Totally Gapless Burning - The One Sure Way

[dex Otaku]

2005-11-13: It's been a while since the original version of this was entered or edited, and it's about time it was updated to reflect the board's discoveries since then.

Both others and I have discovered the hard way that when making line-in recordings, HiMD recorders' auto trackmarking function [which can not be disabled] causes short repeated sections in the recording where track boundaries occur. Actually - sometimes they aren't there at all, sometimes they're only a few samples in length, and sometimes they're as long as 1/3rd of a second [going by my recordings].

This caused me to revise my suggested method for gapless burning, since the old way I had detailed involved using SonicStage's track combine function, which in the case of line-in recordings will be completely unreliable and cause much more grief than simply dealing with individual files in a non-destructive timeline editor such as Audacity, Audition, or Vegas.

Cheers,

D.

------------

There is only one sure way to maintain completely gapless recordings when burning CDs, regardless of what your original source is: edit all contiguous tracks as a single file.

The reason why:

Audio CDs store data in sectors that are a specific length. To fit in that sector size, audio is stored at 75 frames per second [or 588 stereo samples].

Gapless playback can only be maintained if every track on the CD you are burning is an exact length in frames, not samples.. Rephrased: you can NOT edit CD layouts or the audio going into them [if it must remain gapless] with sample accuracy, ONLY with frame accuracy.

Any attempt to run two tracks together seamlessly where the first track is not an exact length in frames will fail.

Some CD software [such as newer versions of Nero] will automatically crossfade tracks to avoid this, but this should generally be avoided; it may work, but the point is being missed entirely - if you're writing an original CD layout, especially a master for one of your own recordings, the layout should be valid [i.e. redbook standard-conformant] to begin with. Anything else is simply asking for trouble, especially if you want to distribute your recording in any way.

Most other [and all older] CD writing software will simply pad the end of a track that isn't a full length in frames. This is where glitches come from.

Note that both MD and HiMD conform to the same framelengths used by CD-DA [audio CDs], and depend on tracks being exact lengths or rounded to the same 75fps to maintain gapless playback.

How to do it:

The only 100% completely reliable, it-will-never-fail way to write gapless CDs from gapless sources is to edit all contiguous tracks as if they are a single file and add track marks in your burning software later.

Note that you can carefully split files into single tracks and maintain "gaplessness" throughout as long as the tracks are kept to exact lengths in frames. I do not use and advise against this method because it's very easy to screw up if you're not careful. A change in the length of a track by a single sample will usually introduce an audible glitch in your recording.

Here is my current upload/editing procedure to give you an idea of how I do things:

<blockquote>* Make my recording

* Upload with SS

* Export all tracks as WAV in a folder called "unedited"

* Do my editing in a non-destructive editor such as Vegas, Audition, or even Audacity

* Render the edited layout to a single WAV file

* Add track marks in my CD authoring software

* Burn my CD to CD-RW

* Rip it as a WAV image with DAO cuesheet using EAC [making sure EAC is set up for correct read offsets]

* Convert the WAV file to FLAC with embedded cuesheet

* Add whatever extra tags are required, including production notes, to the FLAC image</blockquote>

I never, and I repeat, NEVER have problems with gaps, glitches, or anything of the sort, and my end result is something I can burn directly from Nero for distro purposes, including CD-Text labels and all. I can also drop the FLAC disc image into programs like Foobar2000 and it will automatically recognise the embedded cuesheet with full track information and simply play as though every track were a separate file - and gaplessly!

Procedure in detail:

When I make recordings, the track marks on the HiMD itself are basically irrelevant to me as anything other than a guide. Far more important than marking tracks is using group recording mode, where each time you hit stop and start recording again, the unit creates a new group. With most of my recordings, each group created represents the beginning of a single session - i.e. one period of contiguous time - for editing.

1. Upload your tracks.

If you need advice about uploading, see the uploading FAQ.

As an addition to the advice in the uploading FAQ, I find it helps to include the group number and track number when naming tracks on the HiMD or in the SS library. This makes it easier to identify what parts should be combined and which shouldn't. I title each group with the date, group, and track within group - before uploading. This ensures that tracks are meaningfully titled and that files are meaningfully named when you export them to WAV. An example of a list of tracks in a single group I'm going to upload would be:

2005-11-13-test recording-g101

2005-11-13-test recording-g102

2005-11-13-test recording-g103

The next group would be:

2005-11-13-test recording-g201

2005-11-13-test recording-g202

&c. ..

