Developing a simple petition for improved bitrates and increased spectrum allocation for DAB+ broadcasting in Australia

[MLXXX]

20 minutes ago, alanh said:

Yes and the CD is of worse quality

Not for adult human beings. It is found that human beings in double blind tests have been unable to distinguish a 44.1kHz 16 bit stereo CD at usual recording and listening levels from a 192kHz 24 bit original source.  (Dogs might be able to tell a difference because of their higher upper frequency limit of hearing.)

Why you are trying to quibble about such matters as resampling and dither makes no sense to me when what we are discussing are low/moderate bitrate codecs. If you examine the published test results, 64kbps HE-AAC is by no means transparent for human test subjects. And yet CD quality (at usual recording and playback levels, and mastered using dither) is transparent for human test subjects.

 

23 minutes ago, alanh said:

Why are you avoiding my question.

I have given my own subjective impressions many times in the past. I will wait until I (or others) have uploaded some current examples (or linked to some new examples) before offering further subjective impressions. 

Why you are fixated on my subjective impressions I don't know, alanh. It is no secret that 64kbps HE-AAC is not transparent. There are plenty of test result graphs and tables. And I'm sure there are published subjective reports too if you were to just search for them, rather than badgering me!

[Pesto Lovin' Man]

7 hours ago, alanh said:

The 44.1 kHz sample rate is so it can be locked to NTSC video which is no longer used in the USA.

Huh?

Youre crazy Al. Go do the math. 44.1KHz fits nicely into PAL 25 FPS fields, but not NTSC 29.97 FPS fields.

Or if you accept they pushed NTSC from 29.97 FPS to 30 FPS to make it work its just as true to say "The 44.1 kHz sample rate is so it can be locked to PAL video which is no longer used in the rest of the world". 44.1kHz was chosen because it could be fitted into both.

p.s. 48kHz doesnt fit in to PAL or NTSC either.

7 hours ago, alanh said:

More importantly the down sample is synchronised so exactly each alternate sample is removed. Dithering is trying to cover up the distortions created.

Go away and learn about noninteger resampling before you make yourself look more stupid.

7 hours ago, alanh said:

Why are you avoiding my question.

After taking a refresher look at the history of this subject on this stupid forum its clear MLXXX has answered your question many times before. You just refuse to beleive him or others and keep on backing up your own opinions with wild assumptions ("3.5 million listeners cant be wrong"? really?!).

Since hes answered you many time before & you just use it as another excuse to not beleive him I dont blame him for not bothering to answer you again. You want to argue with him over that? Do your own bloody spadework.

[hrh]

13 hours ago, Pesto Lovin' Man said:

44.1KHz fits nicely into PAL 25 FPS fields, but not NTSC 29.97 FPS fields.

Or if you accept they pushed NTSC from 29.97 FPS to 30 FPS to make it work its just as true to say "The 44.1 kHz sample rate is so it can be locked to PAL video which is no longer used in the rest of the world". 44.1kHz was chosen because it could be fitted into both.

p.s. 48kHz doesnt fit in to PAL or NTSC either

I think there could be a problem with that.
48kHz is what is used in the television industry - certainly in Australia at the least, and if the "samples" I have which is sourced from overseas is any indication, so do they.
Here are a couple of interesting reads

http://www.tvtechnology.com/opinions/0004/digital-audio-sample-rates-the-48-khz-question/184354

https://en.wikipedia.org/wiki/Sampling_(signal_processing)   [This link needs the bracket on the end, but is doesn't seem to want to work in the translation here.]

I was talking with a tech/engineer today about this and he said 48kHz is used because it fits in with the video rate whereas 44.1 doesn't - or at least not when the system is setup to use 48.

Edited April 18, 2017 by hrh

[alanh]

hrh,

The sample frequency must be at least 2.2 times the highest frequency to be sampled, but it requires a good filter prior to sampling, it is easier if it 2.5 times.

The AES standard for all audio except the CD has been 48 kHz for a very long time. The audio industry knew that TV was going digital and as a result there will not be any need for a lock to NTSC frequencies.

All HD video formats divide in whole numbers to 48 kHz. Being slightly higher frequency the filter requirements are not as tight.

Alanh

 

[alanh]

MLXXX,

Has it ever occured to you the following

1. The manufacturer who made the DAB Module in your Roberts radio has gone through new 4 models of the chips used to receive the signal.

2. Manufacturers have improved the accuracy of the encoders. Even the designers of HE AAC V2 have a new model specifically designed for speech.

In both cases microprocessors have become faster and more capable.

