So far I thought it is a matter of course that a sound card returns each time the same immediate record of its own equally repeated output. I blame asynchronous clocks for what I observed: At least for the card under test, the phase shift between sent and returned wav file was random for purely sinusoidal signals. Any imaginable influence of speaker and mic was definitely excluded since the card's output was immediately fed back to the card's input.
According to Ohm's law of acoustics, hearing is highly phase- deaf. So phase doesn't matter much except for measurement e.g. of room response derived from comparison between sent and incoming sound. I tested a cheap card from Creative in duplex mode with line out directly fed back into line in, and I was extremely unhappy with the result since gain heavily fluctuated with changing frequency while phase response proved erratic, i.e., even worse. Perhaps, only recording is seriously affected with the SB from Creative. Having little experience with sound cards and no interest in games or artificial sound effects I guess, a Terratec DMX 6Fire would best meet my desire for fairly accurate replay of a wav file and subsequent also accurate recording.
However, I would like to ask for confirmation that I may use this card or any alternative one in a manner which excludes the mentioned irreproducible phase response and for a piece of advice how to do so. Any constant latency is acceptable. If only data aquisition cards will do the job, I would also appreciate pertaining recommendations. Would the Matlab DAQ toolbox be the best option in this case or would I be better off with other systems? I hope for getting guided to an affordable solution.
<P ID="edit"><FONT SIZE=-1><EM>Edited by Eckard on 05/09/03 06:13 AM.</EM></FONT></P>
You have to consider the fact that you're passing the signals through ad and da converters, and they're not transparent, not equal and they are a "path" that implies some time consuming. So few samples of shift are inevitable, and they can't be constant, as the two converters are working in different ways.
That said, if the point of wiew of a professional daw musician can be of any interest, no one of the Creative products is taken seriously for any pro daw.
Decent converters cost each more than a single Creative card.
I'm using Creamware Scope, but there you buy also an incredible dsp Studio, the price is so much higher...but the quality....
Well, the situation might be more tricky than we revealed so far. From my dovish point of view, I would like to make it quite clear that I don't search for absolute fidelity but just for what is indispensible to those who are ready to pay for the option of measuring room response. Professionals are included, of course. However, I feel, many other customers will also strive for getting such a tool. Since the basic requirements are pretty clearcut, in principle, I don't see serious justifications for the obvious widespread lack of knowledge about available systems meeting these requirements and how much do they cost today and how much tomorrow. A 500k license offer is definitely unacceptable for envisioned customers like me. Many acousticans seem to rely on their own DSP based handiwork. In case of some efforts I am not sure whether or not they were already successful. For instance, I wondered why in WinMLS with Digigram VX222 'real and imaginary part of frequency response is not available yet'. I also doubt that published tests of sound cards for Matlab DAQ include correct phase.
Aren't such efforts possibly ineffectively wasted ones, at least as a rule? SBs are cheap because millions of them are sold. What sound card could be modified as to reproducibly provide phase? One idea of mine is to look for an electronic 'knob' that cuts possibly different delay of DAC and ADC to an equal length by means of external synchronisation. Maybe, there are other approaches, too.