Re[2]: 16-Bit Stereo PCM Standard?

Hello Patrick,

Monday, April 5, 2010, 7:02:24 PM, you wrote:


> Ah that is indeed a good way of putting it. But the confusion however
> still stands. I might not have stated it clearly, but I'm curious
> about three things:

> 1) What is the _scale_ by which the [-32768, 32767] get converted to
>   sound. E.g. as you said, using 32767 would mean the speakers cone
>   pushes out as far as it can go, and 0 means it's not pushed out at
>   all, but then does the value (32767 / 2) mean that it's pushed out
>   half way? That would be a linear scale. Or is the scale something
>   else?

The scale is really just whether the DAC on the soundcard puts out it's full -V voltage, or full +V voltage, or somewhere between. The actual dB is determined by the analog stage following (buffer amplifier etc). Typically, a sound card puts out a "line level" signal of 2V RMS p-p. If you drive speakers, there is a power amplifier in use which applies gain. The decibel level resulting depends on the gain of this amplifier and also the sensitivity of the speakers.

> 2) It is slowly making sense to me that there can't be a requirement
>   that says anything along the lines of "PCM value 16,383 yields
>   decibel value 120db"? The user can always change the speaker volume
>   and foil your attempts to produce any particular decibel
>   values. However, this leaves me baffled as to what part of the
>   [-32768, 32767] range a sound file ought to use. Should every sound
>   file use all of the range? This leads into question #3:

Ideally, for maximum dynamic range and greatest SNR, it should make full use of this range. Some headroom is normally desirable in the case where the sound source may be mixed or have other effects applied to it prior to output, but this is not strictly necessary as the source can be attenuated before this happens.

> 3) How does sound get captured from a microphone? Do you get 16-bit
>   PCM input as well? But if you do get something like 16-bit PCM data
>   from a microphone then there _must_ be some mapping going on from
>   "decibels -> 16-bit PCM", since the microphone really can be
>   guaranteed to capture exact decibel levels (even if the speakers
>   aren't guaranteed to output exact decibel levels). Additionally,
>   is there some maximum decibel level that a mic can usually capture?
>   Might that get mapped to something like the 32767 PCM value?


This is the reverse process - the analog input is usually attenuated or scaled (adjustable by the Mic In volume control) to within the ADC's acceptable input range. There, it is sampled, and the resultant values are relative to the ADC's range.


-- 
Best regards,
 Peter                            mailto:[email protected]

_______________________________________________
Openal mailing list
[email protected]
http://opensource.creative.com/mailman/listinfo/openal