24/192 Music Downloads ...and why they make no sense

Practical tips for recording, editing, and mastering.

Moderators: khz, MattKingUSA

mclstr
Established Member
Posts: 48
Joined: Sat Jul 23, 2016 4:10 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby mclstr » Sat Sep 09, 2017 1:27 pm

The biggest problem with lower bit rates and sampling frequencies usually have to with A/D or D/A conversion.

It's not so much a problem anymore, but I got into digital in the early days and noticed that the difference in the sound was more based on the conversion hardware than on the bit rate and sample frequency.

Around the time when 8 bit, 12 bit and16 bit were the norm, I noticed that some 12 bit converters were able to sound exactly like the original source. Where some of the popular and cheaper 16 bit converters butchered the sound.

The better 12bit converter running at 32k would sound better than the cheaper 16bit running at 48.

It's not so true anymore, you can get good AD/DA converters on a chip for a few bucks now.

I personally only work at 32/44 and am proud of it:-)

I am old school and so am in the habit of keeping my levels near top where high bit rates are less important. Digital processing has less effect on data in the higher range.

Yes, 44k does have some phase shift in the higher frequencies, but the new converters are much better. Besides, it is going to end up at 16/44 or MP3/OGG in most cases anyway.

My big argument with lower frequencies and and bit-rates has to with CPU utilization.

I have to work within the limits of my computer. Higher bit/freq means more limitations on the quality of music I can produce. I have to watch my CPU as it is.
Higher bit/freq means more CPUs, more electricity, heat, fan noise and a limit on the amount and quality of plugins I can use.

I consider the lower bit/freq the best compromise for me.

User avatar
GMaq
Established Member
Posts: 1399
Joined: Fri Sep 25, 2009 1:42 pm

Re: 24/192 Music Downloads ...and why they make no sense

Postby GMaq » Mon Sep 11, 2017 2:44 pm

mclstr wrote:I personally only work at 32/44 and am proud of it:-)

I am old school and so am in the habit of keeping my levels near top where high bit rates are less important. Digital processing has less effect on data in the higher range.



+1, same here

It's all getting funneled into human ears which are both limited in frequency perception and attached to a brain that is amazingly able to fill in the blanks.

I truly believe if you blind tested people with a regular 16bit 44.1k recording and then the same recording with some mild spatial effect or phase alignment (ie a BBE) and told them it was at a higher bit rate they would quite blissfully believe there was a meaningful difference. With the recent passing of Steely Dan's Walter Becker I dug into their song catalog with a vengeance, all produced on 2" tape with high-end microphones, it sounds unbelievable good from it's analog source. If I recall correctly 2" tape roughly corresponds in a digital equivalent to 9 or 10 bits (could be wrong on that). Even working at 16bit/44.1k with decent converters is vastly superior and I'm already at my age probably missing frequencies in listening to analog tape sources..

The importance of sample rates and bit depths above 16bit/44.1k is way down the priority list in comparison to paying attention to how the sounds are recorded (proper impedances, good mics, correct mics for the application, strong signal levels, balanced lines where possible etc etc.). A well executed and properly recorded tune at 16bit/44.1k will almost always sound better than a poorly recorded one at 24bit/192k.

This issue can sure waste a lot of good recording time! :lol:

User avatar
CrocoDuck
Established Member
Posts: 1057
Joined: Sat May 05, 2012 6:12 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby CrocoDuck » Mon Sep 11, 2017 8:56 pm

I haven't read the whole thread, but just few cents that I think might be related.

GMaq wrote:It's all getting funneled into human ears which are both limited in frequency perception and attached to a brain that is amazingly able to fill in the blanks.


Indeed.

ufug wrote:It's a challenge for those of us who are not engineers to not react to audio esoterica with a raised eyebrow.

This is a great example of why. People who seem like experts do not reach consensus, and yet both sides approach the arguments with authority. What are the rest of us to think if we can't actually hear the difference?


