi-track tape editing.. in production of CDs and after mixing composition a ADC process turns out the final 44.1khz 16bit audio data from some analog source.. then after 1989 the digital audio workstations- DAWs came to rule since..and started with 48khz/24bit as an industry standard for audio editing in the 90s..

[@geddylee501]

That hans beekhuysen bloke didn't make much sense of it either, wow complicated and I'm still none the wiser, I'll stick to 44.1 and vinyl until this DSD malarkey catches on, thx paul

[@LuxAudio389]

All my bits go to my audio dealer. 🧐

[@richardsmith2721]

Could he be confusing bits with khz?

[@NYSkye]

Things are getting a "bit" to technical lately.

[@D1N02]

The question really is what happens to the samples when a high resolution 384KHz audio file is played at 96Khz because that is all the DAC will support. What is lost? 288Khz :p

[@sudd3660]

there is bitrate and bit depth, bits everywhere.

[@johnnytoobad7785]

He's confusing bits with the classic 60's song "96 Tears"...which sounds just fine with 16 bits.

[@robertspringer9477]

96 tears minus 16 bits = nothing left to hear, just ask ? and the Mysterions.

[@pedrocols]

The more I stare at those speakers the uglier they get.

[@edwinmarcano6741]

Too confusing.

[@greenbow7888]

Is it possible your question has mistaken 384KHz and 96KHz for bits?

[@freekwo7772]

Maybe 44,1×8=352,8 kHz as sample rate against 96 kHz that's even more confusing

[@franciscorompana2985]

Everything is fine but maybe you can make your videos louder. What about some compression? Or closer mic to the source/mouth.

[@geoff37s38]

Paul is as confused as the questioner.

[@YouSoundButtHurt]

Paul's a salesman. He doesn't really need to know what he's talking about.

[@JonAnderhub]

That's because Paul doesn't know what he is talking about.

[@racesla]

Yeah, whatever you say.