Welcome Guest ( Log In | Register )

Reply to this topicStart new topic
Adaptive double-blind listening tests?, Tests that optimise the "samplepoints" where it matters the
post Jul 10 2007, 22:01
Post #1

Group: Members
Posts: 570
Joined: 1-November 06
Member No.: 37047

When I was into this stuff, I remember reading about som kind of "ladder" tests for detecting listener thresholds for audio degradation that are supposedlyd monotonic and gradual, such as decreasing the bitrate of a lossy codec.

The approach was to start out with the "simple" (large degradation), then serve the listener "harder" (less degradation) samples as long as he/she was able to distinguish them. When an error was made, the system would step back 2-3 levels and continue from there.

This way, the test would (hopefully) "home in" in the area of just noticable differences, and give as much data as possible in this range, in stead of wasting valuable test-time for samples that was dead-easy or imposible to distinguish for the particular person under test.

Is this a test on its own or simply a way of organising ABX-tests?

I tried to google this but found nothing. Anyone recognize it?


Either the "backtracking" amount or the "X" et each quality-level would have to be variable, if not, the listener could possibily hear that the "detectable" sample was introduced, and figure out the system.

This post has been edited by knutinh: Jul 10 2007, 22:06
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:


RSS Lo-Fi Version Time is now: 26th November 2015 - 15:33