recording 2GB+ wav's in traction (title edited)

[mmmatt]

Some DAW progrms can be set to split files longer than 2GB. Maybe this is why.

Matt

[valley]

When the header is not useable - so is the file.

Not necessarily true. In most cases (though it is convention not law) the data chunk of a wav file is trhe last part. If the header stupidly defined the chunk as being shorter than it actually is, you'd just get truncated playback. Problems would only arise if the wav reader attempted to decode the content at the end of the file. Most wav reading code thopugh tends to err on the side of lax about such things, 'cos there are so many malformed wav headers out there.

Back when I wrote a load of wav reading utilities, I was constantly amazed just how poorly formatted some of the headers were that I encoutered.

Aiff also has a 32 bit header, if I'm not very mistaken.

That's a bummer then.

[valley]

Some DAW progrms can be set to split files longer than 2GB. Maybe this is why.

Yup.

What I'm trying to research is whether it is possible to use a multisample wav to get around the 32bit addressing issue. If it is, I'm sure some apps will do it.

I seem to have lost my collection of notes on wavs though.

[jens]

seems to be wrong nonetheless...

Author: Aleksey Vaneev
Date: 02-09-05 23:19

It supports files up to 4 GB, but not above this point. Going above would require Wave64 support which is not implemented.

http://www.voxengo.com/phorum/read.php?f=16&i=19&t=19

Aleksey usually knows his shit...

[valley]

actually that's correct, and I should have spotted it[1].

32bit can address 4Gb of data only. HJowever, stereo would double the bandwidth, so a 4gb mono file equals a 2gb (2x2) stereo. Sorry for the confusion

[1] I've been up 36 hours straight trying to get linux to work properly on a laptop, and my mind is slipping.

[aldi]

i'm really sure that depending on the way tracktion handles reading of the files it doesn't matter at all. the header contains the format of the wav and that's the only real important information, after that the readinroutine can jump to the data the wave contains and read it as the format defines till there's no data left (means the file ends) the maxsize only depends on the filesystem (ntfs, fat32..). if jules had liked to, he could also write the information raw to disc without any header.

[Pädy]

@ valley: that's pretty much what I expected - or do you consider truncated clips useable.
@ Jens: 4 GB is fine, but what format 16 bit, 44.1 kHz? Stereo or Mono?

[Mr. Tingle]

but why should they not appear at all?

(from original post)

[valley]

@ valley: that's pretty much what I expected - or do you consider truncated clips useable.

I don't consider it *not* useable.

So, after reminding mysel;f of some stuff:

The actual wav spec (as opposed to some of the extension hacks that people have made over teh years) says that:

a wav file must have one, and only one data chunk. This data chunk can be up to 2^32 - 1 bytes long (4GB).

multi-sample wav files, are both broken, and also bound by the 4GB limit, as the multisaple IFF list is embedded in the data chunk.

4GB at stereo 44k 16bit gives about 7hours worth of recording.

Double the sample resolution (ie 24 or 32 bit) and you halve that time. Double the sample frequency, and your halve it again.

IE, about 2 hours.

[valley]

but why should they not appear at all?

(from original post)

that's a bug - pure and simple.

If the OP is using an NTFS drive, then Tracktion should be able to warn that data cannot be convereted into a wav file, and either store it as a raw data lump, split into multiple files, or offer to downsample it into a format that can be stored.

[jens]

.
@ Jens: 4 GB is fine, but what format 16 bit, 44.1 kHz? Stereo or Mono?

irrelevant - 4GB whatsoever - 4GB is 4GB, or not? - of course the sample-lenght can be twice as long if it is mono (if both have the same bit-depth and sample-rate) but in both cases it's still 4GB

[Pädy]

Okay, let's quickly summerize and then I'm off:

I am right, there's a 32 bit header restriction in the wav format.
This translates into 2 GB worth of data - 4 GB stereo.

Most probably this restriction is the cause for the thread-starter's probs.

Cheers, Pädy

PS: Now, I know again why I stopped coming here so frequently. It's just not possible to give a small bit of information without getting involved in a cockfight.

[valley]

I am right, there's a 32 bit header restriction in the wav format.

yup.

[uote]This translates into 2 GB worth of data - 4 GB stereo.[/quote]

nope - just 4GB. Your 4 doesn;t go so far with stereo.

Most probably this restriction is the cause for the thread-starter's probs.

yup.

PS: Now, I know again why I stopped coming here so frequently. It's just not possible to give a small bit of information without getting involved in a cockfight.

yup...

[jens]

This translates into 2 GB worth of data - 4 GB stereo.

thast's the part of the equation I don't understand!

how can 2GB be 4GB?


Matt wrote that he records at 24/96 - if it is stereo that means 4GB are a bit more that 2hours and ten minutes - and if it is stereo it is a bit more than 4 hours and twenty minutes, right?

[jens]

oh, and Pädy:

so far you've been the only troll in this thread!

Matt writes about a fact he experienced first hand and because you read something somewhere in a manual you feel obliged to write:

Then either Jules has silently implemented the wave64 format or it's a miracle

so who's been the fighting cock then?

and now you're sulking because some of us dared to doubt your words... (and rightly we did)