Since you can't hear >20k, 48k is fine for listening , and 96k for research on insects/birds.

You have stumbled on "Fourier Transform", commonly implemented as FFT. Since it is a transform, no signal is lost.

For SR = 48000hz, you can do:

• FFT size = 480, 100 FFT/sec. Low frequency resolution, high temporal resolution.
  
• FFT size = 4800, 10 FFT/sec. High frequency resolution, low temporal resolution.

No signal is lost in either case, just different properties for different cases.

The best ADCs in the world are only accurate to about 20bits, and if you take into account all the processing that's applied for music with slightly more noise added to each stage, etc, 16b with its noise floor of c. -120dB is fine for listening in most cases. Headroom is good for mixing, it's not needed for listening provided sensible levels are set at the mixing stages.

Conclusion: 44/16 is fine for listening in most cases.

 

when you think back at the times of 8 bit resolution, we were able to produce high quality voltage cosines with analog computers, which would suffer mostly from the bitrate, samplerate, and clock jitter when converting to digital audio, and not so much from not beeing perfect from the beginning.

your "microphone" example is still valid argument though. i guess that is the good old rule "your chain is only as good as the weakest link in it". no ideal recording situation = dont worry too much about your D/A converter.

yeah, it is absolutely suprising sometimes how much can be faked and still works for the ears, while in other situation you wonder why you can still hear a difference after looking at the code or specs.

yes, and for the topic "A/D conversion" itself, it is even worse: there has yet to be built a clock which is good enough to make a recording at 768 kHz sound as good as one with 48.

(many music interfaces with 8 IOs are using clocks with 384 Hz and distribute those across the channels.)

the rate and the depth are just fine today. the main issue what remains is the jitter.

well, 48 bit would represent the errors caused by jitter with a much higher precision, haha.

now this is a bit too theoretical maybe. rate and depth should probably always looked at apart from each other.^^

but when we ignore the jitter issue with the current "affordable" 192 and 384 kHz interfaces, i would like to come back with my "pitch" example - or in other words, a DDD process.

what, if you know from the beginning that you would like to play the recording 4 times slower later? you would of course try to avoid to record this in 44 and choose something higher. even if it is in a lower quality because of jitter.


(i guess what i am trying to say is that experimental artists or bat researchers will more likely require 384kHz interfaces than audio professionals who record brahms concertos for major labels. if you really only record and master, 16 bit 44 kHz AD is perfect for making a CD.)

 

Last edited: Sep 8, 2021

@BEAT16: es sind übrigens 0 und nicht 1. ansonsten klare leseemfpehlung.

 

So you're saying 0Hz is a real existing frequency???

 

Regarding the "no perfect sines" and other imprecisions of the digital world keep in mind that the Fourier, Nyquist, FFT math are for discrete signals in time, not digital.

Discrete signal in time means is not continuous (sample rate and stuff) but the samples/values have perfect precision (both in time and freq). Digital adds discrete precision to the samples value (16, 32 bit etc).
In a discrete scenario, a sine would be perfectly reconstructed no matter the sampling rate as long is below Nyquist.
I think there're some possible problems/side effects when using FFT even for discrete signals, don't remember by heart.

PS. As a side note, that scenario doesn't exist in real life

 

Last edited: Sep 8, 2021

For sure! I also heard that 0hz therapy can cure cancer in only 3 sessions, each lasting indefinitely. Um... no... certainly not.

 

Sorry, thread TL;DR.

44.1kHz used to be for the highest quality consumer product: 16 BIT on Compact Disc.

If you were to work on a project at 16 Bit, you get aliasing above 22 K; cramping at Nyquist.

So NOW we record at 96Khz and 24 Bit to avoid that.
For CD, we still downres to 16 Bit 44.1 kHz though!
(And we have to dither for this reason! And prior to MP3 creation!)

This is all you need to know in a nutshell, imo.

 

After these endless chats there's one thing that i understood, i don't care how smooth those curves are and i don't want to know. All i want is to make music and creativity is my highest priority. if i wanted to study all of those advanced sonic theories, i'm done. they just make me crazy. and i want to thank everybody who've been here for help.

So the conclusion is i will be pretty fine with 44.1 for now which i was used to, i've mixed plenty songs and was pretty happy with those results and will continue to work as i'm used to. so appreciate all of you help.

peace-out

edit: i'm only releasing music for digital streaming platforms and i think i'll be fine with 24bit.

 

Totally. But the benefits of recording at 48kHz are that when we shelve and later dither, it happens way beyond human hearing.
If you record at 24 Bit 44.1kHz, I believe the cramping at 22 K still happens and we have to deal with that if it ever was to be an mp3.
So, at least still dither, is what I'm saying I guess.
But why not switch to 48 kHz for recording?
I admit 96 is probably over-kill, but it is quite standard now, regardless of over-sampling itb with plugins.
You will still have to dither for your less-than-highest versions of your track.