If these low bit rates were so annoying you should be able to quote examples off you head. You have not done this so as a result the bit rates selected by our broadcasters is adequate. This is demonstrated by the fact that no one has joined your petition for increase.

The truth is that if you don't know the source you cannot detect the poor quality caused by the low bit rate.

You should apologise to Digital Radio Plus, the ABC and all the posters on this topic over the last 7 years. You cannot detect low bit rate errors on current broadcasts.

Alanh

 

Edited April 18, 2017 by alanh

[MLXXX]

15 minutes ago, alanh said:

MLXXX,

Has it ever occured to you the following

1. The manufacturer who made the DAB Module in your Roberts radio has gone through new 4 models of the chips used to receive the signal.

2. Manufacturers have improved the accuracy of the encoders. Even the designers of HE AAC V2 have a new model specifically designed for speech.

In both cases microprocessors have become faster and more capable.

You have raised these or very similar points before, alanh. However, turning on a DAB+ radio in 2017 in a showroom and trying out a selection of services will soon reveal quite noticeable artefacts with the lower bitrate services.  If encoders in use for DAB+ broadcasts in Brisbane in 2017 are somehow more effective, it's certainly not obvious to my ears.

[Pesto Lovin' Man]

13 hours ago, hrh said:

I think there could be a problem with that.
48kHz is what is used in the television industry - certainly in Australia at the least, and if the "samples" I have which is sourced from overseas is any indication, so do they.

I was talking about the digital pro audio recording systems that were around before & at the beginning of the CD era that Alan was refering too as his "reason" CD is 44.1kHz. They stored uncompressed PCM audio as a video luma signal on either NTSC (44.056kHz) or PAL (44.1kHz) recorders & couldn't fit 16 bit 48kHz asmpled audio into a NTSC or PAL video field/frame.

edit: at least the ones that were 16 bit did. There were others but they were 12/13/14/15 bit systems at differing bitrates e.g. NHK/Columbia had a 14 bit 47.25kHz system using either 2" or umatic that nearly became common.

His story is wrong or at least very misunderstood, although its a common belief thanks to dozens of websites & wikipedia that get it confused & wrong too.

Modern video studio & broadcast tape formats & digital broadcasts were developed later & are a different matter, & yes 48kHz sample rate is common. Though theres no particular reason that 48kHz fits any compressed tape/broadcast format better - its compressed so theres no strict relationship between sample rate & bitrate, its padded to fit the container if needed & usually timestamped for sync.

Edited April 19, 2017 by Pesto Lovin' Man

[alanh]

MLXXX,

"However, turning on a DAB+ radio in 2017 in a showroom and trying out a selection of services will soon reveal quite noticeable artefacts with the lower bitrate services.  If encoders in use for DAB+ broadcasts in Brisbane in 2017 are somehow more effective, it's certainly not obvious to my ears. "

So when did you do this?. You still haven't stated what is the problem now.

When are you going to a broadcaster with your request to increase the data rates?

What will be your answer when they say to what is wrong with our existing broadcasts? You say go to my posts in the dtvforum in 2009 for the answer. It really sounds like a pressing problem that the broadcasters need to solve. Then you say I only have a pair of signatures for this petition and one is mine. They say that there is 3.5 million listeners and many are listening to the simulcast of their FM and AM programs.

I would like to see the expressions on both their and your faces!

The problem is that you believe there is a problem, you have never blind tested your belief to prove it is just that.

Alanh

 

[alanh]

Pesto loving man.

48.000 kHz has been an Audio Engineering Society standard for a very long time. They set most audio standards.

A chief motivator for 48 kHz is that there is a whole number of samples per frame in 24, 25 and 30 frame/s TV systems along with film. So when time code is used for editing particularly separate sound and vision on a program there is no drop frames. As a result the edit decision lists have no confusion of dropped frames from the video in NTSC based systems and no drop frames in the audio. In addition because it works with 24 frame/s the movie industry has been using it extensively.

Without dropped frames, timecode is used to syncronise the recording and playback of separate sound recorders and the video recorders.

Alanh

[Pesto Lovin' Man]

6 minutes ago, alanh said:

48.000 kHz has been an Audio Engineering Society standard for a very long time.

Not as long as 44.1kHz CD though which is what I was talking about in reply to your garbage statement about 44.1kHz.

Edited April 19, 2017 by Pesto Lovin' Man

[MLXXX]

On 4/19/2017 at 11:29 AM, alanh said:

MLXXX,

"However, turning on a DAB+ radio in 2017 in a showroom and trying out a selection of services will soon reveal quite noticeable artefacts with the lower bitrate services.  If encoders in use for DAB+ broadcasts in Brisbane in 2017 are somehow more effective, it's certainly not obvious to my ears. "

So when did you do this?. You still haven't stated what is the problem now.

