Splitting 16/48 tracks?

[spidergawd]

This should be an easy one, but damned if I can find the answer anywhere. In the past I've always dealt with 16/44.1 and know all about SBE's, but I'm not sure how to go about splitting 16/48 into tracks. What's the most common method? How do people usually listen to 16/48 files, and does it even matter how they're split? A link to a tutorial would be great.

I don't necessarily need software advice, I prefer to understand how stuff works rather than letting a piece of fancy software do my thinking for me, but if it helps I run Slackware Linux.

Thanks!!!

[co9ol]

Audacity can do it (here's a good guide, make sure to read the "Extra notes about burning to CDs" section)

http://wiki.audacityteam.org/wiki/Splitting_recordings_into_separate_tracks

Although CD's don't support 48khz, I'm pretty sure you could still use it.
I think most people listen to them through their computer so SBE's probably wouldn't make much difference, but I feel like its still a good idea.

[spidergawd]

Audacity can do it (here's a good guide, make sure to read the "Extra notes about burning to CDs" section)

http://wiki.audacityteam.org/wiki/Splitting_recordings_into_separate_tracks

Although CD's don't support 48khz, I'm pretty sure you could still use it.
I think most people listen to them through their computer so SBE's probably wouldn't make much difference, but I feel like its still a good idea.

Thanks for the info, and yes I agree Audacity is an excellent open source editor. I'm mainly concerned if there is a sector boundary-type issue with 16/48 like there is with audio CD. In the past I'd always recorded shows to DAT @48kHz, and then downsampled to 44.1 to share. Now that 16/48 flacs are becoming more widely used, I'm thinking of going back to my DATs and upping some in 48kHz. The only thing I need help with is where to split the tracks, or if it even matters. I personally don't yet have the equipment to play back 48kHz files properly, so I really don't know how to proceed, and I'm not having any luck Googling it.

Maybe I'll just PM someone who's seeded a 16/48 show and ask them, and then post here in case it will help someone else.

[rspencer]

SBEs won't be an issue. They only matter if you're doing 16/44.1. If you aren't doing redbook standard, lots of software will give you an error if you try to do anything regarding SBEs.

Most audio players will handle 16/48. foobar does no problem, all the way to 24/96. Maybe higher, but I haven't done any higher. If you've already spread those shows in 16/44.1, I wouldn't bother to do so in 16/48. The difference is negligible, most likely not noticeable at all.

[Audioarchivist]

16/48 (or 24/48 or 24/96 files) technically can't have SBE's per se - those sample rates are not made for CD's. Having said that, if you take higher res files and downsample and dither them to 16/44.1, if they aren't split at the right places, they will possibly have SBE's at CD rates...

What I do when processing my 24/96 files is to only split tracks to the exact second. That is, no fractions of a second. CD sectors mean that each second is split into 75 frames. If you split a file between the points that these frames are, that makes a SBE. It's hard to tell where those other 74 frame boundaries are, but there's one at each perfect second. So, splitting a file to be exactly down to a perfect second with no fractions of a second will ensure that your hi-res files will never have any future compatibility issues with SBE's when they are converted to CD quality...

no fractions = no SBE's

make sense?

[weedwacker]

Samplerate has nothing to do with S.B.E.'s except when converting from one samplerate to another. S.B.E.'s are dependent on the register sizes of the processor processing the audio. It has to do with binary numbers if the procesor register size required is 16 bit based on the bit depth then the audio has to be able to be split into a sample size that is an equal amount of 16 bit samples. If it isn't you get an S.B.E., same goes with 8, 24 or 32 bit. Changing samplerates in simplest terms changes the file size or sample size so a split that was correct when done at one samplerate will not be an even split at another even if the bit depth hasn't changed simply because you add or remove samples from the conversion process. If you change bit depths like from 24 to 16 bit you'll have the same problem. Almost no programs I know of take into account 24 bit sample sizes except for cdwave editor. It is the only program I know of that when it does the splits, it uses a sample size that is a multiple of 16 and 24 so you won't get an S.B.E. regardless of the bit depth used. It is useful for large audio files in that even if you make a cue sheet you can reuse it for all bit depths and sample rates supported, you just need to leave a little bit of silence at the end and lob it off for the final track split.

[spidergawd]

SBEs won't be an issue. They only matter if you're doing 16/44.1. If you aren't doing redbook standard, lots of software will give you an error if you try to do anything regarding SBEs.

Most audio players will handle 16/48. foobar does no problem, all the way to 24/96. Maybe higher, but I haven't done any higher. If you've already spread those shows in 16/44.1, I wouldn't bother to do so in 16/48. The difference is negligible, most likely not noticeable at all.

Thanks. I was concerned that folks were burning to a format (HD-CD, etc) that I'm not familiar with, and wanted to know what the general consensus is on track breaks. I'll take your advice and not bother with most shows, but I do have one in particular, a Little Feat soundboard from 6-26-2001 where the band's tape deck didn't record (I sent my tape to Chris so he could copy it) that might be worth the effort. I originally upped it to usenet in 16/44.1 format, and it doesn't seem to circulate, may be worth the effort. We'll see.

