When I export as WAV - how is the downsample done?

S

sjfoote

New member
If I record a track in CakeWalk at 24-bit, 48kHz - when exported as WAV, how does it convert to standard 16-bit, 44.1kHz file?

Is there a process within CakeWalk that does downsampling to the CD standard? If so, is it just an algorithm that does the conversion, or is there a piece of hardware involved also?
 
There may be some selections you can make in the "Save" window. Or perhaps there's a "Save As" or "Export" menu item that will have the options. Usually, no hardware is necessary.

However, if your destination file needs to be at 44.1, I'd suggest recording at that sample rate and avoid the downsampling.
 
If you save a 48kHz file as a 44.1kHz wav without sample rate conversion, and your program doesn't automatically do it for you, you will change the pitch of the tune. Even if your program does a conversion, the built-in routine might not be the best quality. However there are software algorithms, usually in the form of plugs, that will do a respectable job.

I would try it and see what happens . . .
 
Maybe I didn't pose the question correctly. When I export a track as WAV (single track or mix-down track with all edits/effects) that was recorded as 24 bit, 48k - the file comes out as 16 bit, 44.1k because that is how the software is setup for WAV export. Those parameters can be changed in the software, but since the final destination for these files is a standard audio CD, that is how I have the software configured. I believe that the file must be in this format before burning to audio CD - I hope I have that right.

Now that being the case, I assume then that the software must do some type of conversion/downsampling to change from 24/48 to 16/44.1. Is this the basic procedure for DAW software?

I'm trying to determine if the soundcards AD/DA converters are part of this process, or is it an algorithm that does the conversion. In other words, does my soundcard figure into this conversion?
 
The dithering and sample rate convert are all done in the software. You don't need anything connected to your computer to export your wav files at whatever sample rate/bit depth you want.
 
Thanks to all who helped with this question.

I'm not sure what I did wrong, but someone must have given me bad rep for asking this question. I was only attempting to confirm that the quality of the soundcard was not a factor in this process.

Thanks again...
 
RAK said:
The dithering and sample rate convert are all done in the software. You don't need anything connected to your computer to export your wav files at whatever sample rate/bit depth you want.
Click to expand...
When I change sample rates down to 16/44.1 with Cubase, I have to apply a dithering plug. Does cakewalk do this automatically?
 
NYMorningstar said:
When I change sample rates down to 16/44.1 with Cubase, I have to apply a dithering plug. Does cakewalk do this automatically?
Click to expand...

Essentially yes. There is an options tab where you set what type of dithering you want to use. But once you set that, it does it automatically when you export. Of course you need to set what sample rate/bit depth you want to export at, but you don't need to manually dither everytime. At least that's how it works in Sonar 3 PE and 4 PE.
 
sjfoote said:
Thanks to all who helped with this question.

I'm not sure what I did wrong, but someone must have given me bad rep for asking this question. I was only attempting to confirm that the quality of the soundcard was not a factor in this process.

Thanks again...
Click to expand...

Beats me, here....... have a greenie. :)

What I do is keep the mix file at 24 bit, and then I have burn files that I dither to 16 bit. The reason is that any mastering you decide to do later will benefit from maintaining the bit resolution. Same goes for the downsampling.
 
sjfoote said:
Maybe I didn't pose the question correctly. When I export a track as WAV (single track or mix-down track with all edits/effects) that was recorded as 24 bit, 48k - the file comes out as 16 bit, 44.1k because that is how the software is setup for WAV export. Those parameters can be changed in the software, but since the final destination for these files is a standard audio CD, that is how I have the software configured. I believe that the file must be in this format before burning to audio CD - I hope I have that right.

Now that being the case, I assume then that the software must do some type of conversion/downsampling to change from 24/48 to 16/44.1. Is this the basic procedure for DAW software?

I'm trying to determine if the soundcards AD/DA converters are part of this process, or is it an algorithm that does the conversion. In other words, does my soundcard figure into this conversion?
Click to expand...

You are 100% correct.