This makes it obvious which parts belong together, not only in order but in terms of when they were recorded and what is most likely to be contiguous on the original disc.

Naming tracks *before* upload means that SS will name the actual files with whatever title you've given them, which can reduce the hassle of dealing with SS's default naming system.

2. Export your tracks to WAV. This is simple is SS 3.x - just select a range of tracks in the library, right-click within the selection, and choose "Save in WAV format..".

I'd suggest putting the files somewhere meaningful. Following the above example, I'd make a folder called "2005-11-13 - test recording", then make a subfolder in it called "unedited", and dump the files there. Restated for those of you who can understand paths: I'd put them in "d:\2005-11-13 - test recording\unedited".

For those of you who use FLAC and Cool Edit / Audition, you might want to convert to FLAC now. For speed's sake, I do all my editing with straight PCM files, just to reduce a bit of overhead.

One advantage to using the FLAC method if it's available to you is that you can tag FLAC files, whereas there is no strictly-adhered to standard for metadata with WAV files. This means you can't add information to WAV files that more than a few programs will be able to understand.

3. Backup the tracks in case of catastrophe. It takes space and a bit of time, but can save your rear if you do any destructive editing, whether on purpose or by accident. Making a second copy on your hard disc is fine if you trust your hardware, and burning a DVD- or CD-ROM is another option.

I rarely make backups this way before editing any more because I do all my editing non-destructively. In any case, the originals uploaded into SS become the definitive pre-edit master - you can export as WAV again if need be.

4. Drop the tracks into your editor's timeline in order, then make sure the ends either meet or overlap correctly.

Be sure to check that tracks that appear to be bumped together in your editor actually are - and if your recording was made with line-in, make sure that possibly repeated sections are either crossfaded or intercut correctly. Zoom in at junctions, make sure tracks meet end-to-end, and even zoom in "vertically" [amplitude-wise] if you're editing line-in recordings - you can usually see those brief repeated sections to edit them out.

5. Edit your audio.

The important thing here is for your assembled timeline in your editor to have no gaps in it. If you can play it gaplessly in the editor when you're done, then everything is going right.

An important point to note is that if you edit an entire album as though it's a single file [i almost always do], you can apply effects such as compression, normalisation, &c. to the entire mix from beginning to end. If you, like I do, tend to not alter your record levels once you've started recording [or at least, after the first track or so], your entire recording then has a consistent reference level; you just fix things to the first track [or the point after which things don't change] and apply processing to the entire album in one go. The extra time it appears to take to process an entire album is usually far less than the amount you can waste on processing each track individually.

6. Render [save as.. in some editors] your edited audio to a single WAV file for each section that must be contiguous. I usually render an entire CD as a single WAV file.

For those of you who might be using Sony/Sonic Foundry's suite of editing apps, you can set Vegas or Sound Forge to follow 75fps and add regions which CD Architect can convert dircetly to tracks later. Whether this works between other applications is iffy at best, but may be worth trying if you want to.

For really simple editing jobs, CD Architect can be used directly with the original WAV files exported from SonicStage, as it basically is a timeline editor itself.

Both Vegas and Adobe Audition can be used to make a CD layout and burn discs directly from within themselves, as well. How you do it depends on how you prefer to do it.

I tend to do my editing in Vegas, then my CD layout in CD Architect, because each tool is optimised for its own purposes. The one thing CD Architect in particular lacks is CD-Text creation [which it lacks because CD-Text is technically not redbook compliant and CD Architect is made for making fully-compliant masters].

7. Import the rendered WAV file into your CD authoring software and manually add track marks.

Most CD authoring programs that are even half worth their salt will allow you to insert track marks in a large file, or split a large file into tracks. CD authoring software always automatically conforms to 75fps [they have to, otherwise they're not burning audio CDs], so there's no worry about gaps or glitches.

Programs like Nero allow this. The slowest part of doing so is waiting for Nero to draw the waveforms for your tracks. Once this is done, navigation is usually pretty quick. Tracks are usually fairly obvious on a visual timeline; in the case of continuous mixes like DJ mixes, it'll take some shuttling back and forth to find where you want the markers to be.

That's basically it.

The point behind all this is, once again: if you split your audio up into tracks before putting them into your burning software, you likely will end up with glitches - unless you know exactly what you're doing, in which case you don't need this howto.

