yfei
100+ Head-Fier
- Joined
- Jul 1, 2002
- Posts
- 452
- Likes
- 209
Thanks for the correction, that's very informative. After lookup the usb audio class, yes, "Error detection via CRC, but no retry or guarantee of delivery." Why it is designed this way is because they assume "if a packet or frame was dropped every now and again, it is less likely to be noticed by the listener."
(http://www.beyondlogic.org/usbnutshell/usb4.shtml)
Which means the quality of USB cable is important: lost / corrupted packets are just skipped, and they hope user's won't hear the difference.
Quote:
(http://www.beyondlogic.org/usbnutshell/usb4.shtml)
Which means the quality of USB cable is important: lost / corrupted packets are just skipped, and they hope user's won't hear the difference.
Quote:
Currently, for USB Audio (even class 2), it is not It is a common misconception: USB Audio Class 1.0 & 2.0 deal with isochronous endpoints, which does not feature error correction.
The S/PDif features a low-level protocol that splits the audio into frames, sub-frames and time-slots.
USB audio does not feature re-transmission or error correction. Only the bulk transfert mode allows for error correction (mass storage devices). The Musiland convertors (Monitor 01 & 02) are the only USB devices I know that don't relie on the isochronous transfert mode (as found in all synchronous, adaptive & asynchronous converters). They rely on the bulk transfert mode and require specific drivers (not USB Class complient).