Kyma Forum
  Kyma Support
  Numbers and Clipping

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Numbers and Clipping
tuscland
Member
posted 09 January 2006 11:02         Edit/Delete Message   Reply w/Quote
Hello,

I was wondering under which condition the results of computation were "clipped".

What I mean is if I put a sine wave with enveloppe of 1.0 into a Attenuator with a gain of +10 dB and then a Attenuator with a gain -10 dB, will I have my sine wave back or a distorted one? I tried the experience and found that I get the distorted one.

Let me put this another way : I would like to know if the internal computations are done in 24 bits or a higher resolution (56 bits accumulator)?

I guess the rule is : every number used in the DSP computation have their absolute value between 0 and 1.
Is this true?
Are there any exceptions?

I am a bit confused about the range of values are handled in Kyma, and I want to make sure I fully understand the issue.

Best wishes,
Camille

IP: Logged

SSC
Administrator
posted 09 January 2006 14:26         Edit/Delete Message   Reply w/Quote
"What I mean is if I put a sine wave with enveloppe of 1.0 into a Attenuator with a gain of +10 dB and then a Attenuator with a gain -10 dB, will I have my sine wave back or a distorted one? I tried the experience and found that I get the distorted one."

Yes, a full amplitude sine wave is using all the bits of the sample word. When you put that through a Gain, the waveform is clipped. That information is lost so when you attenuate it again, it still has a square top.

"Let me put this another way : I would like to know if the internal computations are done in 24 bits or a higher resolution (56 bits accumulator)?"

There is a 56-bit accumulator.

"I guess the rule is : every number used in the DSP computation have their absolute value between 0 and 1."

Sound outputs are always in the range (-1, 1).

(The generated value of a SoundToGlobalController can be *any* any number. However, this is true only of the hot value that it generates as a "side-effect". The actual audio output of the SoundToGlobalController when Silent is unchecked will be clipped to the range (-1,1).

IP: Logged

tuscland
Member
posted 09 January 2006 14:44         Edit/Delete Message   Reply w/Quote
Does the 56 bit accumulator "accumulates" results all along the prototypes chain, or is it internally bounded to a single prototype?

I would like to know if I can assume that the dynamic range is 56 bits from the input to output of a Sound. If yes, that leaves plenty of headroom with a dynamic of 337 dB !!!

Apart from that, thanks, this all is very clear now.

Cam

IP: Logged

SSC
Administrator
posted 10 January 2006 08:49         Edit/Delete Message   Reply w/Quote
There is a 24-bit signal path between modules (144 dB).

IP: Logged

tuscland
Member
posted 10 January 2006 08:59         Edit/Delete Message   Reply w/Quote
Right, that makes perfectly sense.
That's good because there's still around 50 dB of headroom compared to 16 bit signals.
I asked these questions because I read the explanation of the Kyma compressor and saw that the signal was attenuated of 10 dB. I wanted to know what 10 dB were meaning compared to the signal/ratio of the signal path. 10 dB in the 16 bits field mean a lot, but in the 24 bits field is fine.

just my $0.02

IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c