There is no mystery as to the "problem". It is what it has always been with a great number of the DAB+ radio services: insufficient bitrate for the audio. Here's a picture taken today of a DAB+ radio at Myers Chermside (Brisbane northside) tuned to 4KQ Classic Hits:

 

Although nominally a 48kbps service, the actual audio bitrate of 4KQ Classic Hits currently is substantially less (about  34.6kbps, HE-AAC v2).

That is a low enough rate for encoding artefacts to be noticeable almost continually for my hearing. What did I hear when listening to the above radio today?  Well, here's a quick list of what struck me immediately:-

1.  Hi-hat cymbals sounded like a mush of high frequencies rather than a clear cymbal sound. The percussive sound of the snare drum was similarly affected.
2. A writhing, phasey quality affected the sound overall. [This is a tell-tale psychoacoustic codec artefact. You get a similar sounding artefact with low bitrate mp3.]
3. A general lack of verve. 

These and other effects were sufficiently marked as to reduce my listening enjoyment substantially. Although I don't mind the genre of the music, the audio quality of this particular DAB+ service would be too irritating for me to be comfortable listening to it for an extended period. I would move on to a higher bitrate DAB+ service, or to an FM service.

There you go, alanh. Some subjective impressions of mine formed today. However there's nothing unique in what I've reported. Nothing new. These types of artefacts have been present since the start of DAB+ broadcasting in Australia.  They have been reported on fairly extensively on this website over the years.  And of course the published literature has plenty of graphs and tables of less than ideal MUSHRA scores for low bitrate HE-AAC. 

Your apparent suggestion that there are no audible impairments is beyond my understanding. Such a suggestion conflicts with official reports of formal testing of low bitrate HE-AAC on human subjects, and it conflicts with anecdotal adverse reports on this forum from people who have tried to listen to low bitrate DAB+ radio services.

Edited December 27, 2017 by MLXXX
Late edit to correct spelling of a suburb.

[alanh]

MLXXX,

Finally you have gone done some listening to a new radio.Then you pick a radio with tiny speakers which does not contain a pair of tweeters which can accurately reproduce high frequencies. Those speakers are trying to reproduce lower frequencies.When are you going to listen to a radio which is not a portable or tabletop receiver which all have poor sound quality regardless of the signal source. As I have previously suggested you listen to a DAB+ car radio they generally have better speakers.

As I have already suggested that you use nationally syndicated stations so we can listen to the same programs.

6IX had similar problems a few years ago, but they moved into a new studio complex and the problems have disappeared, but the bit rate is still 48 kbit/s. You had better to complain to 4KQ.

You are yet to convince lots of people to join your petition.

Alanh

 

[MLXXX]

9 hours ago, alanh said:

You are yet to convince lots of people to join your petition.

No such petition exists as yet, alanh, for Australia, whether put together by me or by anybody else. If and when one is drawn up and advertised, there will be the opportunity for people to subscribe to it.
 

9 hours ago, alanh said:

.Then you pick a radio with tiny speakers which does not contain a pair of tweeters which can accurately reproduce high frequencies.

 That radio at Myers Chermside is a premium standalone radio with speakers quite capable of delivering good quality, pleasant sounding, treble, as they did for me when I tried out some higher bitrate DAB+ stations.

There is no need for ruler-flat extended range speakers just to listen to a radio broadcast.  And if you did, extravagantly, connect up a very expensive audiophile system to reproduce the audio from a low bitrate DAB+ broadcast, that would not hide the artefacts. They would still be there, perhaps even a little easier to hear; perhaps even more irritating!

 

Listening to DAB+ radios in Brisbane showrooms

Most showrooms in Brisbane have too low a signal strength for DAB+ radios to work. One place I've found with a selection of DAB+ radios connected to power and capable of receiving the DAB+ signals from Mt Coot-tha is HN in the Valley. (Earlier this week I found that one of the radios there with rather better quality built-in speakers, a Panasonic SC-HC49DB, could be got to work with the help of a friend to lift the loose wire antenna into the air for you, giving you freedom to try different DAB+ services without reception cutting out!).

 

DAB+ time functionality useful for alarm clock radios

The majority of DAB+ radios on sale by the main electrical goods retailers in Australia have only mono speakers. There is a great number of alarm clock style DAB+ radios. Such radios have the convenience of being able to access the current time using the DAB+ signal. There is no need to set the time manually. It appears these radios can fully resume after a mains power outage without issues, and in particular without the need for a battery.

