MP3 vs FLAC system payload?

[Rednroll]

Originally Posted by fakemaxwell

Right, I said I had golden ears and that this is all about hearing the differences between FLAC and mp3. Oh....wait, nope, once again you're conflating the two different use cases that are being discussed, namely editing and final delivery.

Now you're trying to start making bets about listening tests??? Whether or not anybody can hear the difference between the two formats in a listening test is completely irrelevant to what filetypes one should use while editing and mixing.

If your claim that mp3s have more benefits was true, everybody would record straight to mp3. Please point me to another source that recommends doing so.

Sorry, I must not have received that golden rules handbook of editing and final delivery. Could you post me a link and the particular section you're referring to?

Originally Posted by fakemaxwell

In no world should you be using mp3s in your final uncompressed exports, especially for sound effects, so wondering whether or not mp3 is better for system load is a moot point.

Must be difficult keeping a story straight when you're blowing smoke out your behind.

[Bri1]

golden ears listeners who will split hairs over the minor losses of MP3@320kbps but then have an 180 degree viewpoint on system overhead.

^aye-that's an intersting point-but consider-kbps it' the stream rate yeah? per sec?
+then 1 need consider reaper playback resampling project settings..this can be very low rate..or very high rates..
+i also believe spectro+sonograms to be a much 'better' judge of qualities over 'ears' any day--the numbers simply do not lie--unless they are..faulty,or in error of calculations..
had you ever considered trying .wav adpcm before? the 2 or 4 bit compressions ? <<<what does you ears and disk managements say to that?

[Rednroll]

Originally Posted by Bri1

had you ever considered trying .wav adpcm before? the 2 or 4 bit compressions ? <<<what does you ears and disk managements say to that?

Never tried it. The only other .wav format I'm familiar with besides the standard PCM is broadcast .wav. What would I use to convert close to 100K files to this .wav adpcm you speak of and which programs support its playback?

[fakemaxwell]

Originally Posted by Rednroll

Sorry, I must not have received that golden rules handbook of editing and final delivery. Could you post me a link and the particular section you're referring to?


Must be difficult keeping a story straight when you're blowing smoke out your behind.

lol okay buddy, use whatever you want. Clearly you've convinced yourself that you're correct. If you want to lose 60% of the information of your entire library of sound effects to save a few gigabytes, go right ahead. I just hope anybody else who stumbles onto this nonsense has the good sense to use uncompressed, lossless files when possible until final delivery.

[Bri1]

Never tried it.

heh-well reaper is only 1 i know that does this type of file-it's not supported very much otherwise--and looking again,maybe the ratios v quality are not as good as best rates of mp3-- not a lot of media players even support flac playbacks...so in practice i mainly stick to wavs,wavpacking+ogg's all day long..24 or 32bit wavs/wv for the better keepers,or hq conversions..
a lot of older sample libraries just used 44.1/16 for most practical space saving and system economies..even the small leap upto 48/24 has a quite radical effect on system resources --once your hitting 96k--well most systems will struggle with only a few samples per sec..it's very taxing on playbacks so librarians kinda not adopt it--
even hq video,games and dvd/bluray is squashed a bit to maximum space saving levels i guess for extra content to be available.
trade offs,are trade offs,there's no other way of seeing it really..when it comes to any type of compressions...the payback is in the ratios eh.

*shrugz*

+ there's 2 really easy ways to check playbacks and sources-
1. set up 5 recording input tracks and set each to a different recording format by going to> 'track recording setting/quantize etc' set pre rec--study the differences after recording on playback+ phase invert to also check them.
2. make a wav file in reaper,select it + batch convert it >to multi formats...then (turn down volume 1st!) and duplicate your 1st wav track 100x ...goto 'projectbay' /source media and switch out all instances to your other formats while playback is happening-- view your scopes+ cpu graphs..per sec--ok.

[Rednroll]

Originally Posted by Bri1

heh-well reaper is only 1 i know that does this type of file-it's not supported very much otherwise--and looking again,maybe the ratios v quality are not as good as best rates of mp3-- not a lot of media players even support flac playbacks...so in practice i mainly stick to wavs,wavpacking+ogg's all day long..24 or 32bit wavs/wv for the better keepers,or hq conversions..
a lot of older sample libraries just used 44.1/16 for most practical space saving and system economies..even the small leap upto 48/24 has a quite radical effect on system resources --once your hitting 96k--well most systems will struggle with only a few samples per sec..it's very taxing on playbacks so librarians kinda not adopt it--
even hq video,games and dvd/bluray is squashed a bit to maximum space saving levels i guess for extra content to be available.
trade offs,are trade offs,there's no other way of seeing it really..when it comes to any type of compressions...the payback is in the ratios eh.

