What sample rate and bit rate these days?

[nothingness]

Every music streaming service requires different sample rate/bit rate.
Thus, I'm wondering which one is the best for it.
44.1 khz/16 bit or 48 khz/ 24 bit?

 

[Slouching Raymond]

I'd go with 48/24.
The streamer can re-sample it downwards if they wish.
Nobody can re-sample it upwards.

 

[ecc83]

I know my knowledge of matters digital leaves much to be desired but. If the end result is 16 bits at 44.1kHz (or 48) or worse, then sending the higher word length is pointless and inefficient? Bigger files take longer to upload and since every fekkin' piece of data ever streamed now gets locked away for eternity in a data farm consuming gigawatts of energy and Olympic pools of water we should send the smallest files we can?

16bits (at 44 or 48kHz) is more than good enough for 'domestic' listening.

Dave.

 

[keith.rogers]

Every music streaming service requires different sample rate/bit rate.
Thus, I'm wondering which one is the best for it.
44.1 khz/16 bit or 48 khz/ 24 bit?

Recording, I tend to use 48kHz/24-bit, unless I'm starting with something from someone else, and then I'll not re-sample so long as it's 24-bit. Output depends on my whims, since it's easy to output anything, I suppose. I generally bounce non-lossy to 16-bit strictly for size, but if I'm going to actually distribute a lossy format, I'll convert from 24-bit, and skip the conversion to 16-bit.

But, I'm curious. Which services actually specify just a single upload rate that's incompatible with other services? I thought they'd take most anything, or at least you can find a single one that's common. Now, there may be a difference between creating tracks for a CD and streaming, often not just bitrate, but I'm not completely sure I understand the question, or that you're asking the right one, perhaps.

 

[58andCounting]

I'd go with 48/24.
The streamer can re-sample it downwards if they wish.
Nobody can re-sample it upwards.

Upsampling has been common for decades. Or are you referring to different streaming platforms?

I know my knowledge of matters digital leaves much to be desired but. If the end result is 16 bits at 44.1kHz (or 48) or worse, then sending the higher word length is pointless and inefficient? Bigger files take longer to upload and since every fekkin' piece of data ever streamed now gets locked away for eternity in a data farm consuming gigawatts of energy and Olympic pools of water we should send the smallest files we can?

16bits (at 44 or 48kHz) is more than good enough for 'domestic' listening.

Dave.

Cheers, Dave; when you say 'send,'...what does that mean? Send to streaming services, send to clients? Just curious about the context!

Every music streaming service requires different sample rate/bit rate.
Thus, I'm wondering which one is the best for it.
44.1 khz/16 bit or 48 khz/ 24 bit?

For production? For your recordings? It will come down to the quality of the rest of your equipment. - i.e., in modern converters, the chosen sampling rate is about the least important aspect of the process. The only primary argument (albeit a mostly irrelevant one for higher-end gear) is that processing at higher sample rates will reduce aliasing. Some have variable filters as well. One of my DACs does. 24-bit depth, however, IMO allows the additional headroom needed for quality gear (and analog gear) And while not right nor wrong, 24./44.1 is my preference - after seeing all captured audio above 20khz missing after mastering, my opinion (based on my gear and setup) is that 44.1 sounds best if I use it throughout the process. That does not mean it is best... but rather it sounds best for my workflow.

 

[TalismanRich]

Upsampling has been common for decades. Or are you referring to different streaming platforms?

Upsampling can be done but it can't actually "improve" on the original. Once the limitations have been imposed, they can't be removed.

I won't say that nobody can hear the difference between 44.1 and 88.2K ( I've read long debates on this without a consensus) , and for sure 24 bits is better than 16 bits, especially if properly dithered. At my stage in life, I could probably go to 32K and never know the difference. Millions of people who listen to FM broadcast never knew that the vast majority of stations brickwall at 15kHz! (anyone remember 19kHz mplx notch filters on cassette decks?)

As they say, once it's gone, you can't get it back.

For production? For your recordings? It will come down to the quality of the rest of your equipment. - i.e., in modern converters, the chosen sampling rate is about the least important aspect of the process. The only primary argument (albeit a mostly irrelevant one for higher-end gear) is that processing at higher sample rates will reduce aliasing. Some have variable filters as well. One of my DACs does. 24-bit depth, however, IMO allows the additional headroom needed for quality gear (and analog gear) And while not right nor wrong, 24./44.1 is my preference - after seeing all captured audio above 20khz missing after mastering, my opinion (based on my gear and setup) is that 44.1 sounds best if I use it throughout the process. That does not mean it is best... but rather it sounds best for my workflow.