Edited April 20, 2017 by MLXXX

[MLXXX]

Tonight I've been looking for website material regarding DAB+ perceived quality in Australia, beyond the many references that exist in the archives of this forum. Here are two sources of material that point out negatives:

1.  A detailed webpage from 2010 titled Digital Radio Plus: NOT - Say no to DAB+ and its minuses includes this comment:

DAB+ (on the majority of stations, as of 2010) can be described as an audio signal that has been split into a two bands, with the low (bass to mid) band sounding muffled (as if you had pillows or several sheets over the speaker), combined with an overly accentuated and crunchy high (treble) band for good measure. It may re-create some of the original sound timbre, but the results are disastrous for music such as a orchestral pieces (the stringed instruments lose most of their nuances). ABC Classic FM on DAB+ does no justice to the music.

Unless you enjoy the horrible 'tinnyness', raspy, metallic, youtube-like sounding audio compression used on DAB+ (in the current state of allocations), it would be better to stick to FM.

 

2. An informal thread on Whirlpool forum from 2012 titled DAB+ low bitrate transmitted begins:

Why is DAB transmitting at such low bit rates? the quality is no where near an FM tuner? It is completely un usable on any sort of reasonable gear unless you regard listening to the horse races an a portable am radio stuck to you ear as reasonable. Or perhaps listening to your portable on a tram with your iBuds may be better than an FM radio in the same circumstances and for that sort of use it may well be truly better especially if you only want to hear the news. As far as high quality material in your home, a few minutes is about all one can stand and then it is back to the FM tuner. Why bother? 

 

In contrast to the above, I note that there are many articles that paint a positive picture of DAB+ sound quality, especially those used by the digital radio industry in promoting DAB+.

In my searching this evening I came across an article about DAB+ (that appears to come from 2010) attributed to Steve Ahern. The website states: Steve Ahern is an international broadcast training consultant, broadcaster, author and media commentator, specialising in Radio and New Media.  The article, titled Steve Ahern explains Digital Radio, includes these words:
 

... Digital radio quality is light years better than AM, and better than FM quality, while using much less bandwidth. 

How many bits per second a station chooses to use determines the quality of its signal. The improved form of digital radio used in Australia, called DAB Plus (DAB+), can deliver twice as many stations as the original form, DAB, used in the UK, Europe and elsewhere. Many countries are now examining the possibility of upgrading to the DAB+ standard and are looking at Australia to prove that the new improved system works. So far they are concluding that it works very well.

Most stations in Australia are getting good FM quality signals at a data rate of 48 kilobits per second, one third of the bit rate used to deliver similar quality in the older system.

 

What does the last sentence mean by “FM quality signals”? Is it the quality to be expected from a car radio driven about town in a good signal strength area, the quality from a quality FM receiver in the home connected to an external antenna receiving a strong signal, or the quality to be expected from a portable FM radio using its built-in rod antenna, located in a fringe reception zone or inside a steel structured building? I think a number of us who have listened to DAB+ broadcasts in Australia would disagree that 48kbps gives anything like as good as FM quality, in circumstances where the FM radio is connected to a good antenna and receiving a healthy signal.

And what is meant by a data rate  of “48kbps” in this context? Is it simply the nominal bitrate of the particular DAB+ service, as displayed by a DAB+ radio if a listener chooses to display such information; or is it the bitrate actually devoted to the audio itself, a smaller figure known only to specialists, that takes into account the overhead of error correction, and any bitrate reserved for Program Associated Data?

I suspect many people over the years who have written that "48kbps gives FM quality" have not turned their minds to either aspect.

Nor I suspect have they actually compared for themselves reception of a high signal strength FM broadcast with reception of a "48kbps" DAB+ radio simulcast (be that the nominal rate of the DAB+ service or the actual rate of the compressed audio itself). Because if they had, they might have had real doubts that "48kbps gives FM quality".

 

A forum poll

In the next few days I'll see if I can put together some comparison files of reception of off-air FM [using good signal strength] and simulcast off-air DAB+ "at 48kbps" for the purposes of a forum poll. Should I use a DAB+ service that is at an overall data rate of 48kbps or one where the bitrate for the embedded HE-AAC audio stream is at around 48kbps?  Which interpretation would be more appropriate? 

Edited April 20, 2017 by MLXXX
Paragraph added: "A forum poll".

[alanh]

MLXXX,

Get off the computer and into the real world.

Go to live concerts with real unsampled unamplified sound for example at Qpac and the QSO has a new conductor.

I note that there are many broadcast streams at 48 kbit/s and you only complain of a single faulty one. What about the others. No complaint about any of the 80 kbit/s streams.

