A bitmeter

Grok · Post by **Grok** » Sat Aug 27, 2005 8:16 am

... to check the audio signals bit resolutions within Scope 4...

...It would be instructive, and somehow useful

Shayne White · Post by **Shayne White** » Sat Aug 27, 2005 11:40 am

Scope always operates internally at 32-bit integer resolution. The ASIO/Wave source modules will tell you what the incoming bit depth is for those modules, but I don't know about ADAT or S/PDIF.

Shayne

Grok · Post by **Grok** » Mon Aug 29, 2005 3:33 pm

Thanks for your answer, Shayne.

The difference between the digital audio signal format and the bit resolution of this signal has to be understood.

e.g. you can bitcrush a 32 bit float or 32 bit integer audio signal to say 4 bit; it will remains a 32 bit signal but its resolution will be 4 bit.

A 32 bit signal can have a 24 or 16 bit resolution... It depends.

When I record a Scope synth or a Scope FX in a 32 bit float audio file, the Samplitude's Bitmeter shows me that this 32 bit float audio file has a 30 bit linear resolution. If I record a 24 bit converted A>>>D signal in a 32 bit float file, the resolution of this 32 bit float file is 24 bit linear. If I process this file with a 32 bit float audio engine (say with a "freeze"), the resolution of the freezed resulting file jumps to 32 bit float.

Regards,
Grok

PS: I would like to know for sure what really are the Scope processes; the Creamware's site says the Scope processing is 32 bit float; some others say it is 40 bit integer; you say it is 32 bit integer; I measure in Samplitude the recorded Scope signals to 30 bit linear; I'd really like someone involved to clarify all this. A bitmeter usable inside Scope 4 could be instructive !...

<font size=-1>[ This Message was edited by: Grok on 2005-08-29 16:39 ]</font>

spiderman · Post by **spiderman** » Tue Aug 30, 2005 7:35 am

sometime samplitude give me 32bit (non float !!! ) when I record mixdown through asio etc ..

the bit meter in samplitude has the red float led and some time this led don't shine ! very strange ....

spiderman · Post by **spiderman** » Tue Aug 30, 2005 7:41 am

I pretty sure sfp works in 32 bit integer ; but only someone from creamware could elucidate this as some plug ins like vintage eq works at 64bit !! maybe it's some virtual 64 bit something .

more often when I record a mixdown through sfp , I record it in 32 bit bloat in samplitude . most of the time ( 90% of the time ) the bit meter say 32bit float . and some time it says 32bit integer

astroman · Post by **astroman** » Wed Aug 31, 2005 3:39 am

On 2005-08-30 08:41, spiderman wrote:
..., I record it in 32 bit bloat in samplitude ...

nice typo that really says it all

8 of your precious 32 bits will be arbitrarily mixed up by any membrane of even the world's best speaker cone...

why do you assume that the program is precise in this domain ? it's as suspective to any rounding error as any math proc.

as long as you don't know the details of all processing routines a given bit depth is of few relevance (if at all).
there are tons of ways to do the 'same' calculation

according to specs Sharc DSPs can do anything from int to float and mixed.
Imho the Scope boards use a 32 bit integer datastream for internal communication (as a kind of 'common' format), probably in the way that (if required) bus data is converted from int to float and vice versa frequently.

such kind of conversion has no audible influence at all, even if it is done 100 times along the signal path.

what DOES have an influence is if data is processed in a loop, where the same value is mangled over and over again. And this is a fairly common programming structure.
Any regular routine has dozens of them, possibly with thousands of passes through an inner loop.
That's exactly the location where your signal looses precision - at the very core of the algorithm.

as long as you don't know those (respective) details, you don't need to worry about bit depth - listen

cheers, Tom

wayne · Post by **wayne** » Wed Aug 31, 2005 10:55 am

On 2005-08-31 04:39, astroman wrote:

nice typo that really says it all

Actually , i had a 32-bit bloat the other night, couldn't get through it - i was stuffed after the hors-d'oeuvres

astroman · Post by **astroman** » Wed Aug 31, 2005 11:40 am

tacky tuba tone

spiderman · Post by **spiderman** » Thu Sep 01, 2005 1:14 am

8^)))

I don't worry at all ! I just ask .
i thrust my ears . they are large enough ..

<font size=-1>[ This Message was edited by: spiderman on 2005-09-01 02:17 ]</font>

wayne · Post by **wayne** » Thu Sep 01, 2005 2:34 am

at0m · Post by **at0m** » Thu Sep 01, 2005 9:05 am

Spiderman, it's indeed possible to apply some tricks for higher bitdepth, like STW's Vintage EQ series internal processes are at 64bit.

The DSP chips may run on a higher bitdepth, or the calculations may be, but the audio SFP works with is 32bit linear.