Sampling rates and processing speed

Discussion in 'Computer Science & Culture' started by kula, Aug 12, 2004.

Thread Status:
Not open for further replies.
  1. kula (Memes enclosed) within Registered Senior Member

    Messages:
    148
    Hi,

    Does anyone know the link between sampling rates of music programs and the processing speed of computers ?

    I am trying to record music, tuned to exactly 432hz (and assosciated harmonics), but cannot get accurate results once the sounds are saved as wave files. I have tried different sample rates, which do make a difference, but I would like to be as exact as possible. My guess is that the speed of the processor is out of sinc with the wavelengths of the sounds i am producing. (in a similar way that mp3's corrupt the original sound of .wav files) I would appreciate any comments or ideas.

    thanks

    kula
     
  2. Google AdSense Guest Advertisement



    to hide all adverts.
  3. firdroirich A friend of The Friends Registered Senior Member

    Messages:
    565
    What source are you using for the sound? A similiar frustration is "ripping" songs from vinyl to HD. The anologue\digital thing just makes the sound inaudible. A quick workaround to that problem was the introduction of digital outputs to decks. I never was aware of any other way. Maybe the problem with sampled sounds & processors is the interface between them?
     
  4. Google AdSense Guest Advertisement



    to hide all adverts.
  5. wesmorris Nerd Overlord - we(s):1 of N Valued Senior Member

    Messages:
    9,846
    i can only guess, but I don't see how it could be a processor speed problem except as a small fraction of a percent. most likely it's your sound hardware/software.
     
  6. Google AdSense Guest Advertisement



    to hide all adverts.
  7. kula (Memes enclosed) within Registered Senior Member

    Messages:
    148
    I can produce a pure tone (using wavelab sound editor to generate it) of 432hz and when saved to a .wav it has become 432.02hz, not a big difference, but big enough to make a difference to what i want to achieve. When I recorded 144hz (the note 'D' when 'A" is tuned to 432hz) as a pure tone, it appeared unaffected, but that may be the accuracy of wavelab to measure it.If i write a whole tune using 432hz, the more complex it is , the less accurate the outcome when it is saved. I have tried saving at low quality up to 48khz, but it still looses the accuracy.

    Thanks for replying

    kula
     
    Last edited: Aug 12, 2004
  8. wesmorris Nerd Overlord - we(s):1 of N Valued Senior Member

    Messages:
    9,846
    Sounds like software. If you go to an even lower frequency, I'd guess the error is still there? Yah I'd guess there is a "accuracy vs. frequency" thing, that sets a band where the card or software can create within a tolerance of freq.xx but as you leave the band on either end you go to freq.x tolerance. Just guessing though.
     
  9. Blindman Valued Senior Member

    Messages:
    1,425
    It your audio hardware that does the sampling. Check the max caps of your sound card.

    SB Live!Audio [E800]
    dwMaxSecondarySampleRate 191999

    This is for DirectX. It should be posible to sample up to this rate but I have not tried it yet, it may spit an error. THis is very high for audio but would improve your error.
    I assume it can play back at the same rate.
     
  10. kula (Memes enclosed) within Registered Senior Member

    Messages:
    148
    early morning maths

    Thanks for the advice, i'm going to have a fiddle. The faster the sampling rate I use, the more accuracy I get but as you say, it may be the accuracy of the sound generator in the first place.

    The lower the frequency I generate, the less distortion I get and my guess (I know little about computers, a bit more more about sound, so apologies if this analogy doesnt work) but, the reason i want to preserve an exact frequency of 432hz is to enable the wavepeaks of sound to be in sync with each other, even over a long period of time. With even a small disprepency, the wavepeaks gradually go out of phase. Lower frequecies are less effected so i was thinking that say the computer is sampling at 4321 times a second, and the frequency I want is 4320, that the small difference amounts to the computer sampling slightly behind each wavepeak and that over the period of a song, this adds up to being a larger difference.

    Ive only just woken up, so i hope this is co-herent, (I'm not !)

    wave peaks ^--^--^--^--^--^--^--^--^--^--^
    sample time s. s. s. s. s. s. s. s. s. s. s. s. s. s. s. s

    11 peaks, 16 'moments' of sample, but all the wavepeaks are not sampled/saved as they do not line up.

    Thanks again
    kula
     
    Last edited: Aug 12, 2004
Thread Status:
Not open for further replies.

Share This Page