Comments to date on REAFFIRMATION of AES5-2008, AES5-2008 AES recommended practice for professional digital audio - Preferred sampling frequencies for applications employing pulse-code modulation, published 2013-07-01 for comment.
Comment received from Richard Hess, 2013-07-01
Hello, Mark and team,
RE: AES5: AES recommended practice for professional digital audio — Preferred sampling frequencies for applications employing pulse-code modulation
[I think the time has come to re-organize this standard to reflect the reality of the archival world.
===================================
ORIGINAL:
5.2.3 For applications with an audio bandwidth greater than 20 kHz or transition region in the anti-alias filtering a rate of 96 kHz may be used.
===================================
CONCEPTUAL CHANGES [With comments in brackets]:
5.1.1. Primary sampling frequency
[5.1.1 then reads exactly as the original 5.1 read. Then add:]
5.1.2. Preferred wider bandwidth sampling frequency
For digital audio encoding where a bandwidth greater than 20 kHz and/or a broader transition region in the anti-alias
filtering is desired a rate of 96 kHz shall be used.
NOTE 1 This frequency is widely used in retrospective digitization of analog sources both from tape and from disc.
It has been reported that this increased bandwidth may assist in subsequent post processing to reduce certain types of noise found in the original recordings.
NOTE 2 This frequency is widely used for recording and tracking.
[Replace current paragraph 5.2.3 with:]
5.2.3. For cases where a quad speed sampling rate is desired, the preferred frequency is 192 kHz.
===================================
[OPTION 2: Leave 5.1 alone and make a new 5.2.1]
5.2.1. For digital audio encoding where a bandwidth greater than 20 kHz and/or a broader transition region in the anti-alias filtering is desired a rate of 96 kHz shall be used.
NOTE 1 This frequency is widely used in retrospective digitization of analog sources both from tape and from disc.
It has been reported that this increased bandwidth may assist in subsequent post processing to reduce certain types of noise found in the original recordings.
NOTE 2 This frequency is widely used for recording and tracking.
[What was 5.2.1. becomes 5.2.2.]
[What was 5.2.2. becomes 5.2.3.]
[A new 5.2.4 is created]
5.2.4. For cases where a quad speed sampling rate is desired, the preferred frequency is 192 kHz.
[What was 5.2.4. becomes 5.2.5.]
[What was 5.2.5. becomes 5.2.6.]
===================================
[Commentary follows:]
There are many ways to say the same things. I don't want to get hung up with the precise wording, but would like to work with the committee to see if we can get something like this into the standard.
This will become a more useful tool to convince archivists NOT to digitize oral history cassettes at 96 kHz.
Cheers,
Richard
Reply from John Grant, Chair SC-02-02, 2013-07-03
Thank you for your input. I agree that we should look at restructuring the way we standardise sampling frequencies, but would prefer to include AES11 (which is due for review next year) in that process. AES5 needs to be either reaffirmed or revised this year, and I think it would take longer than that to complete what is likely to be a fairly major revision.
I therefore propose we reaffirm AES5 as is for now, but also begin immediately the process of reviewing AES5 and AES11 with the intention of revising both of them, and possibly also creating an Information Document with recommendations for particular applications such as archiving, by the end of 2014. I hope that is acceptable to you.
Please reply by the end of the comment period if this reply is not acceptable to you. You may also ask us to consider your comments again for the next revision of the document. You may also appeal our decision to the Standards Secretariat.
John Grant
Chair, SC-02-02
Response from Richard Hess, 2013-07-0
Hi, John,
I understand wanting to keep things in sync. However, if AES5 is reaffirmed this year, can it be revised next year or does it have to wait for the revision cycle? It seems from your comment that it can be, which is great.
Is there a way I can stay in the loop for AES11 and AES5? While I probably cannot attend meetings, I think I may be able to contribute dialogue to these issues as they are something I have thought long and hard about.
The only comment I have received, however, from a colleague is that he prefers now capturing his high-end classical recordings direct to 44.1 for CD release and feels that archivists have gone way overboard wanting 96/24 for everything. I am purposely not naming him as I know this is public and I don't know if he wants to be dragged into the discussion.
