Another nail in the 24 bit audio coffin

24 bit audio resolution is vital in modern digital recording. And in mixing. And in processing.

But it’s increasingly looking pointless in music delivery at the consumer level. I’ve previously reported on the inability of listeners to distinguish between high resolution (typically, 24 bit, 96kHz or DSD) audio and the same content ‘throttled down’ to a CD-standard 16 bit, 44.1kHz presentation.

Likewise, I’ve shown that it’s likely that most of the extra eight bits in 24 bit audio are being used to encode noise.

Now I’ve stumbled across the blog of an audiophile who seems particularly rigorous, and he’s conducted a rigorous test with some 140 volunteers in which they were presented with three high resolution (96kHz sampled) music clips, each in two versions: the original 24 bit recording and the same content reduced in resolution to 16 bits. He took a number of careful steps to make sure that the comparison files could not be easily distinguished by non-listening means.

His results: read the whole thing because he slices and dices the results in interesting ways. But in short, for the three clips the results were random. Precisely half were right and half were wrong for two of the clips, while for the third a few more were wrong than were right. Only one group (audio engineers) were consistently more than 50% correct, but with n=34, and the highest result at 55%, this can also be safely regarded as random.

Notable categories to do worse than the whole: those confident in their results, and hardware reviewers.

Of course, some of the participants got all three right. But, then, some got all three wrong as well. As the author notes:

Looking at the individual responses, there were a total of 20 respondents who correctly identified the B-A-A selection of 24-bit samples, and 21 selected the opposite A-B-B. This too is in line with expectations that 17.5 would pick each of these patterns based on chance alone.

One note: I suspect that if 16 bit and 24 bit resolutions are distinguishable, they are more likely so in lower sampling rate audio: 44.1 or 48kHz. Although to be fair, 24 bit audio does tend to be associated with high sampling rates.

This entry was posted in Audio, Codecs, Imperfect perception. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *