Po-Jen Chiang wrote:
> I did a little research on these external audio interface. I think
that
> Edirol/Roland products like UA-1D, UA-3D, UA-5 resample the
digital signal
> because it could not sync to external clock of digital signal.
That is, it
> resamples to 48k and then resample again to your desire sample
rate such as
> 44.1k into the computer. Even 48k digital signal, it still
resamples. Thus,
> not bit-perfect.
I just did a quick test with my Edirol/Roland UA-30. I generated a
synthetic sine signal with a constant frequency of 12 kHz and saved
it into a .wav file with a sampling rate of 48 kHz. The resulting
binary pattern of the resulting sound file is very simple and can be
easily verified (the binary sequence is : 0, +16383, 0, -
16383, ...). I transferred that file onto my SONY TCD-D3 DAT
recorder (sampling rate 48 kHz) via the SP/DIF optical out of the UA-
30 (sample rate selector set to 48 kHz). I then played that test
signal back and recorded it via the UA-30 digital interface (with
the recording software set to 48 kHz). The resulting .wav file
showed very minor changes only. I found some deviations in the
least significant bit of some samples (the binary value changed from
0 to -1 or from 16383 to 16382). This corresponds to a very small
attenuation. Both long-term spectra and spectrograms of the re-
recorded test signal did not show any significant distortions. There
was no phase shift over time, which means, that the clock
synchronization worked fine. However, there were some strong
distortions during the first 250 milliseconds after starting the
recording process on the computer. The synchronization circuit seems
to require that time until it works properly.
Regards,
Raimund Specht
Avisoft Bioacoustics
________________________________________________________________________
________________________________________________________________________
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|