Use your ears and not a search engine for others who copy each other and the myths keep circulating.

Lets face it your petition is a failure, your views are not held by others.

Alanh

[MLXXX]

1 hour ago, alanh said:

Go to live concerts with real unsampled unamplified sound for example at Qpac and the QSO has a new conductor.

You have such a limited memory, alanh. I have told you in the past that I attend orchestral concerts regularly.

I also happen to be a musician and currently perform in a number of different groups.  I hear live, unamplified, symphony orchestra instruments very frequently indeed.  

 

1 hour ago, alanh said:

I note that there are many broadcast streams at 48 kbit/s and you only complain of a single faulty one. What about the others.

It isn't "faulty". Its sound quality is quite typical for a low bitrate DAB+ broadcast. I see that you describe it as 48kbps, the nominal bitrate.

 

1 hour ago, alanh said:

Lets face it your petition is a failure, your views are not held by others.

I would remind you: there is no petition at this stage. The purpose of this thread is to gather together some points that might feed into a petition. 

As for others not holding my views, what an absurd statement! There have been many contributions to this forum over the years, from a variety of members, bemoaning the poor quality of many of the DAB+ services.  And in any case, official published test results indicate audible impairment perceived by test subjects for HE-ACC stereo at bitrates well in excess of 48kbps, let alone at the audio bitrate of a nominal 48kbps service.

Edited April 21, 2017 by MLXXX

[Pesto Lovin' Man]

This should keep everyone happy or unhappy

Subjective and Objective Comparative Study of DAB+ Broadcast System

DAB+ @ "96kbps" sounds better than FM but they didnt check the actual audio bitrate.

Subjective testing HE-AACv2 in all but a couple of cases showed 56kbps actual bitrate is sub-good quality & worse than 96kbps & so on.

Their objective test ViSQOLAudio wasnt a good predictor of subjective quality except maybe for some pop genres.

 

 

[MLXXX]

9 hours ago, Pesto Lovin' Man said:

This should keep everyone happy or unhappy

Subjective and Objective Comparative Study of DAB+ Broadcast System

Thanks very much, Pesto.  When I came across that recently published paper in my searches last night I glanced at it only briefly and mistook it for a paper I'd already read!

 

The paper is unusual in that part of it compares actual simulcast broadcasts, i.e. real life broadcasting of DAB+ with real life broadcasting of FM, albeit that no details are given as to the reception conditions for the FM.

I note that the more detailed results in the paper are for test HE-AAC encodings prepared by the research team, not live broadcast encodings.

 

9 hours ago, Pesto Lovin' Man said:

Subjective testing HE-AACv2 in all but a couple of cases showed 56kbps actual bitrate is sub-good quality & worse than 96kbps & so on.

For readers not familiar with absolute character subjective ratings using a 5-step scale:

 

Rating	Label
5	Excellent
4	Good
3	Fair
2	Poor
1	Bad

 

In the paper, we can see mean opinion scores on page 8 for actual HE-AAC bitrates of 64kbps, 96kbps, 128kbps and 160kbps. Perusal of the histograms reveals that use of the lowest bitrate, 64kbps, led to scores of  around 3 or "fair".  [There is no mention of broadcast processing. These appear to have been straightforward encodings using the HE-AAC codec and nothing else. I note that it is very possible the 96kbps encoding would have used AAC-LC, leaving only the 64kbps encoding to rely on HE-AAC v1, i.e. spectral band replication.]

No part of the paper explores bitrates as low as 48kbps.  At page 7 of the paper it is stated: "Currently, the digital DAB+ multiplex in Poland offers radio programs at 6 bitrates: 64, 72, 96, 104, 112, and 128 kbps.". 

 

9 hours ago, Pesto Lovin' Man said:

DAB+ @ "96kbps" sounds better than FM but they didnt check the actual audio bitrate.

I see that in Figure 1 of the paper [a 2016 paper by Gilski & Stefanski, published in 2017 in the Archives of Acoustics (The Journal of Institute of Fundamental Technological of Polish Academy of Sciences)], the lowest nominal bitrate of the simulcasting DAB+ services (the rate for the regional service) slightly exceeded 96kbps:

We in Australia would be envious of such high nominal bitrates. (The paper does not discuss differences between the overall data rate of the service, which appears to be what is quoted in Figure 1, and the bitrate of the embedded HE-AAC audio stream.)

