Topic: Suggestion for a one-and-half blind resolution test.

[Petrus]

It would be interesting to see if somebody really could tell 16/44.1 from 24/96 recording.

It would be fairly easy to try:

- make a 3-5 min long high quality 24/96 recording with a good recorder (like SD722) using good microphones in a quiet place (to get max dynamic range).
- cut the audio in 30 sec segments
- convert one random segment to 16/44.1 and save it
- reconvert it back to 24/96
- splice all segments back together
- listen and try to tell where the modified segment is.

The file could be placed somewere in the net for people to upload.

[Brian Skalinder]

Good idea, Petur.

Not quite the same test, but I think useful nonetheless:  I have the same sample clips at 24/48 and 16/48 in my dither comparison I created some time ago (with a single comparison of 16/44 since R8Brain wouldn't dither without also resampling).  I ABX'd the original 24/48 v. several of the dithered 16/48 and found it pretty easy to distinguish the differences.  Was the difference jaw-dropping?  I don't think so.  But did I like what I heard at 24b v. 16b?  Yup.  Files here:

ftp:  tapers.org
login:  ftp4all
pass:  ftp4all
dir:  /drive1/_gear_comparisons/dither.flac

Files named *_master are the original 24/48, the other file names are somehow descriptive of the dither algo used to get down to 16-bit (see text file for more info).

For reference:  original download thread, original discussion thread