[linux-audio-dev] Syncronizing Sample Clocks [WAS: A bit of goodnews--paper now available for your viewing pleasure and/or comments]

[RTaylor]

On 2004-05-28 19:49:08 +0000 Fred Gleason <fredg at salemradiolabs.com> wrote:

> On Friday 28 May 2004 15:19, Ivica Ico Bukvic wrote:
>> Hmm, it would be a fun project then to come up with a profiler of various
>> audio cards by recording and then capturing a specific buffer of audio
>> data. Then by comparing them (assuming that this drift is constant) see how
>> many empty samples there are (or if the playback is slower, how many
>> samples are missing), and then create a framework that allows real-time
>> resampling in order to compensate for that discrepancy whenever multiple
>> soundcards are being used :-D
> 
> I strongly suspect that you'd find your results to be non-repeatable.  Many 
> factors can subtly influence the output frequency of even crystal-locked 
> SRGs: ambient temperature, power supply voltage variation, even component 
> aging.
> 
> This issue affects many more applications than just audio.  *Any* system that 
> requires precise replication of clock (as, for example, most any digital 
> telecommunication scheme does) faces this dilemma.  In the end, some form 
> *locking*, slave clock to master, is needed.  A variety of methodologies, 
> such as PLL (phase lock loop) exist to do this, but the bottom line remains 
> that some sort of hardware support will be necessary.

  what if you use a buffer {maybe something that would interactively adjust the 
relative latencies and serve to each card separately.You would have to set 
time somewhere else. {unless you set your time with one card and fed stuff to 
the other relative to that.}