I wouldn't say there isn't consensus. I will quote from Kutruff:

Despite the fact that people are subconsciously aware of the acoustics to which they are daily subjected, there are only a few who can explain what they really mean by ‘good or poor acoustics’ and who understand factors which influence or give rise to certain acoustical properties. Even fewer people know that the acoustics of a room is governed by principles which are amenable to scientific treatment. It is frequently thought that the acoustical design of a room is a matter of chance, and that good acoustics cannot be designed into a room with the same precision as a nuclear reactor or space vehicle is designed. This idea is supported by the fact that opinions on the acoustics of a certain room or hall frequently differ as widely as the opinions on the literary qualities of a new book or on the architectural design of a new building. Furthermore, it is well known that sensational failures in this field do occur from time to time. These and similar anomalies add even more weight to the general belief that the acoustics of a room is beyond the scope of calculation or prediction, at least with any reliability, and hence the study of room acoustics is an art rather than an exact science.


This is about room acoustics, but I think it holds pretty much to everything in audio. Most of it is science, but extremely counterintuitive. For example: why loudspeakers have cones? Because from the acoustic point of view a cone is much more similar to a flat disk than a flat disk itself. And flat disks are a good shape to design and predict sound field radiation. Pretty crazy uh? This without taking into consideration psycho-acustics: that is, the relation between the physical signal at our ears and what we perceive, with the contribution from actually all the other senses. As such, it is easy to find material that seems contradictory when actually it isn't.

I recently "reviewed" a psychoacoustics study on my blog. It is about latency, but it can show how complex it is to actually understand how physical signals relate to perception, and how many things need to be kept under control.
Check my Linux audio experiments on my SoundCloud.
Browse my AUR packages.
Fancying a swim in the pond?

mclstr
Established Member
Posts: 48
Joined: Sat Jul 23, 2016 4:10 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby mclstr » Wed Sep 13, 2017 4:49 pm

CrocoDuck wrote:I haven't read the whole thread, but just few cents that I think might be related.

... These and similar anomalies add even more weight to the general belief that the acoustics of a room is beyond the scope of calculation or prediction, at least with any reliability, and hence the study of room acoustics is an art rather than an exact science.



I studied a bit of acoustics and did some studio space design back in a previous life.

Although there is a some randomness that comes into play, bad acoustics can for the most part be avoided in most spaces.

Some famous examples have been well know concert halls that had mediocre or even bad acoustics.
In most cases, budget gets in the way or what often happens is the chief architect overrides the recommendation of the acoustical engineers.

Good acoustics often clashes with the visual goal of the architect and so acoustics aren't considered important enough, even for a concert or lecture hall.

My own personal experiences usually had to do with construction costs. Non-parallel surfaces cost more to construct than rectangular spaces.
Too much parallel surface and/near identical curved surfaces can lead to bad acoustics, but anything else is out of the budget.

My bedroom music studio is one good example:-)

User avatar
CrocoDuck
Established Member
Posts: 1057
Joined: Sat May 05, 2012 6:12 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby CrocoDuck » Wed Sep 13, 2017 5:56 pm

Yep! On top of that, there also few examples in which calculation of key properties, like reverberation time, went way wrong. Usually due to not using models properly.

I would say that nowadays methods for designing room acoustics are very mature and room acoustic predictions are more and more often similar to the end result. Also because computationally intesive methods such as FEM, BEM, Ray Tracing and Image Source Method can be handled by modern computers (by the way, this is hot).

However, in many pro audio fields it is not the same. For example, I work in design and implementation of active noise cancelling for headphones. Headphones are -a nightmare- to predict. Reliable models hold up to 1 -2 kHz. After that everything is governed by distributed models that involve the wearer anatomy and physiology. I did try to use FEM and similar, and it actually works, but there are few problems, mainly that it requires to know many material parameters (which usually are not known) and it does not fit well in the short time scale of headphone projects. So bad, 'cause I -love- computational acoustics.