Thanks again for the advice.

[spidergawd]

16/48 (or 24/48 or 24/96 files) technically can't have SBE's per se - those sample rates are not made for CD's. Having said that, if you take higher res files and downsample and dither them to 16/44.1, if they aren't split at the right places, they will possibly have SBE's at CD rates...

What I do when processing my 24/96 files is to only split tracks to the exact second. That is, no fractions of a second. CD sectors mean that each second is split into 75 frames. If you split a file between the points that these frames are, that makes a SBE. It's hard to tell where those other 74 frame boundaries are, but there's one at each perfect second. So, splitting a file to be exactly down to a perfect second with no fractions of a second will ensure that your hi-res files will never have any future compatibility issues with SBE's when they are converted to CD quality...

no fractions = no SBE's

make sense?

Makes very good sense, thanks for the tip. That's similar to what I used to do 10-12 years ago, before I knew about CDWave/WaveBreaker etc., I used a cue sheet to split to exact seconds with EAC. In those days I didn't have a digital sound card or a burner in my computer, I transferred DATs to CDR using a Tascam CDRW700 as one long track (letting the Tascam do the downsampling), ripped to PC with EAC, split tracks with EAC, compressed to shorten and uploaded to usenet. Now that I know a _lot_ more about the process I was considering redoing some of the DATs and leaving them @48. Probably not worth the effort, but maybe will use the technique with future recordings.

Cheers

[bluzman]

Thanks for this helpful and informative thread!

[Limulus]

my advice for DAT tapes or any digital media:

always do "unedited" digital 1:1 clones on harddisc and save them like that (unedited) multiple times (on data dvdr, harddiscs etc.). this way you're always goin to have the chance in the future to re-do the original recording of you. like DVD-Audio, CDR, whatever.

as for myself i prefer to create 16bit/44.1khz FLACs wthout SBEs from any audio recording in any bitrate i'm doing....for verifying the show and listening pleasure on CDR or mobile audio devices.
as for DAT 16/48 and downconverting to 16/44.1khz highest quality....my ears fail to hear a difference (and my ears are still good!!).

[spidergawd]

my advice for DAT tapes or any digital media:

always do "unedited" digital 1:1 clones on harddisc and save them like that (unedited) multiple times (on data dvdr, harddiscs etc.). this way you're always goin to have the chance in the future to re-do the original recording of you. like DVD-Audio, CDR, whatever.

as for myself i prefer to create 16bit/44.1khz FLACs wthout SBEs from any audio recording in any bitrate i'm doing....for verifying the show and listening pleasure on CDR or mobile audio devices.
as for DAT 16/48 and downconverting to 16/44.1khz highest quality....my ears fail to hear a difference (and my ears are still good!!).

Very good advice. The unedited clone option wasn't available to me until I got my Delta DIO2496 card in '03, that's one of the reasons I'm thinking of doing my earlier tapes over again. As for CDR, I can't remember the last time I burned a CDR for myself, but certainly understand what you mean. And I know there's no way I could hear a difference between 44.1 and 48, but my ears have been exposed to auto factories and rock and roll concerts for decades. I'm just thinking it's best to do it at as high quality as possible, on principle, even if I don't benefit from it personally.

Thanks!

[spidergawd]

Samplerate has nothing to do with S.B.E.'s except when converting from one samplerate to another. S.B.E.'s are dependent on the register sizes of the processor processing the audio. It has to do with binary numbers if the procesor register size required is 16 bit based on the bit depth then the audio has to be able to be split into a sample size that is an equal amount of 16 bit samples. If it isn't you get an S.B.E., same goes with 8, 24 or 32 bit. Changing samplerates in simplest terms changes the file size or sample size so a split that was correct when done at one samplerate will not be an even split at another even if the bit depth hasn't changed simply because you add or remove samples from the conversion process. If you change bit depths like from 24 to 16 bit you'll have the same problem. Almost no programs I know of take into account 24 bit sample sizes except for cdwave editor. It is the only program I know of that when it does the splits, it uses a sample size that is a multiple of 16 and 24 so you won't get an S.B.E. regardless of the bit depth used. It is useful for large audio files in that even if you make a cue sheet you can reuse it for all bit depths and sample rates supported, you just need to leave a little bit of silence at the end and lob it off for the final track split.

I've been experimenting with splitting 48kHz wav files at exact seconds, and I'm finding that after downsampling to 44.1 shntool's len function shows SBEs even though it shows the length as an exact second. Much of Weedwacker's post goes over my head, but I think the gist of it is that sample rate conversions will alter the data to the point that it's not possible to split 48kHz files such that a downsample will be on a sector boundary for everyone.