Cheers and happy editing.

Edited November 21, 2005 by kurisu

[corduroy]

tried to combine tracks that were in .wav but the sonicstage can't do it because it's not a supported file format, it says...

i can combine only .oma?

can i use soundforge or another software for doing this operation?

is nero a CD authoring software? i don't understand what exactly means CD authoring software

Edited June 29, 2005 by corduroy

[The Low Volta]

tried to combine tracks that were in .wav but the sonicstage can't do it because it's not a supported file format, it says...

i can combine only .oma?

read the post again:

* Export all tracks as WAV in a folder called "uncombined", just in case

* Combine all contiguous (read: gapless) tracks with SS

* Export all tracks as WAV in a folder called "combined"

so import (=oma) backup separate tracks (=wav) combine original oma's, convert combined tracks as wav

can i use soundforge or another software for doing this operation?

yes, but it will lead to gaps (like audactiy, see the downloads section)

is nero a CD authoring software? i don't understand what exactly means CD authoring software

software to write audio-CD's...nero allows to put trackmarks in a long wav, so it will look like two tracks on the CD, but there will be no gap inbetween them (other software could also do this)

Volta

[dex Otaku]

CD authoring = CD creation. Look up the word "authoring."

[Artstar]

Another way (the way I do it) is with CD Wave (draws up a .cue file) for the splits and then CDRwin or Alcohol for the burning. I'm a bigger fan of Alcohol than Nero lately.

[SkttrWave]

Yesterday I made a recording and I forgot to put the machine on hold. I accidentally paused the recording for a second or so. So now I have 1 wave file of 60 and one of about 5 mins. When I try to combine these using SS 3.1, after combining for a while it says: "an error occurred during the combine."

Is this just because there misses a second so they aren't really contiguous or shouldn't that be a problem?

Thanks in advance

[dex Otaku]

It could be that SS doesn't like the length of one of the files, though I have successfully combined multiple files of lengths ranging from 1 second to 75 minutes, with both HiSP and PCM recordings.

Please, if future readers have questions about combining/splitting, &c. - please post them in the software forum. This thread isn't meant for troubleshooting purposes, even if it has been left open.

[riotshield]

It could be that SS doesn't like the length of one of the files, though I have successfully combined multiple files of lengths ranging from 1 second to 75 minutes, with both HiSP and PCM recordings.

Please, if future readers have questions about combining/splitting, &c. - please post them in the software forum. This thread isn't meant for troubleshooting purposes, even if it has been left open.

can you explain the how to make single audio tracks that are 75fps. how many ms equal to 1 frame?

thanks

[riotshield]

can you explain the how to make single audio tracks that are 75fps. how many ms equal to 1 frame?

thanks

ok,i figured it out. its impossible to work it out in decimal view, in ms that is. you have to use the 75fps view in cooledit or any other program and cut at exact frames, or subdivisions of 588 samples.

you opened my eyes,thanks!! i always wondered why there was glitch between tracks.

[dex Otaku]

A note to those of you who read this:

SonicStage v3.1 and 3.2 DO NOT COMBINE HiSP CORRECTLY.

You are better off simply bumping tracks together in your editor after exporting to WAV, without combining them in SS.

[greenmachine]

Neither does it combine PCM perfectly, there are short (<0.5 sec) repetitions around the combining point. I have to try it in an audio editor instead, although that makes the process way more complex than it needs to be. Proper MD recording is becoming a science.

[dex Otaku]

Thanks, greenmachine - I'd experienced the same problem with HiSP, but hadn't used the combine function since then [i.e. hadn't tried it with PCM].

I guess that makes it official, then:

Don't use SS's track-combining functions at all.

[ozpeter]

Trouble is, with auto track marking, you get short repetitions around the track mark point whatever you do.

Leastways, that was my conclusion a couple of months back which I reported here somewhere...

Edit - found it - there's this one and this one on the subject.

Edited October 6, 2005 by ozpeter

[dex Otaku]

Oh? That's news to me - but would also explain the problem I had, since I was recording [30+ hours of material] through the line-in. I must test this eventually. Sounds like a hardware bug to me.

[ozpeter]

dex, note my edit to the above post if you want to dig deeper sometime.

[dex Otaku]

Thanks, ozpeter. It would still appear that the best solution is to avoid combining in SS, and to bump tracks together in your editing software. At the least, this would provide the opportunity to check for the short repeated chunks because the join is done manually.