The main reason I record 88/24 is that the latency is lower, and my system can easily handle it. As for wasting storage space, my pitiful output is nothing compared to the gazillion worthless TikTok videos and pointless memes posted daily. Talk about a waste of time and energy! It's become the days of bread and circuses to keep the masses entertained!

 

[Slouching Raymond]

Upsampling has been common for decades. Or are you referring to different streaming platforms?

TalismanRich has got it.
If you took a photo as 640*480, you dont have the information to then double that resoution afterwards.
I did have a camera which promised higher resolutions than the sensor, but it just interpolated between the data
and blew up the file size, with no real resolution improvement.
If you did increase the sampling rate, it would just increase power consumption, while not improving quality, and therefore upsetting ecc83.

 

[ecc83]

". Millions of people who listen to FM broadcast never knew that the vast majority of stations brickwall at 15kHz! (anyone remember 19kHz mplx notch filters on cassette decks?)" Moreover Rich, programmes were distributed about the country on digital links running at about 13.5 bits! Yes!
BTW, I do mean that work should START at 24 bits! Then the result dithered to 16 bits for 'transport' or storage.

Dave.

 

[jamesperrett]

For production you certainly want to use 24 bits but CD standard 44.1kHz/16bit mastered files are fine for most digital distribution. However, most services will now accept higher resolution files and some will even flag them as HD or something similar so, if the flag is important to you, it would be best to supply 96kHz/24 bit files.

 

[TalismanRich]

". Millions of people who listen to FM broadcast never knew that the vast majority of stations brickwall at 15kHz! (anyone remember 19kHz mplx notch filters on cassette decks?)" Moreover Rich, programmes were distributed about the country on digital links running at about 13.5 bits! Yes!
BTW, I do mean that work should START at 24 bits! Then the result dithered to 16 bits for 'transport' or storage.

Dave.

My first Magnavox/Philips CD player was actually only 14 bits. That was mid 80s. It still sounded better than my cassettes and LPs because it was quieter than my LPs and didn't have a single click and pop. It wasn't too long afterwards that 16bit/4x oversampling came to market. But the 4x oversampling was just to endure that bits were read correctly, not to expand the frequency response.

 

[58andCounting]

Upsampling can be done but it can't actually "improve" on the original. Once the limitations have been imposed, they can't be removed.

I won't say that nobody can hear the difference between 44.1 and 88.2K ( I've read long debates on this without a consensus) , and for sure 24 bits is better than 16 bits, especially if properly dithered. At my stage in life, I could probably go to 32K and never know the difference. Millions of people who listen to FM broadcast never knew that the vast majority of stations brickwall at 15kHz! (anyone remember 19kHz mplx notch filters on cassette decks?)

As they say, once it's gone, you can't get it back.


The main reason I record 88/24 is that the latency is lower, and my system can easily handle it. As for wasting storage space, my pitiful output is nothing compared to the gazillion worthless TikTok videos and pointless memes posted daily. Talk about a waste of time and energy! It's become the days of bread and circuses to keep the masses entertained!

TalismanRich has got it.
If you took a photo as 640*480, you dont have the information to then double that resoution afterwards.
I did have a camera which promised higher resolutions than the sensor, but it just interpolated between the data
and blew up the file size, with no real resolution improvement.
If you did increase the sampling rate, it would just increase power consumption, while not improving quality, and therefore upsetting ecc83.

Of course, but that wasn't the quote.

The quote was -

Nobody can re-sample it upwards.

I wasn't addressing sample rate resolution (information) I was addressing upsampling (adding extra 0s). As an industry, we upsample often for various reasons. It is a common practice. In the worst cases, DVD audio discs are little more than upsampled 16-bit CDs.

In real-time real-world work, 44.1 latency hasn't ever been an issue, nor had any relevance with my systems. I have been running a RME system for 19 years.

There is no trade-off for me at 88 with modern converters. In nearly all cases, oversampling has fewer artifacts, far purer harmonic content, and a vast reduction in aliasing than running at a higher sample rate alone.

 

[58andCounting]

My first Magnavox/Philips CD player was actually only 14 bits. That was mid 80s. It still sounded better than my cassettes and LPs because it was quieter than my LPs and didn't have a single click and pop. It wasn't too long afterwards that 16bit/4x oversampling came to market. But the 4x oversampling was just to endure that bits were read correctly, not to expand the frequency response.