By the way, now that I think about headphones, there is another thing which makes for another quite counterintuitive example: what's most important in headphone acoustics between the speaker and the cushion? Actually a different cushion can easily have a larger impact in the acoustic performances of a headphone than a different driver.
Check my Linux audio experiments on my SoundCloud.
Browse my AUR packages.
Fancying a swim in the pond?

User avatar
chaocrator
Established Member
Posts: 306
Joined: Fri Jun 26, 2015 8:11 pm
Location: Kyiv, Ukraine
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby chaocrator » Mon Dec 18, 2017 4:14 pm

24/192 Music Downloads ...and why they make no sense

„640 kB ought to be enough for anybody“ :mrgreen:

User avatar
CrocoDuck
Established Member
Posts: 1057
Joined: Sat May 05, 2012 6:12 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby CrocoDuck » Tue Dec 19, 2017 2:41 pm

Hi guys, this looks interesting.

Although ultrasonic sounds are not really audible themselves, it seems that they might somewhat affect the perception of ordinary sounds, when they are presented at the same time. Kinda sort of like "ultrasonic masking" maybe.
Check my Linux audio experiments on my SoundCloud.
Browse my AUR packages.
Fancying a swim in the pond?

folderol
Established Member
Posts: 925
Joined: Mon Sep 28, 2015 8:06 pm
Location: Here, of course!
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby folderol » Tue Dec 19, 2017 3:42 pm

There is a recording floating around the net I've been trying to find again... and failing unfortunately This really clarifies the bit depth issue.

It starts with a man talking rather quietly, but turn up the volume and there is no detectable noise, apart from the acoustic noise of the environment - a lockup garage - and his footfalls.

However he advises you to turn the volume down until you can barely hear him. After a pause he then slams a garage door, and the sound is deafening! He points out that this is a standard, unprocessed 16 bit depth recording.

As a matter of interest, I dropped the file into Audacity, and stretched it out looking for any signs on limiting. There were none :shock:

Another time I picked up a pair of recordings of a piano, demonstrating the effect of adding dither to the final processing. This was an 8 bit recording. Without dither the distortion was rather unpleasant. With dither, there was a high noise level, but magically the distortion has all but disappeared! I don't profess to understand how this works, but there is absolutely no doubt it does!

folderol
Established Member
Posts: 925
Joined: Mon Sep 28, 2015 8:06 pm
Location: Here, of course!
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby folderol » Tue Dec 19, 2017 3:43 pm

Not quite sure what happened there. Service failure then two copies - like London buses :lol:

Lyberta
Established Member
Posts: 681
Joined: Sat Nov 01, 2014 8:15 pm
Location: The Internet

Re: 24/192 Music Downloads ...and why they make no sense

Postby Lyberta » Wed Dec 20, 2017 7:30 am

folderol wrote:He points out that this is a standard, unprocessed 16 bit depth recording.


Yes, 16 bit still has very large dynamic range, people just don't use it. But when you have tons of effects, some of it will eventually be lost. So for mastering purposes, 24 bit is better (because some of those bits will be lost during processing) but the final output can be 16 bit.

User avatar
protozone
Established Member
Posts: 181
Joined: Tue May 08, 2018 9:02 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby protozone » Thu Jun 28, 2018 3:34 am

This topic is a never-ending topic on sites such as https://gearsz.com
You could debate it their for your the rest of your existence, or just make music instead.
Since they constantly get new members, and the site is already pretty old and successful, the topic never dies.

Personally, I like to maximize bit resolution so that when I process my audio with DSP there is no generation loss due to repeated DSP use and/or repeated gain changes. Since I freeze everything to audio right away and do a lot of audio manipulation, I am relieved that my main DAW programs are better than in the past in this regard. I used to read a lot of technical articles and manuals about this type of thing, and I even had a few college courses on this exact topic field during the late 1990s/early 2000s. I kept studying digital audio topics before, during, and after that time.

