The 64 Million Dollar Question...What Am I Hearing?

[mark4man]

Crew...

[Sorry to post another topic again so soon...but this one is related to the previous one...& this one may provide the key to solving an age old problem for me.]

As discussed w/ various forum members...SONAR converts 16 & 24-Bit files to 32 FP the instant DSP comes into play.

First...the background:
[8 out of the 9 projects on my upcoming CD are plain ol' 16-Bit (don't ask...it was the lead guitarist's fault.) Many of the tracks have UAD-1 & native Cakewalk plug-ins applied, as well as automation envelopes & track level adjustments...i.e., mucho DSP.

I used to be amazed at how great they sounded played back out of SONAR (back when I used to only have vXL 2.2.) The band members would come to the studio, listen to the 16-Bit projects playing in real-time; & say: "F_ck_ng WOW".

Then, I began to get ready for mastering; & started mixing down track groups & comp sets to 24-Bit stems (to then take to the mastering house, which I did.) I did this thinking that; if the projects sounded that good as 16-Bit...they're going to sound even better in 24...''cause 24 will capture more plug-in ambience, thereby increasing the dynamic range in that respect & making them sound "rich".

But the 24-Bit stems were always lacking a certain "presence"...they never sounded as bright or clear as 16-Bit project playback.

The finished masters came back "muddy".

So...I created 9 24-Bit master projects, in which I loaded in the 24-Bit stems from each composition; & have spent a great deal of time since then adding EQ here & there to brighten them up...trying essentially to make them sound like they used to as played back in the 16-Bit projects.]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Now...what was I hearing (in the original 16-Bit projects)?

SONAR converts 16 to 32 (or 64)...we know that. The processing occurs at that level of internal precision...we know that.

The output signal then has to be pipelined to the audio interface...to the D/A.

Since the projects are 16-Bit...is it a 16-Bit signal that is pipelined to the AI?

Or...my Audio Driver Bit Depth is set to 24...to accommodate the full capability of the audio interface (& I believe this occurs automatically w/ ASIO anyway.) So...is the output signal pipelined to 24-Bits?

Out of a 16-Bit project...what am I hearing...16 or 24? This is the question.

'Cause...if it's 16? I have an experiment ahead of me. If it's 16...I've been hearing truncation; & liking it (unless SONAR rounds or does something else when bit reducing to a fixed-integer output...& I've had no dither option checked, ever.) So if it's 16...I'm going to try 16-Bit mixdowns, to see how they sound.

Any thoughts or answers?

Thanks,

mark4man

 

[Middleman]

SONAR converts 16 & 24-Bit files to 32 FP the instant DSP comes into play.

If the original imported files are 16 or 24 yes. But if you are tracking from scratch it can be as high as 24 (limited by your soundcard drivers).

But the 24-Bit stems were always lacking a certain "presence"...they never sounded as bright or clear as 16-Bit project playback.

The finished masters came back "muddy".

Probably because you upsampled to 24 bit or you chose less than a desirable dither option during the upsample.

Now...what was I hearing (in the original 16-Bit projects)?

Un-dithered sound prior to upsampling which resulted in the dulling. Or, you were hearing the 32 bit internal operations of Sonar prior to creating the stems.

SONAR converts 16 to 32 (or 64)...we know that. The processing occurs at that level of internal precision...we know that.

It doesn't convert i.e. upsample on import, at least I don't think so. It operates at 32 bit internal processing and imports 16 or 24 bit files in that environment. Someone correct me if this is wrong.

The output signal then has to be pipelined to the audio interface...to the D/A.

Here dithering would apply which is why many people recommend turning it off. When you are ready to send it to mastering send it out 24 bit and let them take it to 16.

Since the projects are 16-Bit...is it a 16-Bit signal that is pipelined to the AI?

Depends on the setting in the Audio menu. Is it 16 or 24?

Or...my Audio Driver Bit Depth is set to 24...to accommodate the full capability of the audio interface (& I believe this occurs automatically w/ ASIO anyway.) So...is the output signal pipelined to 24-Bits?

Yes but dithering may be applied to what you are hearing.

Out of a 16-Bit project...what am I hearing...16 or 24? This is the question.

Once again, it depends on your audio setting and the dithering setting.

'Cause...if it's 16? I have an experiment ahead of me. If it's 16...I've been hearing truncation; & liking it (unless SONAR rounds or does something else when bit reducing to a fixed-integer output...& I've had no dither option checked, ever.) So if it's 16...I'm going to try 16-Bit mixdowns, to see how they sound.

16 works in a lot of instances if you have really good convertors. But working at the highest level in Sonar i.e. 32 without dither until you are ready to create your stems will pass along the most information from the original tracks.

If you start out with 16 however, its not going to get any better with upsampling. It can be different but usually you can't get back any more than what you started out with.

Hope that was clear.

 

[mark4man]

middleman...

Thanks very much for your answers; & please understand...there was no dither. Dither was not enabled in SONAR2, whatsoever.

Does that change anything in your responses?

Thanks,

mark4man

 

[Middleman]

Not really. You are still working with upsampling which is a compromise. You can't make things better ITB at a higher resolution if you start out with a lower resolution.

 

[DM1]

Upsampling happens when you increase the sample rate (i.e. going from 44.1Khz to 96Khz.)

When you convert from 16- to 24-bit, you're changing the bit depth, and it's a non-destructive operation. (Sonar just pads the 16-bit words with 8 bits of 0)

SONAR converts 16 & 24-Bit files to 32 FP the instant DSP comes into play

Sonar does everything at 32 (or 64) bits internally, regardless of your project's bit depth, and the bit depth at which you record. Whether or not you have plugins (which I assume is what you mean by DSP) is irrelevant.

But the 24-Bit stems were always lacking a certain "presence"...they never sounded as bright or clear as 16-Bit project playback.

There's no inherent reason this would be the case. From Sonar's perspective, the audio data in your 24-bit files is EXACLTY the same as the 16-bit files from which they were created. (The 8 extra bits of 0 don't carry information.) Did you change anything in the listening environment? Did you add any plugins to the 24-bit stems that weren't present on the original 16-bit mixes?

Do you still have the original 16-bit files to do an A/B comparison with?

Out of a 16-Bit project...what am I hearing...16 or 24? This is the question.

As middleman said, it depends on both your dithering settings and your sound card driver settings.

In your case, you said dither was not enabled in SONAR2, so your sound card driver is the determining factor. If you're running your ASIO driver in 24-bit mode, then Sonar is sending 24-bit data.

I didn't read the other question you referenced, but this post implies that you're no longer using SONAR2. Have you checked the dither setting in whatever version you're using now?

 

[mark4man]

DM1...

Thanks.

I didn't read the other question you referenced, but this post implies that you're no longer using SONAR2. Have you checked the dither setting in whatever version you're using now?

Driver is the same...ASIO in 24-Bit mode. I only use dither (TPDF) when I bounce to a stereo master. For individual stems, they're dither free.

m4