WBF Library: Audio signal Processing Papers/Presentations

Apr 3, 2010
16,022
0
0
Seattle, WA
#1
This posting archives the best papers on audio signal processing. I will keep this first post updated with links and suggestions from everyone.

One of the best papers I have read on the topic of audio sampling rates and resolution from Bob Stewart of Meridian fame: http://www.meridian-audio.com/w_paper/Coding2.PDF

Dither
Most famous paper in this area: Resolution Below the Least Significant Bit in Digital Systems with Dither
John Vanderkooy and Stanley P. Lipshitz
http://www.drewdaniels.com/dither.pdf

The Application of Dither and Noise-Shaping to Nyquist-Rate Digital Audio: an Introduction.
Christopher Hicks, Communications and Signal Processing Group, Cambridge University Engineering Department
http://www.benchmarkmedia.com/sites/default/files/documents/ditherap.pdf
Highly technical paper but does have easier to read sections that talks about noise-shaped dither to increase dynamic range in critical frequencies including Sony's Super Bitmapping (SBM) scheme.

"Theory of Dithered Quantization" by Wannamaker" http://www.digitalsignallabs.com/phd.pdf

Sampling Rate
I found this paper as explanation of 192 KHz sampling not being a good step forward. In reality, I find the paper hard to follow even though it does touch on that topic: http://www.digitalaudio.dk/media/3Tradeoff_of_192_kHz.pdf
Article on DSD/DXD: http://www.merging.com/uploads/assets//Merging_pdfs/dxd_Resolution_v3.5.pdf

Room Correction/EQ
Nice but very technical paper from Dirac on difference between filter types and impact on Room EQ: http://www.dirac.se/media/12044/on_room_correction.pdf

Jitter
Great AES paper by Julian Dunn: http://www.nanophon.com/audio/diagnose.pdf
Great paper covering all aspects of AES digital interface including esoteric concepts such as jitter induced by the cable: http://tvit.org/GuyCD/Reports/Conference/aes-2id-1996.pdf
Excellent tutorial on jitter: http://cp.literature.agilent.com/litweb/pdf/5989-5718EN.pdf
Jitter measurement paper: http://www.msbtech.com/support/JitterPaper.pdf

Cabling
Design of High-Performance Balanced Audio Interfaces: http://sound.westhost.com/articles/balanced-interfaces.pdf
Cable testing: http://www.canare.com/UploadedDocum...phone Cable Perfrmance and Specifications.pdf

Dynamic Range of Audio
Dolby Paper: http://www.zainea.com/Dynamic range.htm
 

DonH50

Member Sponsor & WBF Technical Expert
Jun 23, 2010
3,522
2
38
Monument, CO
#3
This is a GREAT idea, Amir! I have reference texts and articles on these and related subjects, including some of the original noise decorrelation (dither) papers and patents, but (a) for RF/radar, not audio, and (b) on paper, in boxes in the black hole sometimes laughingly called our basement storage room.

A collection of links to key technical papers, all in one place, sweeeeet! <insert drool emoticon here>

Now, if I could just find more time to read them all... :) - Don
 

hdplex

New Member
Dec 16, 2010
5
0
0
#9
If DXD is 24bit/352.8khz, isn't it just a form of PCM using 5bit Delta Sigma modulator?
According to Bob Stewart's paper, PCM with 58Khz sample rate is fine to achieve transparency. Higher sample rate is not justified and considered a waste of bandwidth.
On the other hand, Merging said higher samples rate e.g. 352.8Khz make it easier to define pulse response, lower sample rate will require steep filters and will have effect of pre and post ringing.

Quote from Merging article:
"Two factors define the pulse response — the roll-off frequency of the sampling filter (anti-alias filter) and the slope of the filter roll-off. A high roll-off frequency gives a high amplitude pulse because of more high frequency content being present. A slow roll-off slope will produce less pre and post ringing of the pulse and thus a more precise time point for the pulse. Due to its relaxed filter slope and high bandwidth DSD has a perfect pulse transparency. The results are different for the different PCM rates. These have a higher pulse when the sample-frequency increases, but the steep filters needed to comply to Nyquist (i.e. having full attenuation around the half sample frequency) will have the effect of pre and post ringing. At the higher rates a more relaxed approach can be made towards the filter steepness since the risk of having audio contents folded down is very small as the efficient amplitude of the audio is very low at rates beyond 50kHz, for example. "


Can I say Bob Stewart's conclusion is based on theoretical assumption without taking real world filter design into consideration?

Also in Merging's article Sony acknowledged DXD as a "legitimate" acquisition format for DSD.;)
I am reading Sony basically admit super high sample rate multi bit PCM is better than 1bit DSD.:D Wasn't that the key argument put forward by SACD opposers in AES 2001 conference?


DXD's PCM nature make it editable without conversion and significantly lower high frequency noise help improve conversion if needed.
 

FrantzM

Member Sponsor & WBF Founding Member
Apr 20, 2010
6,469
0
0
#11
Apr 3, 2010
16,022
0
0
Seattle, WA
#20
Thanks JJ. Great to see you here. At first I thought the mp3 files at that site was some kind of sample but couldn't figure out why they would be there. Then I realized they are recordings of your presentation :). I thought we had a link to your presentation some place here. Great to have the talk go with it.