As said before, it's all software....sound card plays no part in dithering. You're doing it right. :)
 
Thanks All. I will continue to export this way and one day soon, maybe I'll have enough new tracks for a CD. :)
 
Just a question...

Don't you think software would do a better job at SRC/dithering since it's just an algorithm and not hardware/clock based?

You would think it would be more accurate.
 
danny.guitar said:
Don't you think software would do a better job at SRC/dithering since it's just an algorithm and not hardware/clock based?
Click to expand...
Ah, but the software *is* hardware/clock-based. It's running on a hardware computer and using the motherbboard clock.

The only real difference is that the instructions are held on external disc storage and loaded into RAM and executed by a general purpose central processor instead of loaded on a ROM chip and executed (maybe) by a dedicated processor and dedicated clock. In fact, one could argue in that light that the dedicated hardware would be "better". Certainly the clock probably would be.

But it really comes down to the software instructions themselves more than anything else. Whether they are burned into a ROM or loaded into RAM from a disc does not matter, it's still software, and it's qualty depends upon the quality of programming and upon the algorithm that the programming is executing. You could put lousy software in a dedicated box and put great software on a personal computer, and vice versa.

Now, on a more practical sense, the dithering and/or SRC algorithms that come built into most audio editors are not for the most part going to be A-list any more than the tires or CD player that come with your average car are going to be the best you can get. But that's for marketing and economic reasons only, not because of anything having to do with potential hardware or software capabilities.

G.
 
I don't see how the algorithm would use the hardware at all.

All the numbers (amplitude/frequencies) are stored in the WAV file itself. It's just a matter of looping through the WAV file and performing the necessary math on the numbers.

In fact, I'm pretty sure you can edit WAV files (dithering/downsampling) without a soundcard installed on your computer as long as you have the codecs.

Maybe I'm wrong, but from my brief experience in programming, I don't see how the algorithm would need any kind of clock/hardware whatsoever to perform math functions on a WAV file.
 
danny.guitar said:
I don't see how the algorithm would use the hardware at all.

All the numbers (amplitude/frequencies) are stored in the WAV file itself. It's just a matter of looping through the WAV file and performing the necessary math on the numbers.

In fact, I'm pretty sure you can edit WAV files (dithering/downsampling) without a soundcard installed on your computer as long as you have the codecs.

Maybe I'm wrong, but from my brief experience in programming, I don't see how the algorithm would need any kind of clock/hardware whatsoever to perform math functions on a WAV file.
Click to expand...
Does it use the soundcard? No.

But when you're talking about digital processing, it's always a combination of software and hardware, whether you're using a dedicated curcuit card or a general purpose personal computer.

When you run a SRC either included with audio editing software or in the form of plug-in software, you are basically temporarily turning the computer into hardware "dedicated" to the purpose of sample rate conversion. Without your CPU, the software could do nothing, and without the system clock, the CPU could do nothing. (And, FTM, without the software, the CPU and clock would be useless as well :) )

A dedicated digital signal processor (e.g. a digital reverb, a UAD card, etc.) is nothing more than a dedicated digital computer with it's own software either built into the firmware (e.g. ROM chip) or partially loaded off of the PCI data buss. The only difference is that the processor chip(s) in the dedicated circuitry are usually not quite as powerful and not wired to use any other software than the stuff it's designed to - it's not a general-purpose computer.

As such, both are just manipulating the ones and zeros of the digital waveform information based upon instructions designed and written in software and executed by digital hardware. On that level there is no difference. The main differences are in the quality of the software itself and possibly in the interface between the hardware/software and the outside world.

G.
 
Thanks for the explanation Glen.

It would seem that an entry-level program like CakeWalk's Music Creator 3, would probably not have the same quality SRC capabilities as would their top of the line SONAR program. Do you think this is correct?

If this is correct, then perhaps I would be better served (since I use MC3) to record at 44.1 and not introduce any artifacts from a rate conversion.

That only leaves the 24 versus 16 bit question - again on an entry level program - what happens when the file is saved as standard 16 bit WAV? Are the remaining bits discarded?
 
sjfoote said:
That only leaves the 24 versus 16 bit question - again on an entry level program - what happens when the file is saved as standard 16 bit WAV? Are the remaining bits discarded?
Click to expand...

