Report of the meeting of the SC-02-01 Working Group on Digital audio measurement techniques of the SC-02 Subcommittee on Digital audio, held in New York, NY., US., 2013-10-18
The meeting was convened by chair T. Kite.
The agenda and the report of the previous meeting, held in Rome, 2013-05-05, were approved as written.
AES-X102 Liaison with IEC TC100 MT 61606
No action has been taken since Rome. The secretariat will draft a liaison statement to IEC TC100. We expect to offer proposals for updating IEC 61606-3 once the update to AES17 is complete.
AES-X118: Liaison with ITU-R study group 6C
Since Rome, Kite updated the electronic copy of AES-R7 with the new information about 0.4fs, and the observation that the peak error will typically be much lower on real program material. M. Yonge will convey our confirmation of 4x as the recommended interpolation ratio to the ITU.
AES-X187A: Revision of AES17
Kite gave an overview of the changes that have been made to AES-X187A since Rome. The document is now close to completion, with only a handful of measurements still to be rewritten, and a small number of TODOs to resolve.
TODO #1: It has been pointed out by H. Morfett-Jones and others on the reflector that the overload behaviour measurement for EUTs with an analog input is not reliable. Morfett-Jones has demonstrated a reasonable case in which an EUT which rolls over and a different EUT which clips can result in the same measured value in the overload behaviour measurement. The measurement therefore needs to be updated.
After some discussion suggesting that a time-domain observation (for instance using an oscilloscope) might be the best way of performing this measurement, we eventually came round to the idea that a better way might simply to measure the THD+N of the EUT when overdriven. A clipping EUT will show fairly high THD+N, but an EUT that rolls over will show extremely high values.
Kite offered to look into this idea and come up with a proposal on the reflector. If anyone has further ideas along these lines, please feel free to suggest them on the reflector.
TODO #2: The issue of why we have two methods of establishing the maximum input level (namely, the THD+N method and the compression method) was resolved by I. Dennis, who recalled that the compression method was most useful for devices that soft-clip. (Note that soft-clipping devices are not excluded by the scope clause of AES17 that excludes dynamics compressors. Soft clipping is typically a memoryless modification of the EUT's input-output transfer function, not a compression of signal dynamics.) There is therefore a need to keep the two methods. It was decided to include in the update a note on which method is recommended for hard-clip and soft-clip EUTs. It will also be made clear that one method or the other should be used. Thus there is no need to monitor the compression while using the THD+N method, and vice versa.
Window-width filters: Kite explained that the issue of whether an FFT analyzer should be part of the standard equipment was resolved by requiring window-width filters for certain measurements. (The use of window-width filters implies an FFT analyzer.) The "noise problem" with window-width filters (described in Annex 8 of the current draft of AES-X187A) was again discussed. Dennis expressed some concern that filters designed for constant bandwidth regardless of FFT bin width would no longer be truly "window-width", since bins beyond the main lobe may need to be included to obtain the desired bandwidth. Kite will attempt to remove the confusion in AES-X187A with judicious use of language.
Kite also requested that members of the group look at Annex 7 of the current version of AES-X187A. This is the section on fast methods for production testing. It is new to AES17, and Kite would appreciate any thoughts on it. Please post comments to the reflector.
AES-X206: Frequency Response Data file format
In Rome, Kite presented an XML schema defining the .frd file format, and an XSLT transform to turn compliant XML documents into legacy .frd text files. (The group had felt that the exisiting text-based .frd file format was outdated, and too ad hoc to be codified into an AES-quality standard.) After the Rome meeting, we understand that the originator of this project rejected this approach.
At this point the group feels that our work on this project is complete. The schema and transform work as advertised, and bring the file format into the modern era. However, we have no control over whether they are adopted.
AES-X217: Performance standard for archival ADCs
In Rome, the group accepted the project proposed by C. Lacinak to create a standard specifying both the measurements that must be made, and the limits on those measurements, to establish a minimum level of quality for archival ADCs. No work has been done since Rome.
The draft proposal needs to be separated into the methods of measurement, and the performance criteria for those measurements. It was decided that Yonge will show Lacinak how to do this by starting the work and then handing it over. Lacinak will continue the work.
Lacinak's intent is to establish several levels of performance criteria for different applications. A question was raised about whether this was necessary, given that even reasonably cheap converters have excellent performance. Lacinak responded that multiple levels may not be necessary; that remains to be determined by a survey of devices on the market.
Lacinak also informed the group that he expected that the archivist (that is, the person or company tasked with performing the archiving) would be responsible for performing the tests. It would therefore be beneficial to make the tests as easy as possible to perform.
Whilst most of the tests in Lacinak's proposal are taken from AES17, some are not. Kite offered to take a look at Lacinak's proposal and determine whether any of these new tests should be incorporated into AES-X187A. The group felt that it would be ideal for all the tests in the final performance standard for archival ADCs be tests from AES17.
Liaisons
Liaisons with other groups were discussed in other projects, as reported above.
New Projects
Lacinak brought up the issue of errors that occur between the audio inputs and the mass-storage device when using general-purpose computers as digital recorders. Apparently dropouts can occur randomly, presumably when the operating system is busy working on something else. There is a need to at least detect these errors (error correction was deemed unlikely because of the size of the dropouts that can occur).
Some ideas along these lines were discussed, for instance embedding a pseudo-random sequence in the LSB of the audio stream. Discontinuities in a pseudo-random sequence are easily detected. This could be done offline by processing the saved data. For high resolution (24-bit or 32-bit) audio, the LSB is below the noise floor. Modulating it with a noise-like sequence should have a negligible effect on the recorded audio.
The Working Group decided to explore this problem. Lacinak will share with the group any information he has on this topic. Members are encourage to send their ideas to the reflector.
New Business
There was no new business.
The next meeting will be at the 136th AES Convention in Berlin, Germany, 26-29 April 2014.