So, unless someone can give good reason otherwise, it seems that the answer to my original question is that it really doesn't matter where 48kHz files are split. Anyone listening to them at 48 won't care, and anyone downsampling to burn to CDR will have to fix them anyway.

Thanks again to all who have replied.

[faninor]

I just use CD Wave to split tracks regardless of sample rate. All the usual sample rates (except 32kHz) can be divided evenly by 75 so I just split everything at the nearest 1/75 second to where I want.

When working with high resolution audio I usually make a 16bit/44.1kHz copy too, so I make that copy before splitting tracks, open either copy in CD Wave to set the track marks, save a cue sheet, and apply that cue sheet to all versions. It's not much extra work for me. In theory you can probably just make the high resolution copy and later convert each track to 16bit/44.1kHz for CD without any trouble. If the tool you're using does not alter the duration and handles the very beginning and end of the file nicely -- I've never bothered to look into it.

[spidergawd]

I just use CD Wave to split tracks regardless of sample rate. All the usual sample rates (except 32kHz) can be divided evenly by 75 so I just split everything at the nearest 1/75 second to where I want.

When working with high resolution audio I usually make a 16bit/44.1kHz copy too, so I make that copy before splitting tracks, open either copy in CD Wave to set the track marks, save a cue sheet, and apply that cue sheet to all versions. It's not much extra work for me. In theory you can probably just make the high resolution copy and later convert each track to 16bit/44.1kHz for CD without any trouble. If the tool you're using does not alter the duration and handles the very beginning and end of the file nicely -- I've never bothered to look into it.

I've tried converting a 16/48 show that I downloaded, and it didn't work out the way I expected. The tracks were split at the same times as an SBE-free 44.1 version, but after downsampling the 48kHz version to 44.1 the tracks were not split on sector boundaries. Perhaps if I'd used a different CPU and/or software it might have been, I don't know. If the converted file has just one extra/less sample (per channel) than a multiple of 2352 byte sectors that would be 1/44100 = 0.00002268 second too long or short, not easily detectable and probably not at all unlikely in a sample rate conversion. (When I mentioned "exact second" in my previous post shntool len was only displaying the length to 3 decimal points)

If anyone has converted a 48kHz show to 44.1kHz and found it to be SBE-free, I'd appreciate a link to the torrent.

[faninor]

That probably has something to do with the software you used. I haven't tried it, so I can't tell you what program would handle it perfectly.

In theory, a program that's careful about it should be able to give almost identical results regardless of if you downmix before or after splitting tracks. The only real difference should be that the very last sample wouldn't be interpolated using the data from the next track. I think.

[Audioarchivist]

As an example of it working properly for me, the six show run of Heart shows I saw 1 year ago (today is anniversary of one of them) were recorded at 24/96. I uploaded 24/96 master clone versions here of all 6, all split on "perfect seconds" in Wavelab. I used these files as the basis of my remastered versions that were EQ'd, then downsampled and dithered to 16/44.1. The finished files had no SBE's or any other errors of note.

I remember that I did some experimentation, mastering from the full wav files as well as from the split wav files. The end result was the same. Maybe the program that you used to split files isn't as accurate as Wavelab? I don't know. I can report, though, that I've never had a SBE from splitting on a second - it takes some extreme zooming in to find the no-fraction zone...

Just about everything I've done for the last couple years is put in the computer as 24/96, and split "to-the-second" and it's always correct. I split and master with Wavelab in 24/96, then use Sox resampler via Foobar to downsample my files, and iZotope RX advanced to dither. The 16/44.1 files that come out the other end have no errors. maybe I'm lucky? No. If it's done right, it works...

[spidergawd]

Ok, it was the software. I'd used Audacity for everything. The "snap-to>seconds" function worked perfectly, the error came from using Audacity to resample. I resampled the same file with command line Sox and the result was perfect. Both results have a basic 44-byte header, but Audacity's output is 4 bytes larger than Sox's output, so Audacity "may" have added one extra sample per channel.

So, it seems that the best method is to split on seconds, or possibly 1/75 seconds (will experiment), and leave it up to anyone downstream wishing to convert to CD to use appropriate software.

Thanks for the schoolin'

[spidergawd]

....and it also worked properly when using Audacity to split 48 kHz file at 75 fps and then using Sox to resample to 44.1

[spidergawd]

....but does _not_ work when splitting the 48kHz file with wavbreaker and resampling with Sox. I don't have a means to try CDWave.

[Drgiggles1]

@ Audioarchivist -- Wavelab 6 I know does everything and more that CDWAVE does as long as the settings are setup correctly. I would assume even if you did not cut on the exact second you would still be SBE free assuming your settings are correct. If that's the case you may be wasting time zooming in and all and doing the way you proceed. I know CDWAVE cuts on sector boundary regardless of where you split.

[dorrcoq]

I always split my 24/48 or 24/96 files with CD WAVE, never had an SBE when I downsampled to 16/44.1.