Taperssection.com
Gear / Technical Help => Ask The Tapers => Topic started by: phanophish on July 20, 2012, 10:48:02 AM
-
It seems to me that a lot of people choose to tape at 24bit/48kHz sample rate. I don't really understand why? I totally get the higher bit depth and understand the value of the additional headroom, but the 48k sample rate makes no sense to me, unless there is no plan to ever dither the recording to 44.1 for "standard" playback. I would think from an audio quality perspective you would be better off running 24/44.1 or 24/88.2 so that the resulting file is a more simple dither and relies less on an algorithm that will require interpolation of the actual wave form to take a odd re-sampling of the sample rate and to the final format. I suppose if you are using a PC for playback and want to play back the 24/48 recording then that might be a argument to be made for the higher 48k sample rate, otherwise it seems to me to be a loose/loose (larger file size & lower end quality) decision.
-
I would think from an audio quality perspective you would be better off running 24/44.1 or 24/88.2 so that the resulting file is a more simple dither and relies less on an algorithm that will require interpolation of the actual wave form to take a odd re-sampling of the sample rate and to the final format.
I think a lot of tapers aren't targeting CD as the final product anymore.
So the highest practical resolution is used. Going to the higher (88/96) sample rates will just fill up your hard drive and make file processing take longer.
And - I think "simple math" thing has been debunked somewhere here before...
-
I would think from an audio quality perspective you would be better off running 24/44.1 or 24/88.2 so that the resulting file is a more simple dither and relies less on an algorithm that will require interpolation of the actual wave form to take a odd re-sampling of the sample rate and to the final format.
I think a lot of tapers aren't targeting CD as the final product anymore.
So the highest practical resolution is used. Going to the higher (88/96) sample rates will just fill up your hard drive and make file processing take longer.
And - I think "simple math" thing has been debunked somewhere here before...
yeah, "simple math" has been debunked.
The original reasoning that I heard was to extend the hardware aliasing rolloff further up and then do a software resample later which (if you have a good SSRC) has better control of that top end (>20khz). I'm not sure modern gear has that problem, so it may now be for DVD-style archiving or stuff of that nature.
-
I've been recording at 24/44.1 for years for that exact reason, but oddly enough, for just the past few shows I've been recording at 24/48.
For my own listening, for quite some time now it has been mp3's on an ipod/iphone/ipad for car and on-the-go listening, 24 bit listening on my computer (I have a pretty good computer playback system), or 24 bit listening via a Squeezebox (which does 24/48 natively) for my main playback system.
Given that I have no need for 44.1k sampling, I've just started recording at 48k, somewhat as an experiment to see if I'm ok adding in the additional 48>44 processing step for those who want CDR listening. That mainly as a time and workflow issue, not a sound quality issue. I guess what I really should do is start sharing my 24bit filesets along with the 16bit filesets for those who care for the highest quality.
Frankly, I seem to be in the very small minority who feel that 16bit flacs are actually lossy versions of my recordings, not lossless (um, I threw out 33% of the information, how is that lossless?). In that vein, I've somewhat stopped worrying about having the best quality for 16bit filesets. As an example, I picked up Wave Editor which has Ozone's MBit+ dither algo which I find to be excellent. But I never dug WE as my editor, so for a while I tried editing in Amadeus, then loading the 24bit file into WE for MBit dithering. I found I just wasn't motivated for the extra time, so now I just dither with the free mda dither plugin available in Amadeus.
I have no idea how good that dither routine is, and really don't know whether a 320kbs mp3 made directly from a 24/48 fileset sounds better than a 16/44.1 flac fileset that has been dithered down with the probably pretty crappy mda dither routine and then re-sampled to 44.1 with the basic resampler in Amadeus. I have no doubt I could do a better job putting together a 16/44 fileset if I used only the best dither routines and re-sampling programs, but at that point, isn't it easier and better to just provide the original 24/48 fileset to those who care about the quality?
No offense intended to those who want quality but are limited to 16bit listening. Then again, it is getting pretty easy to do 24bit listening these days, perhaps time to upgrade to 24bit if you are in that camp.
-
i haven't mastered anything to 16/44.1 in a long time.
24/48 strikes a nice balance of quality and file size
-
Interesting to know the "simple math" re-sample thing is sort of a moot point, anyone have links to those discussions? I suppose then it all boils down the to quality of the re-sample algorithm which is probably much more subjective and I can only imagine could be a pretty extensive debate.
I pretty much agree with where you are regarding playback Todd. I'd say I really have 2 forms of playback. Direct from my native FLAC files on a quality playback system where I think quality really does matter (and could see the argument for 24/48) and in my car on the go. If I am listening in my car it is MP3 playback and quality is less of an issue. MP3 is good enough. I'm kind of overall wondering if 16/44.1 playback is even worth targeting/considering any more. I'm inclined to say yes because of the ubiquity of CD players in vehicles and everyone familiarity/ability to rip to the mobile device of their choice from CD, I usually just burn a CD for anyone except fellow tapers. Trying to explain FLAC playback etc is just not worth it. Sounds like a lot of people are taking the 24/48 to reach the best Quality/filesize balance, MP3 for casual listening, and I'll dump out 16/44.1 at a whatever quality happens by accident for the 16bit limited folks who probably don't know the difference anyway.
-
Video. 48K is a standard for DVD and HD video.
-
i haven't mastered anything to 16/44.1 in a long time.