But getting back on topic,
I really like working at 32-bit IEEE floating point WAV. It's a nice guarantee that I'm not likely to lose any data. Also, it's the default datastream for VST effects and instruments. I use Cockos REAPER alot with the engine configured for 64-bit IEEE floating point and everything else (renders, freezing, glueing, recording) to 32-bit IEEE floating point WAV. There's even the Broadcast WAV and WAV64 possibility which is nice to have even though I don't need them. I don't even bother using mono files most of the time either, unless it's a kickdrum track that I want to be able to see at a glance needs to be mono.

But after a lot of listening tests and plenty of successfully received tunes and not much RAM, I convert my entire drum sample library to 16-bit files. Percussion is already not as detectable for fidelity unless it's tonal, and due to psychoacoustic masking, usually the tails are not going to be noticed much for the types of sounds that I use. I've gotten away with a few 32 kHz sample rate percussion sounds, but it is detectable (although some like it that way, less listening fatigue from high frequencies).

So I usually use AudioMove (32-bit Windows freeware) to convert my whole drum sample library collection to 48 kHz 16-bit WAV after I've done all the sweetening and editing from within a 32-bit IEEE float (engine) editor such as OcenAudio (LINUX & Windows). A lot of people just use the 24-bit recording defaults in REAPER, but that's kind of legacy from drums and vocals recording and it makes sense because audio interfaces can't currently do any better than 24-bit because it's the current physical electronics limitation. They physics don't currently allow for better than 24-bit audio of electrical signals as we currently use them. Light waves, I dunno, radio waves, I dunno, sonar waves, I dunno. But for pro audio, 24-bit is the limit. For 2-inch tape, 20-bit is the limit.

Whether PCI or USB or Firewire or other, 24-bit is where inputs and outputs max out in terms of fidelity.

But anyways, since I downgrade my drum samples to 16-bit, alot more of them fit into RAM. In fact, my entire drum sample library, which is rather large by my own personal standards, shrinks to a 7zip file of only a few MiB's. That's glorious for me--it makes re-using that archive and storing it really easy. Others have gigabytes or terabytes of drum samples, but that's overkill to me. Perhaps they are using loops though. I'm referring to "one-shots".

In terms of sample rate, I stick with 48 kHz. For a while I liked the idea of 88.2 or 96 kHz so that I could do fancy pitch shift experiments several octaves with zero quality loss, but i've since quit doing that. Actually, a lot of the time, 8-bit 11 kHz audio sounds phenomonally interesting due to all the aliasing if it's done right, very much like decent bit-crusher effects.

I started out many years ago using 8 kHz 8-bit mono sound files on a Mac SE, so I know what they sound like (phenomenal!)

Since FLAC, WAVpack (WV), OPUS, M4A/AAC, OGG, and MP3 can all each do 48 kHz, I like that rate. Also, r8brain can work with that. And it's the DVD standard, so it's pretty popular that way. And it's compatible with Vimeo and YouTube other sites and services. And it's compatible with camcorders too. It's better than CD (RedBook) quality, yet still highly compatible and the filesizes aren't humongous. There's a slight improvement for latency at a relatively small cost in terms of CPU. It guarantees the whole humanaly-possible audio hearing range (10 Hz - 24 kHz / 20 Hz - 20 kHz) and it won't bother cats and dogs and bats or fish. It's the closest thing we've got to 64 kHz which some say would be ideal. 88.2 might be better, but most hardware that's kinda old, can't do 88.2 kHz, but can still do 48 kHz. Some old hardware actually works better at 48 than 44.1 kHz also! So I'm sold on 48 kHz. Any larger would be using up too much RAM, CPU, convertion time, and drive space of which I don't have a lot.

