In This Section
- Open Control Architecture - Part 3: Protocol for TCP/IP Networks; AES70-3-xxxx DRAFT proposed for comment
- Open Control Architecture - Part 2: Class structure; AES70-2-xxxx DRAFT proposed for comment
- Open Control Architecture - Part 1: Framework; AES70-1-xxxx DRAFT proposed for comment
- Audio-over-IP network interoperability; AES67 revision published
Comments on DRAFT REVISED AES5-xxxx
Comments received on DRAFT REVISED AES5-xxxx, AES recommended practice for professional digital audio - Preferred sampling frequencies for applications employing pulse-code modulation,
published 2008-08-10 for comment.
Comment received from Kevin Gross, 2008-08-22
- Consider removing section 2 or collapsing it to read, "There are no normative references for this standard."
- Note in 3.2 is arguably unnecessary as this is covered thoroughly and immediately following in section 4.1.
- "Sample rate conversion" (or acronym: SRC) is now the popular terminology used for what's labeled "Sampling frequency conversion" in section 3.3. Consider changing section title to "Sample rate conversion".
- Suggest naming names in introduction of section 4.1 give readers something to ask the librarian (or Google) for if they need more background information: "The Niquist-Shannon sampling theorem requires that ..."
- The term "anti-aliasing filter" is used without introduction in section 4.1. This may give rise to the impression that there's some sort of magical device that gets us around the problem of aliasing. There is no shortcut; signals need to be bandwidth limited prior to sampling. Suggestion: "The form of a bandwidth-limiting filter (also commonly referred to as an anti-aliasing filter in this context) may be selected ..."
- Remove 10 PPM reference in section 5.1. By my careful reading I don't see that the figure serves as any sort of constraint. The reader is, however. led to expect that there's some magic about the figure; There is never any further elaboration. Suggestion: "... instantaneous ratios may vary at any moment within the length of the material."
- I'd like to see more context for note 2 in section 5.1. After reviewing AES3, here's a suggestion: "NOTE 2 pull-down frequencies can be specified by setting the Sampling frequency scaling flag in byte 4 of the AES3 channel status (see AES3-2003 p.17)"
Response from John Grant, chair SC02-02, 2008-09-10These responses are numbered to correspond with the original comments, above:
- A clause 2: Normative References" is a key for navigation in IEC standards style. The first paragraph will be deleted, as you suggest, as an editorial matter.
- This note will be removed as an editorial matter.
- This terminology is deliberate and has been derived after much debate. For the process of sampling or re-sampling (in ADCs, DACs or sampling-frequency convertors, for example) the term "frequency" is appropriate because the interval between sampling instants must be constant. When we talk about the flow of data through a piece of electronics,or through a transport stream, the aggregate rate of flow is important while the period between data elements may not need to be constant. Accordingly, we try to use either, "sampling frequency" or "data rate", as appropriate. While we recognise that the term "sampling-rate convertor" is in wide commercial use, this fact doesn't make that terminology useful in our context.
- Agreed. The text will be adjusted to clarify the useful keywords, "Nyquist-Shannon sampling theorem" as an editorial issue.
- Agreed. This will be handled as an editorial matter.
- When the current draft is compared with the earlier published version of AES5, you'll note that we removed a requirement for sampling-frequency tolerance earlier in 5.1. We should also have removed this second reference. I propose to remove everything in this third paragraph after the word, "Synchronization" so the the complete paragraph reads, "These ratios of frame rate to samples per frame shall be maintained as an average over the length of the material which is to be held in synchronization."
- I propose we change the note to read: "NOTE 2 Division by 1001/1000 can be signaled in the sampling frequency scaling flag in AES3 Channel Status byte 4 bit 7." A bibliographical reference to AES3 will also be added to Annex A, "Informative References".
Chair, SC-02-02 working group on digital interfaces
Reply from Kevin Gross, 2008-09-14
This is all quite acceptable to me. Thanks for the thoughtful responses. I hope my comments were helpful.