I think as the AES we have a duty to help people who are non-technical most effectively use the tools of our trade. The sampling rate and bit depth decisions are often made by archivists and librarians.
Cheers,
Richard
Reply from Mark Yonge, AES Standards Manager, 2013-07-04
Hi Richard,
You wrote: "Is there a way I can stay in the loop for AES11 and AES5? "
Certainly. I have registered you to AES Standards working group SC-02-02 who are responsible for AES5 and AES11. You will receive all emails on this (and other) discussions and will be able to contribute as you see fit.
regards,
Mark
Comment received from Kevin Gross, 2013-07-01
Consider removing 32 KHz recommendation. We did not find any evidence of significant use of this rate in X192 development. Radio uses 48 or 44.1 kHz. Television uses 48 kHz. AES5 references ITU-T Recommendation J.53. This recommendation was apparently published in 1990 and has not been updated. I am unable to find a copy to assess it but I suspect a 23 year old unmaintained digital audio recommendation this is of limited relevance at this point.
Consider deleting or trimming discussion of sampling frequency conversion in section 4.2. Readers may get the impression that oddball conversion rates compromise sound quality. This is not the case. Sampling frequency conversion at certain integral ratios is less computationally intensive than a general ratio. I believe the difference is at most 2x and is probably relevant only in specific optimized software implementations. Computational efficiency is much less significant consideration today than it was when AES5 was first published.
Kevin Gross
+1-303-447-0517
Media Network Consultant
Reply from John Grant, Chair SC-02-02, 2013-07-03
Hi, Kevin. This is my formal response to your comment, as chair of SC-02-02.
I don't have a problem with deleting 4.2 except that by doing so we change the process from reaffirmation to revision, so would have to withdraw the document and issue a CFC for the revision instead. The same applies to removing 32 kHz, and there's also the possibility that there are applications still using 32 kHz which for some reason didn't make it onto X192's radar -- we ought to try to find out whether there is any reason (other than inertia) why 32 kHz is still in J.53.
I think we need to look in depth at both AES5 and AES11 (which is due for review next year), and possibly also create an Information Document with recommendations for particular applications. AES5 needs to be either reaffirmed or revised this year, and I think it would take longer than that to complete what is likely to be a fairly major revision.
I therefore propose we reaffirm AES5 as is for now, but also begin immediately the process of reviewing AES5 and AES11 with the intention of revising both of them, and possibly also creating a new Information Document, by the end of 2014. I hope that is acceptable to you.
Please reply by the end of the comment period if this reply is not acceptable to you. You may also ask us to consider your comments again for the next revision of the document. You may also appeal our decision to the Standards Secretariat.
In reaffirming AES5 I would not like to burden the user with too many examples. However, there is reference to broadcast audio and audio with video, as well as transmission with restricted channel capacity where a bandwidth of 15 kHz is considered adequate, and yet there is no mention of archival audio for preservation or of audio forensics applications. One can only assume that these two significant areas come under section 5.2.3 as applications having a bandwidth greater than 20 kHz.
I wonder if it should be mentioned somehow that when choosing a sampling frequency, consideration should be given for the implications of that choice on the lifecycle of the content beyond its primary purpose. For instance, when an oral history is more than just a voice recording requiring only intelligibility for transcription.
Reply from John Grant, Chair SC-02-02, 2013-07-03
Thank you for your input. I agree that there should be more detail on the sampling frequencies that are appropriate for different applications, but would prefer to include AES11 (which is due for review next year) in that process. AES5 needs to be either reaffirmed or revised this year, and I think it would take longer than that to complete what is likely to be a fairly major update.
I therefore propose we reaffirm AES5 as is for now, but also begin immediately the process of reviewing AES5 and AES11 with the intention of revising both of them, and possibly also creating an Information Document with recommendations for particular applications such as archiving, by the end of 2014. I hope that is acceptable to you.
Please reply by the end of the comment period if this reply is not acceptable to you. You may also ask us to consider your comments again for the next revision of the document. You may also appeal our decision to the Standards Secretariat.