Last but not least, I read a nice article in some computer magazines that modern-day motherboard manufacturers now regularly have 24-bit 48/96 kHz ADC's and DAC's by default. They might not be as good as pro converters, but they are guaranteed to be better than previous generation converters in terms of noise and some capabilities. Yet again, maybe not all would work out in the real world at 96 kHz, in terms of the strain, but 48 kHz is still easily high-fidelity without bottlenecking the systems.

Windows Vista,7,8,10, and WINE can all handle 48 kHz. Macs can handle 48 kHz too, if I remember correctly. Many antique laptops and desktops can nowadays still do 48 kHz since the ones older than that are now usually in the landfills or hopefully recycled.

I like CD's a lot, just bought some recently, but I can see at the record stores that they aren't as popular as they used to be.

On the downside, many hardware lossless audio players that also play MP3's can't do 24-bit FLAC's unless they are modified by the RockBox project. But they can still do 16-bit FLAC's. Usually they can do 48 kHz.

So we are in good hands these days. 96 kHz still might have it's officianados, but i'm not one of them.
Once I owned BPM, the drum machine software, and it was very tedious because the EXTREMELY BLOATED soundsets were ALL at 96 kHz. But it was ridiculous because many of those drum samples had hiss and AC hum in them in a bad way. To make matters worse, they were in a proprietary format, and the software itself had latency and CPU problems as well as iLok fussiness.

If they had simply put the sounds into WAV or at least AIFF format, then at least I could've still been using the sounds which I liked, instead of throwing the whole thing away. It was a mess. Too bad, because the software was great in other ways, but not much greater than FL Studio's step editor, in fact. Beat Thang kinda made similar mistakes, but at least they tried to make a collectible hardware device.

User avatar
protozone
Established Member
Posts: 181
Joined: Tue May 08, 2018 9:02 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby protozone » Thu Jun 28, 2018 3:56 am

At the risk of DH-flogging, I forgot to mention...

I work at 48 kHz 64-bit/32-bit IEEE stereo floating point WAV, and then render to the same for mixdowns.
Then I use a linux-based converter utility to convert to 24-bit stereo FLAC for my personal listening files.
I also keep backup archival copies as 32-bit IEEE stereo floating point WAV. I usually don't want to risk damaging those and they can't store metadata anyhow, so I use 24-bit FLAC files as sources for converting to 320 kbps stereo 48 kHz MP3's which I upload to https://hearthis.at/protozone

When people download, they are getting the best format I can offer (320 mp3) until I purchase an account. Then, I'll go up to 48 kHz stereo FLAC like I used to have at SoundCloud. The files people can download aren't 128 k like some videos.

Also, I specifically installed Fre:AC (free audio converter), a really nice 32-bit Windows freeware program just for the ability to specify that the MP3's should NOT be intensity encoded as joint stereo. Intensity Joint Stereo is only for loudspeakers where supposedly bass panning can't be detected. This is NOT GOOD for headphones or for any music and/or listeners who can detect bass stereophony or even low-mids stereophony. Also, most lossy codecs convert to mid-side somewhere in the chain and give higher priority to the mid (mono) than the sides. I like my tunes to be WIDE sounding, so Joint Stereo is NOT what I want, even though it's the default setting in 90% of the MP3 encoders out there. It's built into the spec. I would use Dual Mono if I could. I've done that in the past. I think it's worth it. Also, I disable any filters, since by the time I mixdown, I've already done any type of DC blocking.

Last but not least, I still use the very reliable FLAC FrontEnd and the FLAC.exe to convert to and from FLAC's and WAV's. It works perfectly in WINE-staging for me on Linux and always worked for me in Windows too. It's really tiny and easy to use. I got it from http://rarewares.org I think.

If I could, I would switch from MP3's to OPUS files, but hardly anybody uses those yet even though they do show up in YouTube video downloads, which is wonderful. I don't really use MP4 technology that much. From what I've read, it's a nightmare for programmers and is caught up in proprietary licenses too. I also forgot to mention that ALAC works with 48 kHz too. But I'd rather be using FLAC and/or M4A/AAC.