An aside: I am beginning to think that for my sample files for Australian DAB+ vs FM I would be justified in using nominal DAB+ bitrates rather than actual bitrates. Doing so would align with the tendency to refer to real world DAB+ broadcasts by their nominal bitrate. I think that when members of the public read a statement about DAB+ radio such as "48kbps gives comparable to FM quality" it is reasonable for them to take that to mean the bitrate information displayed on their radio,  or the rate they might see publlshed on the internet if they search.  It's quite rare to see the actual audio bitrate of a DAB+ radio service disclosed.

 

64 kbps (actual) in Australia - what could be expected?

In Australia, the bitrate of 80kbps nominal is used by the ABC for programs that consist mostly of music. That bitrate roughly translates to 64kbps for the audio depending on how much bitrate is reserved for Program Associated Data. Here is an extract from details I posted on this forum recently (at http://www.dtvforum.info/index.php?/topic/96036-digital-radio-stations-list/&do=findComment&comment=2093845) for DAB+ in Brisbane:-

9C 206.352MHz	CODEC	Nominal	Audio	Comment
BR abc&SBS RADIO		bitrate	bitrate
ABC Classic FM	HE-AAC v1	80	63.8
ABC Country	HE-AAC v1	80	47.6	large PAD
ABC Jazz	HE-AAC v1	80	47.6	large PAD
Double J	HE-AAC v1	80	47.6	large PAD

 

It is interesting to note the following comments in the paper, which appear at page 7:

Quote

When it comes to analysing speech or singing samples, the most important issue is the clarity and transparency of the audio material, since information contained in the voice must reach the listener. According to the subjects, the sound colour was significantly worse for samples coded at lower bitrates, especially 64kbps. This impression was given regardless of the music genre.

... Furthermore, as the listeners indicated, in case of signal samples from category 3-4, spatial attributes of sound, including spaciousness, sound perspective and localization stability, were reported as annoying or even unacceptable for bitrates lower than 128kbps. This effect was less common for electronic music pieces, as it was for classical or popular music.

Edited April 21, 2017 by MLXXX

[alanh]

MLXXX,

Pity you don't understand the articles you quote

1. Why is talk 2 getting a worse rating than talk 1 when the bit rates are identical.

2. Where are the quality assessment graphs of all the samples using of FM as processed for pre-emphasis and then FM modulated and then received?

3. There is no mention of the audio processing applied to their DAB+ and FM signals prior to HE-AAC compression. Is it the same as used in Australia? Did they use any as it is common place in radio stations here. Our radio stations do not wish to pay for operators to control levels and commercial stations particularly want their station to be louder.

4.There is no references from the research institute in Germany which designed the family of AAC of compression standards.

I note that you have only complained about a faulty DAB+ stream where the radio station has been converting compressed signals back to analog and then recompressed using different algorithm. As a result there is link to lock the sample clocks together which causes the phasing effect you complained of. So since 4KQ is feeding an analog transmitter and hence speakers these effects will not be present. Digital radio stations have been warned not to do this.

There was no mention of the phasing effect you complained of. I note that the source signals are sampled at the same sample rate as used by DAB+ and there is no lossy compression prior to DAB+ encoding.

Even playing CDs which is lossless compression, the digital radio station needs to use a broadcast professional CD player which outputs its signals in AES 48 kHz sample rate signals. Internally the CD player will have to play the disc at 44.1 kHz to maintain the correct pitch and duration. This digital signal will have to be interpolated using a 48 kHz signal. A phase locked loop is used to lock the two oscillators. The CD sample rate is divided by 441 and the AES reference by 480. This means there is a phase comparison every 10 ms. This will stop the phasing effect you complain of. I note that other broadcasters don't have the phasing effect.

Have you complained to 4KQ about their poor sound quality yet?

I noticed you have not complained about the audio quality of any other broadcaster.

Alanh

 

[MLXXX]

8 hours ago, alanh said:

1. Why is talk 2 getting a worse rating than talk 1 when the bit rates are identical.

One possibility is different sizes of PAD.

I'm sorry, alanh. I don't have the time or inclination to attempt to address the other points you raise other than to say again that sample rate conversion is a routine process. In my opinion it has no significance for this thread.

Within the context of a low or moderate bitrate psychoacoustic codec used for radio broadcasting, any alterations in sound quality are negligible that might arise from converting PCM source material at 44.1kHz (such as digital sound recovered from a red book CD disc) to 48kHz. If you don't believe me, I suggest you spend time with the freeware Audacity converting sample rates,* and see whether you can hear any differences.

_____________

* One exercise is to use Audacity to save a 48kHz sample rate file at a 192kHz sample rate [by temporarily changing the project sample rate]. Reload that saved file and displace it 1, 2 or 3 samples on the time line (your choice), and then save again but at a 48kHz sample rate. Compare the original 48kHz sample rate file waveform with the waveform resampled with a phase offset relative to the 48kHz sampling. The files will look the same shape in the editor apart from the time displacement, and you will find that they sound the same when played.  

