[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: measuring music playback quality
There are three, fundamentally different music play back quality measurement approaches :
1) HERE+NOW, you strive for the illusion that what you hear is now present in the room where you are listening
2) THERE+THEN, you strive for the illusion that what you hear is the same as was present on the place where the recording was made
3) NO REFERENCE
A loudspeaker that sounds perfect for HERE+NOW will not reproduce THERE+THEN. So a subjective test will always be biased by what material is used and what the exact question is to the subject.
One should realize that quality is the perceived distance between Realisation and Idealisation and that you cannot measure quality when the ideal is unknown. This also results in different measurement approaches in the objective domain (PESQ, PEAQ, POLQA)
(See also http://www.linkedin.com/pub/john-beerends/29/340/52 )
By the way HERE+NOW can be achieved by using an anechoic recording + reproduction by a single loudspeaker having the same directivity behaviour as the recorded source (e.g. high quality voice reproduction will use this approach). For THERE+THEN you should apply independent diffuse field equalization, see .
 K. L. Kantor and A. P. Koster, "A psychoacoustically optimized loudspeaker," J. Audio Eng. Soc., vol. 34, pp. 990-996, (1986 Dec.).
From: AUDITORY - Research in Auditory Perception [mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of Christopher Hummersone
Sent: 26 January 2012 11:58
Subject: Re: measuring music playback quality
Some colleagues of mine ran tests like this a few years ago. See:
Beresford, K, Ford, N, Rumsey, F and Zielinski, SK (2006) Contextual effects on sound quality judgements: listening room and automotive environments, 120th AES Convention, Paper 6648, http://epubs.surrey.ac.uk/565.
Beresford, K., Ford, N., Rumsey, Francis and Zielinski, Slawomir (2006) Contextual effects on sound quality judgements: Part 2 - multi-stimulus vs. single stimulus method, 121st AES Convention, Paper 6913, http://epubs.surrey.ac.uk/526.
They found a single stimulus method to be invalid under their particular experimental setup.
I think the point is that everyone has an internal reference: an idea of what sounds good, based on their experience. This can vary dramatically between people, and can be biased by many factors. The identified reference in multi-stimulus experiments attempts to minimise this effect by providing everyone with a common reference.
On 24 Jan 2012, at 07:48, Marc Schoenwiesner wrote:
> Dear all,
> I am trying to get an overview of different methods that have been
> used to measure the perceived quality of a music recording with the
> goal to evaluate different loudspeakers, playback systems, earplugs,
> I have seen a number of papers that compare two or more setups by
> asking listeners which recording they prefer. I have not come across a
> measure of absolute quality rather than quality comparisons, for
> instance by asking the (expert) listener to rate the quality of a
> single system. Is anyone aware of such an experiment? (I am
> particularly interested in absence of spectral colouring and
> distortions, but any example will do.)
> This message was sent using IMP, the Internet Messaging Program.
This e-mail and its contents are subject to the DISCLAIMER at http://www.tno.nl/emaildisclaimer