*shrugz*

+ there's 2 really easy ways to check playbacks and sources-
1. set up 5 recording input tracks and set each to a different recording format by going to> 'track recording setting/quantize etc' set pre rec--study the differences after recording on playback+ phase invert to also check them.
2. make a wav file in reaper,select it + batch convert it >to multi formats...then (turn down volume 1st!) and duplicate your 1st wav track 100x ...goto 'projectbay' /source media and switch out all instances to your other formats while playback is happening-- view your scopes+ cpu graphs..per sec--ok.

Actually, I just did something similar and opened a FLAC and MP3 version of the same sound FX. Just a random FX with a lot going on in the mix. I put each on different tracks, made sure they were time aligned as much as I could using the max zoom level and then flipped the phase on one of the channels and rendered them. All audio of the Sound FX was completely gone, except for what seemed to be left was tape floor hiss noise where the sound FX was likely originally recorded on analog tape. I'm unsure which version had more tape floor noise than the other, but maybe MP3 removed some of the unwanted noise floor from the original recording which actually may be a benefit in some instances.

[Rednroll]

Originally Posted by fakemaxwell

lol okay buddy, use whatever you want. Clearly you've convinced yourself that you're correct. If you want to lose 60% of the information of your entire library of sound effects to save a few gigabytes, go right ahead. I just hope anybody else who stumbles onto this nonsense has the good sense to use uncompressed, lossless files when possible until final delivery.

So no link to this handbook? How will I ever become a professional like yourself that knows everything?

What's hilarious to me is that I haven't even decided if I'm going with the MP3 or the FLAC versions of the Sound FX files, yet here you are criticizing me and getting all bent out of shape for even considering using the MP3 files while basing your opinion on nothing more than speculations and some unpublished golden rule. I bet you must be a real joy to have a conversation with and get along with in real life. The phrase "closed minded" has really comes to light.

[Bri1]

All audio of the Sound FX was completely gone, except for what seemed to be left was tape floor noise where the sound FX was likely originally recorded on analog tape. I'm unsure which version had more tape floor noise than the other, but maybe MP3 removed some of the unwanted noise floor from the original recording which actually may be a benefit in some instances.

heh-aye---what your noticing is the sometimes very very 'subtle differences--they are the differences the 'golden ears and eyes' notice i guess
try rendering at 8bit-you may be surprised by the results even.. people used to make music with 8bit..allll daaay loong..lolz.and even sell it. =)
beauty,in all it's forms,is in the ears and eyes of each beholder-<this = true imo.

[Rednroll]

Originally Posted by Bri1

heh-aye---what your noticing is the sometimes very very 'subtle differences--they are the differences the 'golden ears and eyes' notice i guess
try rendering at 8bit-you may be surprised by the results even.. people used to make music with 8bit..allll daaay loong..lolz.and even sell it. =)
beauty,in all it's forms,is in the ears and eyes of each beholder-<this = true imo.

LOL! Yeah, you don't have to tell me. When I 1st became an audio engineer like 20 years ago, I was kind of closed minded like some of the individuals in this conversation. I was working with some Hip Hop clients who were bringing in their SP1200 drum machines which sounded nasty to me while we had a much cleaner sounding MPC-3000 drum machine in the studio. I naively asked, why don't you use this MPC it sounds so much cleaner? They looked at me with fright in their eyes. A couple months later an engineer who did more hip-hop tracks than me, explained how a lot of the hip-hop artists loved the sound of those 12bit/26Khz grainy/gritty codecs in the SP1200 and didn't want that more polished clean sound. I'm now very familiar with those techniques and sounds you described with that 8-bit rendering. I find it hilarious how someone is criticizing me for even considering not following some golden ear workflow. That just shows me they're likely very new to all this and still unable to think outside of the box which are just introductory to studio engineering 101 basics......don't draw outside the lines! LOL!