 

Results after the polls closed:

1.) 44.1 kHz / 24 Bit = 10% (depends on the individual case)
2,) 48.0 kHz / 24 Bit = 90% (depends on the individual case)

The diversity of forum users:

- People who are practitioners and do not like theories
- People who always had a one in math and physics
- People who do mastering professionally
- People with imagination and humor
- People who think and act rationally
- People who say "God willing"
- People who learn something

 

Last edited: Sep 8, 2021

maybe i can record at 48khz and downsample to 44.1 with a good converter?

 

Your input question was 44.1 kHz or 48 kHz, many votes for 48.0 kHz, so where is the learning effect with you @BigM?

Best Way is:
Don't rip wav files use Flac or Apple losssless --> Ripping Program: dbPowerAmp
www.dbpoweramp.com (is on the sister site)

48 kHz is the winner...!
The sample rate should be 48 kHz, unless you are explicitly producing audio CDs.

 

Last edited: Sep 8, 2021

yep, lot of people mentioned 48 For Recording. But mostly i'm a producer i'm using samples and i don't want to convert all of them since on logic i get some pitch issues. also i've noticed some artifacts with spire when i'm using with 48. for producing they sound same.
since i only produce electronic music i'll be fine with 44.1. but as a honour to your opinions i will try 48 for sure.

 

Last edited by a moderator: Sep 8, 2021

As i saw some artist/mix engineers using 48 i also saw a lot using 44.1 without any issues. for recording higher sample rates may sound good but for producing i think 44.1 is ok.Correct me if i'm wrong.

and i still can record on higher sample rate and downsample them to my project SR to get that better quality.

 

Spire is a virtual synthesizer and if some presets are not worked properly, you can sometimes hear artifacts and / or clipping. So bad preset sound design. I know that because I made presets for the Spire myself.

 

Last edited: Sep 8, 2021

Alright, i also want to do the same. Lemme try 48 for sometime. If it's better i will use it. But i have to downsample to 44.1 for spotify.

 

Spotify lowers the quality of the music it streams by compressing the audio. It can also cause distortion to the sound by decreasing the bit-depth during the encoding process. Due to these issues, Spotify will often deliver degraded audio. Because Tidal uses a lossless format, one as high as 96kHz, 24 – bit, the issues Spotify has aren’t cause for alarm when using tidal. Although you may prefer the sound of Spotify, odds are, the best possible sound will come from Tidal. Amazon also offers tracks at 24-bit/192 kHz quality as part of its Amazon Music HD service.

Hello @BigM - see the " FLAC " Format:
www.dbpoweramp.com (is on the sister site)

Audio file formats for Spotify
Deliver your audio files to us in FLAC or WAV format (we highly recommend FLAC).

When we get your audio files, we do the following:

• Check they aren’t corrupt and the format container is known
• Transcode the file into different delivery formats for our audio quality options:
  • Ogg/Vorbis (96, 160, 320 kbps)
  • AAC (128, 256 kbps)
  • HE-AACv2 (24kbps)
• Calculate the loudness of the whole release and individual tracks using the ITU 1770 standard

We also encrypt the audio files before they’re delivered to the platform.

 

Last edited: Sep 8, 2021

Gonna try 48khz and FLAC definitely. But what are the best audio settings when exporting with video editors like premiere pro.

 

Import FLAC audio files to Adobe Premiere Pro


I bought today a music collection and the audio files are .flac format, Adobe Premiere doesn't support .ogg and .flac formats, so a plug-in is needed in order to import these formats.

AdobeOgg
It is a plug-in Adobe Premiere, is not a new project is from 2015, but it does the job pretty well.

These are the steps to install the plug-in:

1. Download the file from the github repository for your system Mac or Windows.
2. Copy the plug-in to the shared MediaCore directory:

Mac:

sudo cp -r Ogg\ Premiere.bundle /Library/Application\ Support/Adobe/Common/Plug-ins/7.0/MediaCore

Windows:

Ogg.prm -> C:\Program Files\Adobe\Common\Plug-ins\CSX\MediaCore

Replace CSX with your version. For the latest Premiere Pro CC release, it will be 7.0.

Usage
Once installed you'll be able to import .ogg and .flac audio files into Premiere.

Source: https://www.wdiaz.org/import-flac-audio-files-to-adobe-premier-cc/

 

Last edited by a moderator: Sep 8, 2021

good point.

well, at least 0,000000001 Hz is, that is sure.

but a continious stream of value 0.0 or -0.33 can als be represented by digital audio, and there is no real reason to limit it to "1Hz", no idea where has that from.

there are few other inaccuracies in the article (but it is still a good read)