Are the converters on your sound card/interface 24-bit? If not, then I would record in 16. If you have enough CPU/disk space to do 24-bit then use that.

When converting down to 16 from 24, you usually dither the WAV file. Do a search and you're bound to find (a bunch) of info on dithering.
 
sjfoote said:
It would seem that an entry-level program like CakeWalk's Music Creator 3, would probably not have the same quality SRC capabilities as would their top of the line SONAR program. Do you think this is correct?
Click to expand...
Not having used the SRCs included with either one, I could not comment on which one is better.

It really all depends upon where the company gets their SRC algorithm from, who writes their software, and what they decide to include with any given version of the software.

For example, in your question there is the assumption that because Cakewalk's MC3 is entry level that it's SRC is not as good as the one in Sonar, which is Cakewalk's flagship software. That could very well be true. It could also very well be true that the Cakewalk development team has only bothered to write a single SRC software module that they use in all their software; that the SRC used in MC3 and the one used in Sonar are one and the same.

This kind of re-use of software is commonplace, especially when it deals with features a company is not marketing as different or improved. To my knowledge (correct me if I'm wrong), Cakewalk does not advertise Sonar as actually *sounding* better than GTPro or MC3, only that it has more or better features and capabilities. Huge examples from other companies include Steinberg, who uses the exact same audio engine for both CubaseSX 3 and Nuendo 4, and Sony, who's Acid Pro and Vegas software are practically conjoined twins in the WAV audio and interface departments.

I would tend to think that unless they make a point of including an improved-sounding converter (or at least "improved sound" in general somehow) in the feature list for Sonar, that they probably just use the same SRC in all their stuff. It would be wasteful to dedicate development time to seperate SRCs and not try to make their development costs back by advertising the difference.

That is only an educated guess on my part, though. I really don't know if they use the same SRC or not, or if one is actually better sounding than the other or not. Maybe they *did* develop more than one SRC at Cakewalk and they quietly gave the better one to Sonar. I don't know.

But hopefully this explanation can illustrate how there is no guarantee on the individual feature level just what software one is getting behind it based simply upon product positioning. Sometimes the spare tire and jack supplied with a Chevy Impala is not better or even different than the one supplied with a Chevy Malibu.

G.
 
There's a lot of information in this thread but I think some of it is not quite right.

As I trust we all know, digital audio is a stream of samples taken at set intervals.

The sample depth - we're talking here about 16 bits and 24 bits - refers to how much information each sample contains. The higher the number, the greater the possible dynamic range, i.e., the sample is capable of representing "quieter" information. Conversion from a higher sample depth to a lower sample depth is a matter of loading the sample into a memory location and masking off (throwing away) the extra bits. This is very simple and can be done in software or hardware with NO difference in results. Also, different sample depths do not result in a change in file size.

The sample rate refers to how many samples there are over a given period of time. 44.1 khz means there are 44,100 samples per second. A conversion in sample rates is more complex than one in sample depth. To convert to a lower sample rate, the process needs to throw away entire samples; to go from 48 k to 44.1 k, 3900 samples need to be discarded each second.

In the simplest of rate down-conversion algorithms - no dithering - you simply determine an interval and make a counter which counts samples. When the counter reaches the interval, you do not write that sample out. This is rate conversion without dithering. Conversion in sample rates DOES result in a change in file size.

Dithering in essence involves looking at the discarded sample AND the samples around it, and altering the samples you are keeping to provide a smooth transition to make up for the lost information. It is a form of averaging, and the quality of the result depends on the algorithm used, i.e., how many samples do you alter, and how?

Any math and logic can be implemented in hardware OR software. The earliest digital computers were programmed by changing the wiring. As Glen implies, a conversion algorithm in hardware is, these days, likely actually to be "firmware," a computer program burned onto a ROM chip.

Obviously, the ultimate hardware solution is to have perfect A/D and D/A converters and resample the analog information!
 
You must log in or register to reply here.
Back
Top