Why does Audacity have to transcode when splitting?

[susanne]

I have a WAV file that is 23.4MB in size and 1h 41min long. And I would like to split it in two files. The split point should be at 38min 02sec.

However I would like keep the file in its original audio format and NOT transcode it nor save it other formats.

So far I have used mp3splt for splitting mp3 files on Ubuntu, mainly because it does not transcode the files. Why on earth would I want to lose quality and by re-encoding the file anyway? Can someone explain it to me?

I googled for "audacity split" and followed the instructions here: http://audacityteam.org/help/faq?s=files&i=split

The result: I did get two files, but in total they were much larger than the orginal file. Like 70MB or something. Why did it do that?
Its just a standard WAV file. Can Audacity just split it without re-encoding or doing other unnecessary stuff?

Please help

[steve]

Audacity does not edit files directly but works with "Projects".
The main reason for this is to allow the program to work efficiently and at high quality. For example, imagine if you had a 3 hour recording in Audacity and you wanted to delete a couple of seconds from somewhere in the middle. If Audacity worked directly on one 3 hour audio file, then that would require that the entire file would need to be read from the hard drive and rewritten back to the hard drive. Considering that a 3 hour recording is likely to be around 2 GB or more of data reading and writing the entire file would be really slow. However, because Audacity works with small (1 MB) blocks of data, deleting a couple of seconds from the middle of a 3 hour recording can be accomplished almost instantly.

MP3 files are encoded, which makes it impossible to do much more than very basic trim and split type edits unless the audio is decoded first. If you only want to do basic trimming or splitting of MP3 files, then programs such as MP3Split are far more suitable than Audacity. However if you want to be able to more advanced editing and processing features, then the data that is being edited must be uncompressed.

As you say, there will inevitably be some amount of sound quality loss when decoding/re-encoding audio that is in a lossy compressed file format, and for this reason it is highly recommended that the source material should be in an uncompressed format if possible. Encoding to MP3 should be reserved as the last step in a production. (MP3 is a delivery format rather than a production format).

By default, Audacity is set to work in extremely high quality 32 bit float format. This allows extensive editing and processing to be accomplished with a very high degree of precision. However, when you export your edit file you can select exactly what format you want the finished file to be in.

I have a WAV file that is 23.4MB in size and 1h 41min long

What is the rest of the format? A 1h 41min stereo WAV file at CD quality (44100 Hz 16 bit) would be approximately 1 GB.

[susanne]

Wow, thanks for the detailed explanation!

Well, I do have the source file in its original format (WAV). It's a recording I made with my mp3-player which has a built-in microphone. Because of it's recording quality is pretty bad, I just got myself a new Olympus dictaphone.

I have a WAV file that is 23.4MB in size and 1h 41min long

What is the rest of the format? A 1h 41min stereo WAV file at CD quality (44100 Hz 16 bit) would be approximately 1 GB.

It's a mono WAV file at 8000Hz, I think. The sound quality is awful, but it was the best I could do at the time.

But can't Audacity auto-detect the format and just split it in two files without asking which format to save to? I mean like mp3splt does in Ubuntu? Just splits the file in two, while keeping everything the same.

[steve]

The beauty of MP3Spit is that it does not decode the MP3 so it does not need to "remember" what the original format was - it just splits the original file and creates the necessary file headers so that the output file(s) work. This is a completely different way of working from Audacity. In Audacity you can record and import multiple tracks and once the audio is in the project Audacity does not care about where the data originally came from. So for example you could record a track from a 16 bit sound card, import an MP3 and import a 32 bit WAV file into the same project, then mix and edit them as you wish. When finished you could export as a 16 bit WAV file for burning to CD, or as a 128 kbps MP3 to e-mail to your auntie Mary, or as a 250 kbps mp3 for your iPod (or all three, or any other supported format). In this example Audacity would have no way of knowing what format you wanted the final exported file to be.

It's a mono WAV file at 8000Hz, I think.

Depending on what you want to do with the file you can select the sample rate that you want the final exported file to be.
If you want your exported file to have a sample rate of 8000 Hz, set the "Project Rate" (see the bottom left corner of the main Audacity window) to 8000 before you export.
Any sound quality loss due to converting to a higher sample rate and then back down to the original sample rate should be negligible. 8000 Hz is a very low quality setting and will restrict the frequency response to below 4 kHz, so it's impossible to have sound in this format that does not sound rather muffled (telephone quality at best).

[ignatz]

"Slice" from NCH Software seems to do what you want. I just downloaded it and experimented.

I use NCH's "Switch" to convert files to various formats, so I have some faith in their products, but use at your own risk, etc.

Original file 2'22" in length, 24.0 MB, WAV.

I loaded it into Slice after setting the WAV output preferences to PCM uncompressed, 44100 HZ, 16 bits, stereo.

By default, it was split into 2 WAV files, each 1'11" and each 12.0 MB.

Took a split second to split the file.

Does that mean there was no transcoding?? I don't know.

But you can customize the length of each split portion.

It's free from here:

http://www.nch.com.au/splitter/index.html