1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

The (Actual) Process of Recording/Wrtiting Audio Data ???

Discussion in 'Recording' started by mark4man, Oct 14, 2005.

  1. mark4man

    mark4man Active Member

    People,

    Was posting in a Hardware forum, inquiring about the benefits of RAID vs. SCSI vs. ATA drives for DAW's...& someone came back & asked: "don't most audio apps read as much of the file into ram as possible prior to the user working with it?"

    This got me thinking...what actually happens inside the box when the red button is on? What is the sequence & path of the data flow; & how is it implemented from the interface thru the multitrack recording software to eventually the hard disk? The data moves sequentially, in one continual stream, right? Does it detour thru RAM? How does the software code guide the signal?

    I don't believe I've ever seen this aspect of digital audio recording discussed in the forums...everybody's seemingly interested in gear & performance...& not too much on process. Are there any good white papers or informationals around? (or can someone put it into an outline)? Has anyone ever done an animated graphic on this?

    Thanks,

    mark4man
     
  2. Mr-Nice

    Mr-Nice Guest

    In a nutshell....

    When you record into a computer the signal passes from the input on your audio interface to the host applications input bus and sent to the tracks channel for monitoring.

    At the same time it writes the file to your HD when you record. I dont know if it passes through RAM (at least I dont think) during this process but it does pre-load into RAM during playback.

    Of course there are other factors that come into play such as bitrate, sample rate that determine the resolution, clarity and size of the recorded material. And can affect playback, because the higher the sample rate the more RAM and CPU power your computer is going to use to play it back.
     
  3. KyroJoe

    KyroJoe Guest

    mark4man,


    Signal Path simplified (and some extra info):

    (there are many! CPU interrupts etc, and instructions and interactions between hardware and software going on behind the scenes that are left out below for simplicity)



    From device (instrument or mic)

    -> Line In (analog - voltage! SINE Wave)

    -> A/D (analog to digital converter) MOST IMPORTANT for good quality sound!!

    -> Voltage is sampled by the A/D xx amount of times per second. This is your 44.1kHz, 48kHz, or 96kHz sampling rate.

    -> Analog voltage is PCM encoded here (Pulse Code Modulation) Digital data 0's & 1's as SQUARE wave are used to "represent" the original SINE Wave (a snapshot in time of the position on the sine wave). Amount of samples per second (sampling rate) and the BIT depth (8,16,20,24 - which affects the convertable amount of amplitude, dB pressure (volume) - 16bit has a range of 96dB, and 24bit has about 144dB, before digital clip) Sampling Rate and Bit Depth combined affect how close that "representation in digital" is to the original SINE wave input.

    -> Digital data is moved from sound card/interface A/D circuit into RAM Buffers (designated by audio software, drivers & OS's memory manager)

    -> Software determines when the buffers are full

    -> CPU instructs Hard Drive Controller to write out and empty the full RAM Buffers

    -> data is moved to Hard drive controller circuit

    -> data is written in FILE on Hard Drive


    Nyquist's theorem states that your audio sampling rate needs to be about 2.2 times the highest wanted audio frequency. To capture the entire 0Hz-20KHz frequency spectrum (audible sound) that means you need about a 44kHz sampling rate to accurately encode and represent the SINE wave.

    The quality of your stereo A-D converters are important.
    If you have a stereo A/D converter working at 44.1kHz with a 16bit output
    the total data generated every second is:

    2 (channels) x 44100 (samples) x 16 (bits) = 1,411,200 bits per second or 176,400 Kilobytes per second

    In just one minute there will be about 10.5Mb of PCM encoded audio data to store!
    If you jump to 48kHz sampling at 24-bit, an hour long recording would require just over 1 Gb of storage!



    RAID is a way of using one of the following interface types SCSI (scuzzy), IDE/ATA, Fiber Channel (optical), or SATA drives.
    RAID very simply means that you are using multiple drives in an array, a way that adds fault tolerance through parity storage or redundancy for disaster recovery and some speed increase by spreading data across multiple disks and reducing the workload on each individual hard drive unit.

    Drive speeds are affected by inteface type, controller throughput, Seek Times, and RPMs.
    Typically an IDE/ATA 5400 RPM drive is going to result in clicks, dropouts, and all kinds of problems with audio. They simply are not fast enough. 7200 RPM is the way to go. Over 10,000 RPM is awesome, but usually overkill depending on how many tracks you need to record simultaneously.







    KJ
    ----------------
    Kyro Studios
     

Share This Page