Occasionally I encounter AC3 from video files too. I think those are 48 kHz also, usually.
FL Studio used to have a warning that 48 kHz mode shouldn't be used, but I don't think it matters anymore.
They are somewhat annoying, though, because the FL Studio WAV's are a proprietary form of WAV-encoded OGG Vorbis!!!! Or at least they used to be. You could see it for yourself if you had Windows and the freeware codec revealer PlayTime.exe

I would use WAVpack more, but Audacious has some problems with WAVpack. But I will probably use it anyhow. It seems to be the best format to me in terms of what I need and want most of the time.

I'm hoping that in the future I will just work at 48 kHz stereo 32-bit IEEE floating point WAV, and render to WAVpack (at same), and upload at same, and distribute at same, and everyone will be happy and even my VST instruments won't ever be clipping.

The anti-clipping feature is probably the best thing about floating point. You can normalise DOWN to 0 dB full scale. But if it's already where you want it, you can convert to 24-bit and accept the hopefully inaudible clipping, and then normalise to -0.1 dB full scale and be good. A lot of people these days in electronic music don't even worry about clipping because it's so regular and either part of the sound or just not noticed.

In terms of psychoacoustics, clipping has to occur for a certain duration of time to even be detectable by human ears at any frequency or broadband combination. For my music, it's not a problem. Lately I don't even use much compression/limiting anymore.

So yeah, I got opinions too!

Lyberta
Established Member
Posts: 681
Joined: Sat Nov 01, 2014 8:15 pm
Location: The Internet

Re: 24/192 Music Downloads ...and why they make no sense

Postby Lyberta » Thu Jun 28, 2018 3:27 pm

protozone wrote:They are somewhat annoying, though, because the FL Studio WAV's are a proprietary form of WAV-encoded OGG Vorbis!!!! Or at least they used to be. You could see it for yourself if you had Windows and the freeware codec revealer PlayTime.exe


No, that's actually an official feature.

Also, I hope you do understand that 32 bit float has 24 bit mantissa so it can encode pretty much the same amount of information as 24 bit integer.

User avatar
protozone
Established Member
Posts: 181
Joined: Tue May 08, 2018 9:02 pm
Contact:

Re: 24/192 Music Downloads ...and why they make no sense

Postby protozone » Thu Jun 28, 2018 4:31 pm

Lyberta wrote:
protozone wrote:They are somewhat annoying, though, because the FL Studio WAV's are a proprietary form of WAV-encoded OGG Vorbis!!!! Or at least they used to be. You could see it for yourself if you had Windows and the freeware codec revealer PlayTime.exe


No, that's actually an official feature.

Also, I hope you do understand that 32 bit float has 24 bit mantissa so it can encode pretty much the same amount of information as 24 bit integer.


Naw, 24-bit integer or 24-bit packed integer is still less range than 32-bit float because of the scientific notation style exponent use. That's why 24-bit can clip at full scale, and it's extremely difficult, almost impossible to make 32-bit float or 64-bit float clip at full scale (unless somethine else earlier in the chain is causing it). You can look up the numbers if you like. There's some awesome math websites; I can't remember the names offhand though. There's even some bitwise animations of exactly what happens within IEEE floats. Maybe you are thinking of 32-bit integer.

Just to be safe though, perhaps the precision is reduced while the range is expanded? Is that what you meant? But in terms of decibels, 32-bit and 64-bit float can go I think 20 dB (maybe more?) above 0 dB full scale. You can test it in Ocenaudio if you like. Or in CoolEdit, or check the VST specs from Steinberg. Check it in Wavosaur, or Audacity.

I make tracks constantly, so I'm used to if they clip a bit if I disable the limiter on the 2-mix buss and render to stereo float I can still open the file up in OcenAudio and normalise it DOWN to 0 dB fs if I feel like it. Usually I don't though, because I'm using stuff like sMexoScope to make sure that the waveform is pretty normal and not heavily clipped. I mean a dozen of clipped samples randomly interspersed in a 5 minute song are not going to be heard at 48 kHz. A single clipped sample doesnt't occur for a long enough amount of time to be heard. Even a couple of milliseconds straight of clipping won't be heard if the tune is loud and percussive enough.