24/48 strikes a nice balance of quality and file size
I'm the same, I stopped making 16bit files about 2 months ago, and I'm never turning back. Well, except when the band wants a CD. For me, there were 2 major reasons for this: 1) as tapers we serve a purpose as historians, the highest quality files are the ones that should be archived. 2) I listen to 24/48 flacs and think they sound better that way so it's how I prefer folks to listen to them. Not to mention that it does seem to fend off the reposters a bit. My friends that are serious music fans have all learned how to play back my files, and a couple are OK with making mp3s for themselves with xACT etc.
-
Maybe I'm just hearing things, but when I listen to 24/48files (what I record to, typically) and convert them to 16/44.1, the files sound distorted. I've noticed that I don't have that problem when I downsample to 16/48, though, so I go with 24/48 first and foremost and go from there. I usually upload to LMA & etree at 16/48, as well.
Obviously, there's no problem listening on my home speakers, which are connected to my computer, but now with the ability to play FLAC on the Winamp Pro app on my Android ($4.99 to be able to listen to FLAC in my car is a super deal, IMO), I don't see much reason to have ANYTHING ready for CD playback.
-
Maybe I'm just hearing things, but when I listen to 24/48files (what I record to, typically) and convert them to 16/44.1, the files sound distorted. I've noticed that I don't have that problem when I downsample to 16/48, though, so I go with 24/48 first and foremost and go from there. I usually upload to LMA & etree at 16/48, as well.
that sounds like a conversion or resampling error (or a bad SSRC). Also, make sure you don't amplify audio all the way up to 0dbfs and then run it through the converter, that could yield clipping as well. Another possibility I've seen is where computers receive audio in any sample rate, but pipe out a specified rate and do the conversion on the fly. Food for thought.
Well, except when the band wants a CD.
There is the big hitch preventing me from dropping redbook audio compliant files. It used to be a two-fer with bandwidth; I've never uploaded 24 bit files because they are huge for the bandwidth I have and there is almost zero ROI for me to go for the bigger file size. Now that I don't seed much stuff, it's just the band copies. I have one band that I need to send over the original pre-processing 24bit wavs for, but everyone else (including some members of that band) wants redbook compliant cd audio if they even want that. Some are content wtih V0 mp3s. :P Last, when recording a PA, the noise floor in the room is higher than any of the equipment I use, and when I'm recording an onstage production in a quiet room, I'm applying compression just about every single time when I polish it. If bands just wanted mp3s, I'd consider dropping the dither only because it would shave a few minutes of processing time off of the render function, but thats the only reason I can come up with.
So because I'm generally targeting 16bit already, generating a 24bit set (which is just greater dynamic range) for my own personal listening is sort of pointless. It's important to record in 24bit because of the additional dynamic range that I encounter during onstage recordings (before all of that hits the sound guy's compressor), but for listening I didn't get a benefit.
I don't listen to music loud (period) and generally try to take care of my hearing. I get my hearing tested professionally every other year or so and wear a minimum of -25db plugs to anything remotely loud (and -30 plugs to some shows). I'm just under 30, and I can't hear above 17khz, so 44.1 as a sample rate is very sufficient.
-
Strongly recommend that folks get up to speed on foobar2000 -- It allows for the use of a variety of resampling/dithering options including a nice implementation of SOX -- I load my native 24 bit flacs (I record at 96 k) that I've already tracked and tagged into foobar2000 and use the appropriate conversion preset to create 16bit 44.1 files or 320kb 48k mp3s in a temporary directory of my choice. Takes my older quad core computer only a few minutes to convert and save in either of these formats with the metadata tags intact. This lets me process everything in its native resolution and quickly "dumb down" on demand whenever needed for my iphone or a cd.
-
Video. 48K is a standard for DVD and HD video.
exactly DVD's are usually mastered in 24/48
it works well
I only do conversions since many still want them, and use sample manager
however I also like a final 16/44.1 for my rockbox'd iPod to listen to in the car...
-
24/48 is the audio for video standard - I would not record that way for CD.
For CD I record 24/88.2 and dither down to 16/44.1 after mastering.
On a small portable recorder with internal mics I record at 24/44.1
-
----------
24/48 is the audio for video standard - I would not record that way for CD.
----------
Few times I used 16/44 sound for HD video in mkv. Works fine. Is something wrong with this?
-
----------
24/48 is the audio for video standard - I would not record that way for CD.
----------
Few times I used 16/44 sound for HD video in mkv. Works fine. Is something wrong with this?
nope, for digital files there is a ton of leeway. John's referencing hard media like dvds.
-
Something that hasn't been mentioned is that editing in post using 24/48 files allows you to, for instance, raise very low levels and not introduce any significant his. I find that the editing advantages make this format desirable. 24/48 files are not too large, so that editing speed is not hampered. I imagine that higher numbers would yield even better editing results, but then speed will be an issue. Just my 2 cents ... :-)
-
Something that hasn't been mentioned is that editing in post using 24/48 files allows you to, for instance, raise very low levels and not introduce any significant his.
when compared to 16bit, sure, but if you're talking about in comparison to 24/44.1, I find that very difficult to believe...
-
Video. 48K is a standard for DVD and HD video.
thats why i do it
and 24/96 is overkill to me at this point in time
-
It has been a few months since I last recorded a show. Recording 2 channels at 24/48 takes up 1gb/hour correct?
-
Correct, 24/48 = 1 gig per hour, 24/96 = 2 gig per hour.
-
Correct, 24/48 = 1 gig per hour, 24/96 = 2 gig per hour.
^^^
Obvious perhaps, but just to be clear, that's 2 channels (stereo) @ 24/48 = 1 gig per hour.
A minor advantage of 24/48 is that it makes for simple calculation of the memory required when recording multiple channels-
~0.5 GB / channel / hour.
[edit- my bad mr. kitty, I didn't see that you had specified 2 channels when you asked]