frankR
Head-Fier
- Joined
- Feb 16, 2008
- Posts
- 68
- Likes
- 0
Quote:
I’ll refer back to my earlier analysis posted last month.
http://www.head-fi.org/forums/f46/24...ml#post4017905
Quote:
Matlab is high-level scientific mathematics analysis package. It can import PCM audio. I analyzed short samples of various lengths (millisecond to several seconds) and of different instruments from the high-res-low-res comparison audio clips. I performed power frequency spectrum analysis of these samples using fourier transform. The fourier transform gives a time-independent analysis of the audio frequency spectrum.
What I found was consistent with my earlier crudely performed analysis. There were significant differences between their respective power frequency spectra in the audible frequency range (20 hz to 20 khz).
Quote:
I’ve read similar studies on the subject. There is a belief that these types of double blind audio tests are inherently flawed because, unlike images, the human brain has difficulty retaining fidelity differences, though differences are perhaps perceived.
Nevertheless, these findings do not nullify what my own ears and analysis makes me believe.
An interesting experiment at the upcoming CES in Las Vegas will compare live play-back to recorded material using suburb equipment.
Ampzilla, VMPS and Live Music at January CES
Originally Posted by nick_charles /img/forum/go_quote.gif Interesting, your results simply do not agree with a large scale carefully controlled study in a peer reviewed journal. Can I ask you for some more details ? |
I’ll refer back to my earlier analysis posted last month.
http://www.head-fi.org/forums/f46/24...ml#post4017905
Quote:
Originally Posted by nick_charles /img/forum/go_quote.gif Can you describe how you did this ? - what procedure did you use to get the two samples into Matlab - what were the measured distortion figures you found ? Matlab is just a computational environment and language in't it ? what audio analyzing functions does it support ? |
Matlab is high-level scientific mathematics analysis package. It can import PCM audio. I analyzed short samples of various lengths (millisecond to several seconds) and of different instruments from the high-res-low-res comparison audio clips. I performed power frequency spectrum analysis of these samples using fourier transform. The fourier transform gives a time-independent analysis of the audio frequency spectrum.
What I found was consistent with my earlier crudely performed analysis. There were significant differences between their respective power frequency spectra in the audible frequency range (20 hz to 20 khz).
Quote:
Originally Posted by nick_charles /img/forum/go_quote.gif Key findings, not one subject (n = 60) was capable of detecting the difference between high res and 16/44.1 at a sufficient level of confidence (95%), the total # of trials was 554 and correct answers 276 (49.819%). |
I’ve read similar studies on the subject. There is a belief that these types of double blind audio tests are inherently flawed because, unlike images, the human brain has difficulty retaining fidelity differences, though differences are perhaps perceived.
Nevertheless, these findings do not nullify what my own ears and analysis makes me believe.
An interesting experiment at the upcoming CES in Las Vegas will compare live play-back to recorded material using suburb equipment.
Ampzilla, VMPS and Live Music at January CES