Or set the project rate to 48kHz. Generate a 20kHz sine wave, and save it with a 192kHz project rate. Open that saved file and shift the waveform slightly on the timeline. Then set the project rate to 48kHz and save. The displaced phase is fully preserved when resampling to 48kHz. This flows from Nyquist. You can fully ascertain a waveform by sampling at no less than twice the highest frequency component of the waveform. In the physical world, there is no need to lock your ADC to the same phase as your DAC if a DAC is generating the waveform being sampled with the ADC and they both are operating at the same nominal sampling frequency but free-running. You get the same result. It also doesn't matter if you use a different nominal sample rate for your ADC (as long as it isn't too low a sample rate for your highest frequency component being sampled). The reconstructed waveshape  for the new digital file will be the same as it was for the old digital file, and will be indistinguishable when listening under double blind conditions. [Well if you use actual hardware DACs and ADCs there may be very slight differences from a conversion depending on the model of ADC or DAC. Conversions in the digital domain with software should be transparent to the human ear.]  At first blush this result is not at all intuitive for many people. Digital sampling is an extremely powerful technique!
 

It may help for an intuitive appreciation to consider that in nature a cycle of sound never exists in isolation. There is always an onset and a decay. [And if you try to use an electronic synthesizer to generate a single cycle you will still have the delay of the response of the human cochlea  to contend with.] As for digital to audio converters (DACs) and Analogue to Digital converters (ADCs) these oversample, i.e. operate at much higher rates than the nominal sample frequency. A decimation process can then provide a very nicely smoothed reconstructed waveform. The proof of the pudding however is in the eating: the 44.1kHz CD has been remarkably successful!
 

Edited April 24, 2017 by MLXXX

[MLXXX]

On 4/17/2017 at 1:58 AM, MLXXX said:

I think another thing to consider to increase the bitrate available for music services would be to pare back the allocation for PAD. I'm not sure exactly how this works but I suspect that if the PAD bitrate were reduced it would still allow a basic slide to be displayed. It might just take longer for it to load when a user selects the service.

It does indeed seem to be the case that the image will load eventually. In the example below of a DAB+ service in Crakow, Poland (Kraków, Polski) called "FM+" the image took 47 seconds to load:

Questions that come to mind include:

1. Is 47 seconds too long for a user of a DAB+ service to wait for a 320 x 240 slide image to appear on the display of their DAB+ radio?
2. What impact on sound quality would halving that time have?
3. What impact on sound quality would dispensing with slides altogether have?

In relation to 3, if the nominal bitrate of the service were very low, then dispensing with slides altogether could noticeably improve the sound quality.  I note that DAB+ radios on sale in Australia have displays that show the name of the song (similar to FM radio RDS text) but rarely if ever have a display big enough to show a slide.

In relation to 2, if the nominal bitrate of the service were very low, then doubling the bitrate for slide acquisition could noticeably reduce the sound quality.

In the particular case of this DAB+ service provided in Poland (which happens to be the country in which the DAB+ FM comparisons were carried out for the 2016 paper that alanh raised queries about in this thread yesterday) the nominal bitrate was high for DAB+ radio, being at 144kbps. The bitrate for Program Associated Data of 6.8kbps represented only 5.2% of the nominal bitrate and would have had only a minor impact on the subjective sound quality. The bitrate available for the DAB+ audio itself (using the AAC -LC codec) was high for DAB+ broadcasting, being at 123.7kbps. These figures can be seen at 4m 24s into a Youtube video uploaded in September 2016. Here is detail from that time in the video:

 

Here is a link to the full YouTube video, but set to start at 4m 24s into it. You can witness a significant delay before the slide is fully loaded and displayed:- 

 

 

Edited April 25, 2017 by MLXXX

[alanh]

MLXXX,

You must be desperate using Uni student assignments to prove your case. Where is the Professor's assessment of their work? All the subjects were young, probably fellow students who are hardly average listeners. There age is well under the average Australian age.

When are you going to give examples of non-faulty Australian broadcasts which are fed to all mainland capital cities such as the ABC?

I don't know why you keep on about this. Brisbane is already using 3 DAB+ transmission channels when you add one each for Gold Coast, Sunshine Coast, Wide Bay, Darling Downs, and Lismore you have used all the 8 available channels. (TV uses all the others that DAB+ receivers tune). The ABC/SBS will have to use a single frequency network for Qld and another channel for Lismore as the programs are not identical particularly in Daylight saving in NSW.