Talisman - here is a video for you. Alot of useful information here. Not to mention FabFilter is an exceptional company.

 

[TalismanRich]

Yeah, 58, I've seen those, and all videos from Xiph.org. It's unlikely that I'll be having much of an issue with aliasing, as 90+% of my recording involves microphones, and I can tell you that most of those mics have rolled off well before I run into aliasing. I don't record synths which you can generate any signal you want. A guitar amplifier rolls off massively above about 8k. And, as a rule, I tend to roll off the higher frequencies, contrary to what things like the Focusrite Air mode does (on a Clarett, it boost +4dB at 24kHz! That's the perfect way to increase aliasing! Will someone explain why this is good???)

There are apparently quite a few issues that are alleviated by using higher bit rates within a plugin, I've never programmed one, so I can't comment on that, it's only what I've learned reading comments by some of the designers.

It's a bit like horsepower in a car, it's unlikely you'll ever use the full 600+ BHP output of a turbocharged V8 Ferrari, but it's there if you ever needed it, with the accompanying penalty of fuel usage. But putting a 1.8L 150BHP turbo in that same Ferrari means you're going to run out of steam at some point. I'll stick with my 88/24 settings and buy another SSD for $60 if I run low on space.

 

[Slouching Raymond]

I wasn't addressing sample rate resolution (information) I was addressing upsampling (adding extra 0s). As an industry, we upsample often for various reasons.

Adding 0's is not re-sampling upwards.
If you have a crappy 8-bit sampled audio signal, adding 8 0's to each sample is not going to make it CD quality.
The only defence for adding 0's would be if you were going to further process the signal, but that just helps to stop it being worse than 8-bit.
You might as well process it as 64-bit floating point, but it will never sound better than its true 8-bit quality.

 

[ecc83]

This thread seems to have morphed into a debate about the relative merits of various sampling rates and word lengths?
That was not the OP's question? I took him to mean "I have done my work at xx and
yy rate etc...NOW, in what format do I upload it?
My answer was the most efficient way commensurate with high quality and 16 bits at 44.1kHz is as good as anyone needs.

Unless told differently by the recipient.

Dave.

 

[Papanate]

I use 48k 24 bit on everything - for a little bit I used 96K 24 bit but couldn't hear a noticeable difference.

 

[PorterhouseMusic]

I use 48k 24 bit on everything -

Yep.

 

[TalismanRich]

Every music streaming service requires different sample rate/bit rate.
Thus, I'm wondering which one is the best for it.
44.1 khz/16 bit or 48 khz/ 24 bit?

This thread seems to have morphed into a debate about the relative merits of various sampling rates and word lengths?
That was not the OP's question? I took him to mean "I have done my work at xx and
yy rate etc...NOW, in what format do I upload it?
My answer was the most efficient way commensurate with high quality and 16 bits at 44.1kHz is as good as anyone needs.

Unless told differently by the recipient.

Dave.

To address the OPs question, I would think the best answer would be to record at least using the highest bit rate/depth as required by a service you'll be using, and then down sample to the sites that request lower specs.

 

[Ujn Hunter]

I'm also in the 48 khz/ 24 bit for recording purposes (unless I'm working with someone else's files or by accident) and then I'll render a 44.1/16 as the final for uploading to the streaming services.

 

[58andCounting]

Adding 0's is not re-sampling upwards.
If you have a crappy 8-bit sampled audio signal, adding 8 0's to each sample is not going to make it CD quality.
The only defence for adding 0's would be if you were going to further process the signal, but that just helps to stop it being worse than 8-bit.
You might as well process it as 64-bit floating point, but it will never sound better than its true 8-bit quality.

I suspect there may be a slight translation issue here. Yes, the process of resampling ‘upwards’ (your language) absolutely adds zeros - ‘stuffing zeros’ is a common process and term. It’s a bit of ‘tongue in cheek,’ meaning it doesn't do much more than that. But as per this thread, we aren’t ( I was not) referring to the ‘quality’ of audio; we are referring to rate and depth. The purpose of upsampling or converting the rate and depth can be multifold; CD/PCM to FLAC or DSD is very common. Or upsampling to import audio recorded at a lower rate into a project at a higher rate for further recording, processing etc. The only point is that it’s common and has been around for a long time.