[Bri1]

heh-oh fo'sure-- i am hiphop--higher infinite power healing other people!!
for me,it's the most abstract collaging music form in modern history--much history was made by 'hiphop' producers+turntable artists-because they pioneered the mechanical studio techniques and most were on serious low level budgets---heh,invention out of necessity,so to speak.
imo-if you don't have to cut corners,why cut'em

[Breeder]

Originally Posted by Bri1

imo-if you don't have to cut corners,why cut'em

All tools are to be used, no need to search for limits, they do come on their own, haha

Just use lossless format when recording. No need to worry about generational wear (render to render to render) nor audio processing (because mp3 changes the sound, your inserts might not do what they can because they will lack information that was cut out when doing lossy compression.). Remember, lossy compression uses psycho-acoustics to do all sorts of things to cut out size...it changes the audio and it's no longer an original.

[Kyd Kasters]

FLAC.

Unless you don't give a shit about your audio.

If that is indeed the case then just compress the fuck out of it with whatever you can find. Shove it in a zip file then convert it to mp3, have a fat bitch (brainwreck might be available) to sit on it and cram it in a carry-on bag. XITE will then transport it to his coven of "attractive" whores where you can subsequently parcel it out, package it in condoms, and shove it up their collective asses. Then, for a fee, they'll transport it across the border, pour taco sauce on it, wherein it will commence with the "expanding" process. After that, he'll summarily chop it up with a rusty razor blade and snort it all up inside his prodigious proboscis.

Then he'll shit it out onto a Beiber cover track.

Trust me, just go with the FLAC.

[Kyd Kasters]

You're welcome.

[drumphil]

I must be a glutton for punishment, but here we go.

I recorded 100 tracks in flac format, 25% CPU usage on playback.

I recorded 100 tracks in mp3 format (target quality best 256k), 25% to 27% cpu usage on playback.

At 128kbps the CPU usage is about 1% lower than FLAC, and that's with 100 tracks.

At 320kbps the cpu was overloaded recording so I can't test playback.

WAV playback was between 19% and 24% cpu usage.

Unless you don't have space for the FLAC or WAV files there is no reason to use MP3 with an SSD. 100 tracks of uncompressed 44.1K 24 bit audio requires about 13 megabytes a second of storage bandwidth, so there is no bandwidth bottleneck there with a SSD.

Case closed.

It took me 5 minutes to do that test. You could have done it. And it confirms what I already told you. It's almost like I've done this before.

[Bri1]

lolz@ Kyd Kasters = heh,that's some human feelings about it
1 may never know how the o.poster decided-until they share their findings,or not..

[Bri1]

it confirms what I already told you. It's almost like I've done this before.

lolz-yeah m8- see i reckon the 'problem' is some people simply do not like being 'told'..?
and yes-you may have done this before,in a dream as we do...a do do do shoo whup da wooo..!

[Bribedant]

Interesting thread. About compressed audio and cpu payload in Reaper, isn't an imported MP3 file converted to a "virtual copy" with the audio spec of the Reaper project? i.e no decoding going on at playback and no extra CPU load.

Another example. If I import an 88.2 khz audio file into a 44.1khz projekt the "info" on the file item says "RESAMPLED from 88.2 khz". This tells me this representation of the audio file in Reaper is already resampled to 44.1khz and no conversion is going on in real time playback.
Have I got this wrong?

[xpander]

Sample rate conversions are done live unless you render or glue the items into the project sample rate. Playback resample mode in Project settings affects how heavy this live conversion is. Render resample mode affects the quality of the render/glue.

[drumphil]

Originally Posted by Bribedant

Have I got this wrong?

Yes, you've got that wrong. The resampling is done in real time.

To work the way you stated, reaper would have to render new resampled versions of all the audio in a project that didn't match the new sample rate you changed to every time you changed sample rate.

[Reason]

My unscientific two cents:
Go with FLAC. You can always add more storage space or replace the drive with a bigger one, but it'd be a huge hassle to recreate your projects with lossless items instead of lossy. If it's worth doing, it's worth doing right.

I am not convinced I'd personally notice a difference, BUT, I'd WORRY that I would, and moreover, I'd worry that I wouldn't notice until far down the line. Illogical, sure, but it's my reality, so I take steps to avoid that worry.

Therefore, if I was in your shoes, I'd go with FLAC.

[Bribedant]

Originally Posted by drumphil

Yes, you've got that wrong. The resampling is done in real time.

To work the way you stated, reaper would have to render new resampled versions of all the audio in a project that didn't match the new sample rate you changed to every time you changed sample rate.

As a principle rendering a new instance of a imported "wrong format" audio file to current project settings would be the way to go. I believe Cubase actually do just that. And besides, how often do you change the sample rate of an ongoing project? Reaper's item info "RESAMPLED from..." is kind of misleading I think.