[incidentally - I'm still glad I combined tracks before editing in the case of at least one disc - 6 hours of audio in 485 tracks. TOOOOO much time to do manually.]

[ozpeter]

Hmm, combining that number of files using whatever that combining package was I found would be an interesting test. Of course the key thing now is that one can go about it different ways during testing without SS deleting the files at the start of the second run! Which makes me think maybe another round of testing may be warranted. I'll see what time allows.

[dex Otaku]

Bump. Updated 2005-11-21.

[SlowMo]

hi,

here is "my way" of doing it:

i was primarily interested in a more convenient (or any) way to get formerly gapless albums that had been ripped to mp3 (not gapless~) to play on my md gapless again, with trackmarks and tracklist info. specificly mixed albums.

So i import those mp3 tracks into soundforge (44.1/16) altogether, create a new file at the same rate and copy/paste the tracks into the new one in the right order. The important thing here is, to set a marker at the end of each track before pasting the new ones. Then i remove the gaps.

I save the file, open Nero (v5.5), create a new audio cd, and put that file in it. Now i right klick the file ->properties, and go to the tab "indices, borders, divide" (i don't know the exakt wording in the english nero version, mine is german. it's the middle tab.)

The cool thing here is, that Nero gets the marker positions inserted by Soundforge, so i only have to press "split at index positions" and i'm done. I set the pauses between the tracks to zero and burn it as an image.

Image gets mounted, open Simple Burner, and hopefully Gracenote works, so i get the album- and tracknames...

Don't know if that helps anyone, but i'm pretty happy with it.

P.S. i wondered about the 75fps limitation, and thought i'd avoid it by putting a whole single file into nero, but fact is, that i didn't pay attention to where the markers were. still all three albums that i've tried play seamlessly in my md.

[dex Otaku]

What you're doing, SlowMo, is basically the same thing described above. It's just slightly more work with Sound Forge [since it doesn't allow manipulation of multiple files on a single timeline].

[BlastM]

If you're ripping from a CD, use the optical output from a CD player or DVD player (nearly every cheap DVD player has optical out). You can enter all the track information in Sonic Stage. The tracks will be totally gapless and the CD/DVD player sends a signal letting your Minidisc player know about when each track starts.

[dex Otaku]

This howto isn't about ripping. It's about keeping your own recordings gapless when burning.

[boojum]

2005-11-13: It's been a while since the original version of this was entered or edited, and it's about time it was updated to reflect the board's discoveries since then.

Both others and I have discovered the hard way that when making line-in recordings, HiMD recorders' auto trackmarking function [which can not be disabled] causes short repeated sections in the recording where track boundaries occur. Actually - sometimes they aren't there at all, sometimes they're only a few samples in length, and sometimes they're as long as 1/3rd of a second [going by my recordings].

This caused me to revise my suggested method for gapless burning, since the old way I had detailed involved using SonicStage's track combine function, which in the case of line-in recordings will be completely unreliable and cause much more grief than simply dealing with individual files in a non-destructive timeline editor such as Audacity, Audition, or Vegas.

Cheers,

D.

------------

<SNIP>

Or, you can pull them all into foobar2000 and select convert to a single file in the same folder, click the button and you are done. That is how I do it. Now that's simple!

[King Ghidora]

The cool thing here is, that Nero gets the marker positions inserted by Soundforge, so i only have to press "split at index positions" and i'm done.

Nero also recognizes cue points from wav's edited with Goldwave as index positions.

Edited February 13, 2007 by King Ghidora

[Ferenc]

I always combine tracks in the recorder. So when I upload there is already a single file.

[lyn119]

Hi, I am a spammer so lame that I can't even create my own content. I have to copy and paste from other posts!

Please ban me, I suck.

[Avrin]

Adobe Audition is very convenient for combining. Just click File->Open Append, select files to combine, and it will open them on a single timeline. Then you can save the timeline as one big WAV.

[SileEeles]

They way I do it, and I have only had to do it once, since not much of my music requires gapless, is just let the album play in Media Player. Media Player supports gapless, and all you have to do is push the T.Mark button on the recording device so it creates a new track. Obviously this method isn't too accurate, but for my purposes I wasn't trying to be. You can go back and separate tracks later and I imagine it would be fine. Saves combining all the tracks in a different program, saving it, adding track marks and so on before you even start the recording to minidisc process. Well, works for me anyway.