Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

32 bit sound

Options
  • 15-09-2005 6:47pm
    #1
    Registered Users Posts: 95 ✭✭


    I'm developing a prototype sound application that requires a dynamic range (amplitude) of 120db or greater. The application uses direct sound to interact with sound card and is working as expected. The problem is with the sound card!

    As far as i can see the best sound cards available only support a 24 bit dynamic range (i.e. 24 bit cards) while this offers a theoretical dynamic range of 144dB (its 6 dB per bit!) it realistically only gives an output of 108dB when you take into account signal to noise ratio.

    Therefore i require a 32 bit card minimum as even with the signal to noise ratio it should be able to deliver over 120dB, particularly if the card supports floating point data. Has anyone had any experience of this? and also is it possible to get a 32 bit card (Preferably external!)?

    Thanks


Comments

  • Closed Accounts Posts: 4,943 ✭✭✭Mutant_Fruit


    have you tried using cards with optical in's and out's to reduce signal-noise ratio? Isn't most of the noise when you convert to analog, and optical is digital?


  • Registered Users Posts: 4,003 ✭✭✭rsynnott


    Mr.StRiPe wrote:
    I'm developing a prototype sound application that requires a dynamic range (amplitude) of 120db or greater. The application uses direct sound to interact with sound card and is working as expected. The problem is with the sound card!

    That's ridiculously loud, and not suitable for human consumption.


  • Registered Users Posts: 95 ✭✭Mr.StRiPe


    I haven't tried any sound cards yet apart from the 16 bit card I have in my pc which I’ve pushed to the max! Was hoping to do a bit or research (get advise!) on it before actually buying a new sound card.

    As far as I can see it’s the transistors that add the noise therefore the more bits the card has the more transistors and the signal-noise ratio also increases.

    The threshold of human hearing is 120 dB and this application is required to test it!


  • Registered Users Posts: 455 ✭✭Ferror


    Have you tried using a preamp? similar to what would be used with a mic (low output) before it reaches an amplifier...

    Should boost the signal power signifactly.


  • Closed Accounts Posts: 4,943 ✭✭✭Mutant_Fruit


    Some sources quote 120 dB as the pain threshold and define the audible sound frequency range as ending at about 20,000 Hz were the threshold of hearing and the threshold of pain meet.
    Is that what you want to be testing?


  • Advertisement
  • Registered Users Posts: 131 ✭✭theexis


    Mr.StRiPe wrote:
    I haven't tried any sound cards yet apart from the 16 bit card I have in my pc which I’ve pushed to the max!

    What does that mean exactly? All you're getting with higher bit resolution is greater fidelity in the amplitude steps you can represent - are you detecting the sensitivity of a persons hearing or something that requires extremely accurate amplitude recording?


  • Closed Accounts Posts: 703 ✭✭✭SolarNexus


    The Creative Audigy 4 Pro is a 32bit soundcard ( link )
    Audigy 4 Pro Processor
    • 32-bit digital processing, which maintains a theoretical 192 dB of dynamic range


  • Registered Users Posts: 131 ✭✭theexis


    SolarNexus wrote:
    The Creative Audigy 4 Pro is a 32bit soundcard ( link )

    32 bit processing, but its 24bit AD/DA.


  • Closed Accounts Posts: 703 ✭✭✭SolarNexus


    theexis wrote:
    32 bit processing, but its 24bit AD/DA.
    I wasnt going to mention this, but afaik windows sound only accomadates 24bit, I could check the platform SDK again, but if its true your either looking at a very expensive peice of kit or something that simply doesnt exist.


  • Registered Users Posts: 131 ✭✭theexis


    The DirectSound channel via KMixer has always worked with IEEE 32 bit floating point as its native format (hence most sound cards are "32 bit processing then dither down to 24bit during AD/DA) and WinXP SP1 supports up to 200Khz sample rate via the same path.

    Whats not clear is why 24bit accuracy is not sufficient to capture / represent the steps in DbU over the range: I'm intersted to hear more about the background to the project!


  • Advertisement
  • Registered Users Posts: 95 ✭✭Mr.StRiPe


    Originally Posted by theexis
    What does that mean exactly? All you're getting with higher bit resolution is greater fidelity in the amplitude steps you can represent - are you detecting the sensitivity of a persons hearing or something that requires extremely accurate amplitude recording?

    Its a digital Audiometer that will be used to test human hearing (can't say to much more about it!) hence the need for accuracy and a maximum amplitude of 120 dB
    Originally Posted by theexis
    ...most sound cards are 32 bit processing then dither down to 24bit during AD/DA.

    This is the problem and I’ve yet to see a card that has 32 bit AD/DA. The sampling rate is not a problem as the highest frequency it has to generate is 8K Hz therefore anything higher than 16K is more than enough. It’s presently running at 46k.
    Originally Posted by theexis
    Whats not clear is why 24bit accuracy is not sufficient to capture / represent the steps in DbU over the range
    As far as i can see the best sound cards available only support a 24 bit dynamic range (i.e. 24 bit cards) while this offers a theoretical dynamic range of 144dB (its 6 dB per bit!) it realistically only gives an output of 108dB when you take into account signal to noise ratio.


  • Closed Accounts Posts: 4,943 ✭✭✭Mutant_Fruit


    Maybe i'm not quite understanding what you want to do... so tell me if i'm wrong.

    But isn't the amplitude irrelevant if you're testing human hearing? Amplitude just determines how loud the sound is. Obviously then if you go to 120dB it's going to be quite quite loud. If i remember correctly from my physics book, isn't 120dB something like hearing a plane take off from about 30 metres away?

    Shouldn't the only important factor be the frequency? Human hearing goes up to about 20k hz i thought, so isn't that what should be being tested?


  • Registered Users Posts: 95 ✭✭Mr.StRiPe


    Your average nite club would be around 100dB!

    Amplitude is the key part! It is true you can only hear frequencies between 20 - 20k Hz

    If you have perfect hearing you will hear these frequencies with an amplitude between -10 to 0 dB

    But as you get older or suffer hearing damage you won't be able to hear these frequencies unless they are amplified to a higher level.

    Therefore to find the degree of damage in a hearing test you keep amplifying the different frequencies (by a small amount!) until the person can hear it.


  • Registered Users Posts: 131 ✭✭theexis


    I think you're out of luck for using a regular sound card - the highest dynamic range I've seen is 123dB on the MaXiO XD ($2000).

    However, since you're interested in a discrete range can't you just scale you system accordingly? e.g. if you have a soundcard that has a dynamic range of 100dB, set up your replay system (headphones I guess?) to amplify to 0dB; this way 1dB increases of the soundcard volume is 0.1dB in physical volume giving you a massive range. Thats probably an over simplification due to the logarithmic nature of the scale but I guess you can figure that bit out ;).


Advertisement