Report of SC-02-01 meeting, 2003-10

Report of the SC-02-01 Working Group on Digital Audio Measurement of the SC-02 Subcommittee on Digital Audio, meeting held in New York, NY., US., 2003-10-11.

The meeting was convened by Chair R. Cabot.

The agenda was discussed and agreed with the addition of two items of new business: S. Harris asked that measurements on Class D amplifiers be discussed. C. Travis had also requested discussion of a standard for jitter measurements.

The report of the previous meeting was unavailable.

Open Projects

AES-6id-R: Review of AES 6id-2000: AES information document for digital audio - Personal computer audio quality measurements.
The chair noted that the document is up for review in 2005. If any changes are to be made they will have to happen in the next year.

Harris had been the principal author of the original AES-6id information document. The original document was written seven or eight years ago and deals principally with Windows-based personal computers. It originated as a tutorial for computer people and covered how you actually measure the audio sub-systems. Harris felt that there was a lot of good material in the document but that some was out of date. For example, all word lengths are specified at 16 bit and the highest sample rate considered is 48 kHz. There is no mention of newer PC audio standards. It includes nothing about compressed audio formats

D. Ranada observed that extending AES-6id to more modern signals would need to deal with issues like the relevance of conventional weighting measurements with low bit rate coded audio?

T. Kite volunteered to take on the task of revising AES-6id and was asked to prepare a schedule for the document development. Harris noted that he would gladly share the many notes about the existing document he has accumulated.

AES17-R: Review of AES17-1998: AES standard method for digital audio equipment - Measurement of digital audio equipment
AES17-1998 was now due for immediate review under the AESSC five-year rule. The meeting agreed that the standard should be reaffirmed as soon as possible. Further development could then take place without undue time pressure.

Development Projects

AES-X088: Objective evaluation of Perceived Audio Quality, PEAQ
M. Keyhl noted that this project is currently dormant.

He felt that it could be developed into a framework to produce a document for loudspeaker applications. The chair suggested that such a departure from the original scope would require a fresh project proposal.

The original focus was on quality testing according to PEAQ standards which is an ITU standard regarding audio quality testing and which was primarily developed for rating low bit rate codecs. However, codec testing is now quite state of the art so the topic for this group is: how can PEAQ be used outside of codec testing?

The lower-quality end of the scale has been tackled by using subjective test procedures developed by the ITU (and EBU) called MUSHRA. It is not an absolute measure but rather a relative comparison. This is being discussed under project AES-X74 by SC-06-04.

This project will be reviewed at the next meeting to consider whether it should continue.

AES-X102: Liaison with IEC MT6106: Audio and audiovisual equipment - Digital audio parts - Basic measurement methods of audio characteristics
The chair invited the IEC project leader, M. Furukawa, to give a quick summary to this group. The general (IEC 61606-1) and consumer (IEC 61606-3) portions of the standard have been approved and released. The professional equipment portion of the standard remains to be drafted.

Furukawa asked that the AES group generate the draft of the professional document. It would differ from a consumer document in basic things like connectors and levels.

Several group members noted that it also needed to differ in some testing specifics due to some shortcomings of the current IEC documents when considered from the viewpoint of current professional measurement practice. Furukawa expressed a desire that measurements should not be specific to company A or user B. They should yield the same plain value.

AES-X118 ITU study group 6
The chair asked S. Lyman to summarise the ITU-R project and its recent developments.

The intent was to find an objective method of measuring loudness. The ITU-R Working Party 6P Special Rapporteur Group 3 (SRG3), chaired by C. Todd, solicited program samples from five different labs around the world and performed subjective evaluations of them. The results of the subjective evaluations and sound samples were sent to the Communications Research Center (CRC) in Ottawa. The various meter proponents also sent their meters there. The CRC used the meters to measure the sound samples and looked at the results to see how close they came to the subjective assessments of those same sound samples.

As a separate issue, digital media overload abruptly and thus even momentary overload should be avoided. The ITU are looking for audio metering characteristics to indicate overload. The chair noted a recent paper at the AES conference in Denmark in the summer. The amount of overload in some real-world CDs far exceeds what you get from simply not correctly sensing peaks between audio samples and indicates a degree of operational disregard for this parameter. Todd has requested this AES standards group to develop a method of indicating the onset of digital overload.

Because of the differing issues surrounding loudness metering and peak overload measurement it was suggested that the work be handled in two projects, one for each.

Lyman pointed out that an e-mail reflector has been set up by SRG3 to support the loudness measurement work within the ITU-R. Relevant documents will be copied to this working group for information. The separate topic of peak level measurement will be progressed as an AESSC project.

New Projects

Class D amplifier measurements
Harris noted that both his customers and his company face difficulties measuring Class D amplifiers, specifically types with a digital input and analog output. He felt that the existing AES17 does not adequately cover this technology and in some areas may be inappropriate.

The Class D amplifier can be thought of as a high-powered D/A converter and may be addressed by normal audio measurements like THD and signal to noise ratio. They also involve typical D/A converter measurements like linearity verses amplitude. Some Class D amplifiers, just like some D/A converters, are designed to mute their outputs with an all-zero input. To counter this, some of the test methods are there to insure the amplifier is operational all the time you are trying to measure its dynamic range. Other tests are useful at looking for tones as many Class D amplifiers are quite capable of producing tones just like DA converters. There are also some power related issues, like amplifier power rating, how many channels are going to draw significant power in a multi-channel environment, how long before the unit overheats, etc.

D. Ranada elaborated on some of these points from the point of view of measuring audio equipment for magazine reviews. One very useful test is a slow ramp of DC. Evaluation consists of high-pass filtering the signal very well, reproducing the result on speakers and listening for any tones or noises. An existing test for noise modulation uses a decaying sine wave envelope. The original signal is filtered out to reveal all the garbage coming out and to see if its level varies with input level. The variation in level is called noise modulation: if the noise is pumping up and down as the signal changes it is believed to be more audible. The object is to get zero noise modulation; Ranada observes this to be less acceptable than a higher - but steady - background noise value. The noise level measurements are made with A-weighting.

Ranada also uses a noise limiting resolution test. The amount of dither in the digital input is decreased from the equivalent level down to the 24-bit level over a calculated time span. The measurement notches out the probe tone and looks at the decay of the dithered background noise level. At some point the measured noise will level off because the noise of the system will be higher than that of the digital input. The effective resolution of the system can be calculated as if it were a digital system with triangular dither. This is a very attractive kind of measurement from a consumer stand-point. All the test tones Ranada uses now are dithered.

Jitter measurements
Travis indicated that he is developing measurement methods to characterise sampling jitter on audio signals. He expects to have a submission to the group in time for the Berlin meeting in May 2004,

New Business

There was no new business.

The next meeting is scheduled to be held in conjunction with the AES 116th convention in Berlin, Germany, 2004-05.

AES - Audio Engineering Society