The human hearing doesn't even start to really detect pitch fully until a full 30 milliseconds. Other types of auditory perception occur around the 18-24 millisecond boundary. Some phase/filter stuff can't be easily noticed below 6 milliseconds. Anyways, I'm going off topic again, sorry. I regularly use the technique of translating from 32-bit float to 24-bit integer to deliberately clip at exactly 0 dB full scale. That way, when I perform a normalisation down to -0.1 dB full scale (a rather tiny amount) the waveform doesn't jump down several decibels just because of the transient peaks.

I'm giving away a special technique, but by doing that, you can boost the 18-20 kHz a lot if you like, which will make the transients go super high for some parts, causing superquick over 0 dB full scale but otherwise making the audio sound clear. But such superquick transients aren't even audible. They might be dangerous to your ears if prolonged though. So I don't do it too extreme, and after the peaks are clipped by going to 24-bit, by lowering to -0.1 dB full scale, they won't be the 8 to 20 dB over full scale as they otherwise would be.

Of course, you're right that the DAC's are only going to be 24-bit maximum in terms of what is heard, despite even if the WAV file is 32-bit float. But that doesn't really affect the retention of the VST instrument-generated waveforms which can go directly into the float files without being limited by DAC or ADC bit resolution.

Just an aside: I used to use MultitrackStudio Pro Plus, and in the early versions freezing a VST instrument track might clip if any format other than 32-bit float was chosen. So I started reading about the stuff then, talked to the developer too. He's a pretty nice guy usually. The CoolEdit 2000 manual had extensive information about WAV formats, dither, sampling rates, aliasing, quantization loss, generation loss, FFT windows, etc. I used to read more about that from collegiate sources too. I suppose you could email Xiph.org too. They'd probably explain it really well. They know about bandlimiting and all that.

Anyways, I'm not trying to be a know-it-all. I realise I probably spent too much time on this site yesterday, but I was trying to pass some time doing something interesting while construction workers at my apartment are jackhammering the wall right next to my bedroom!

As for FL Studio's proprietary OGG Vorbis WAV files, it's only a feature if it means they can cram more sounds into the already bloated download.
Historically, those WAV's are incompatible with some other WAV editor programs. But maybe it doesn't matter much anymore unless people are recoding too many times to other lossy formats. I did some experiments with generation loss of ATRAC on minidisc back in the 1990s, and I remember what it sounds like--slightly subtle, but not good. That's why it sucks if a person loses all backup copies of their tunes and only has somebody's MP3 copies--it's just not the same, especially for cymbals sounds, bass width, and subbass impact. It's even true for M4A's re-encoded lossy again.

But I like FL Studio. I really like how without FL Studio, Image-Line's VST's tend to work in WINE. I purchased Toxic Biohazard and Harmor and they were great. I think how they manage the licenses is clever and helpful. But I'm not a fan of downloads the size of CD-ROMs or bigger. I like how Wavosaur and OcenAudio and Audacity and Reaper and LMMS and others aren't too big.

But we can respectfully disagree too. Maybe you know some stuff that I'm just not getting yet.

Lyberta
Established Member
Posts: 681
Joined: Sat Nov 01, 2014 8:15 pm
Location: The Internet

Re: 24/192 Music Downloads ...and why they make no sense

Postby Lyberta » Fri Jun 29, 2018 2:53 pm

The convention for floating point is that values are in the range [-1;1]. Everything outside is not guaranteed to work.

WAV audio header has a format field and there are dozen of formats you can put there. But most editors don't support stuff that is not PCM.


Return to “The Producer's Area”

Who is online

Users browsing this forum: No registered users and 14 guests