So the the only way you can get more data rate is to push for DRM+ which is capable of 186 kbit/s however that will probably be used for multiple programs like the DAB+ services.

Your pole is a failure.

So forget it after you complain to 4KQ which will probably have to replace its sound library with a new one like 6IX did.

Give it a rest.

Alanh

 

[MLXXX]

15 hours ago, alanh said:

Your pole is a failure.

Alanh, you've mispelled that word on this forum before. Please note the distinction in spelling: the north pole, an opinion poll.

However this thread is not a poll thread. It is for gathering together of points/information.

 

15 hours ago, alanh said:

You must be desperate using Uni student assignments to prove your case. Where is the Professor's assessment of their work? All the subjects were young, probably fellow students who are hardly average listeners. There age is well under the average Australian age.

The particular paper was brought to our attention by Pesto Lovin' Man.   I don't know the age ranges of the people who made up the panel of test subjects. The paper is rather light on, for detail, in a number of respects.

In terms of the off-air part of the study, the finding expressed in the paper is that the DAB+ broadcasts sounded subjectively better than the FM broadcasts. That part of the study does not assist me at all.  It might offer comfort though to broadcasters who use relatively high bitrates. 

I think there is a need to read between the lines with this paper. On the surface it says that according to the listening panel's individual subjective assessments (analysed collectively and statistically), real life DAB+ broadcasting [with whatever sound processing the broadcaster chose to use for that signal] surpassed simulcast real life FM broadcasting [with whatever sound processing the broadcaster chose to use for that signal] in Poland. But of course -- and this is not highlighted in the abstract for the paper or the summary part of the paper -- that was for relatively high bitrate DAB+ broadcasts. As for the FM, no detail was provided of reception conditions.

The nominal bitrates of the DAB+ broadcast services selected for the study were all at the high end for DAB+ broadcasting, when compared with DAB+ broadcasts in other parts of the world. Certainly the rates are very much at the high end compared with what is common in Australia. Even the lab-prepared sound files part of the study refers only to relatively high bitrates.  For that part of the study we are provided with a range of more detailed results, revealing serious issues with the AAC codec.  I consider that there is an implication from the laboratory files part of the study (not made explicit in the summary section of the paper) that DAB+ broadcasters need to pay attention to the bitrate of the encoded audio. It appears that the writers of the paper were content to indicate that the subjective sound quality of DAB+ broadcasting can compare favourably with FM broadcasting, without being too specific about bitrate (or for that matter the circumstances of reception of the FM signa). Or perhaps the paper was watered down at the last moment, on the advice of others, to avoid controversy. We are unlikely to ever know! 

For the part of the study dealing with unprocessed sound files of CD quality simply encoded to HE-AAC at different actual bitrates (as distinct from "nominal" bitrates), the reported results confirm what previously reported tests have revealed, that the subjective quality is less at lower bitrates. At the risk of repeating myself too many times, I will say again that although HE-AAC is a very efficient codec for modest bitrates it is by no means transparent at modest bitrates.  It does a remarkably good job under adverse conditions. It doesn't necessarily sound "good". It just sounds better than many competing codecs.

If Mr Trump gets his way setting a goal of a manned landing on Mars, a specialist low bit-rate codec for speech, such as a further refinement of extended HE-AAC, might be the choice for voice communications from the spacecraft back to earth, to wring out intelligible performance for human speech with the very low bitrates available for use over vast distances in outer space. However in a first world country like Australia, for local radio broadcasting, we should aim for a high standard, I suggest. Coding artefacts should be barely if at all noticeable most of the time. To achieve that, I think we need something of the order of 96kbps (actual), or better.

Edited April 26, 2017 by MLXXX

[MLXXX]

There have been so many excellent posts to this forum in the past regarding DAB+ sound quality. Here is one I came across tonight while searching for something else. I think it explains the audible effect of reducing the bitrate, in a very easy to follow manner:-

[alanh]

MLXXX,

If you read the link, the ages of the subjects is quoted. As I said you don't understand your google searches.

As I have already pointed out there are other factors which determine the sound quality from a broadcaster other than just bit rate tests in a lab in a University.

If the problems are as bad as you claim why can't you quote Australian examples, with broadcaster, date and time of transmission and a description of the sound and what is wrong with it.

You cannot hear any of these problems you keep quoting on syndicated Australian radio so how can you point out to an Australian broadcaster what is wrong with their digital sound.

 There is no likelyhood that any broadcasters will change their bit rate if you cannot demonstrate the problem with their received output on a set of studio quality speakers.

Have you complained to 4KQ yet?

Alanh